Scoreboard

By the close of the 2027 upfront negotiations in June 2027, no top-10 US advertiser will let an AI agent shift budget between Google, Meta, and Amazon fully autonomously without a human approval step in the loop.

Why McEachern of Basis said directly that agents moving budget across platforms on their own is directional, not standard practice, and that the moment they get "hands on the wheel," guardrails become the whole conversation. The 58% of execs in the MadTech survey expecting agentic buying to scale within a year are describing optimization and assistance, not unsupervised money movement between rival walled gardens who have every incentive to bias an agent toward their own inventory. Large advertisers with shrinking CMO tenures and audit exposure will not hand three competitors an unsupervised budget dial that fast; the opposite outcome would require both the protocol to mature past read-only and enterprise legal and finance teams to accept autonomous cross-platform spend inside eighteen months, which is faster than any prior buying-automation shift has moved.

Right if: no top-10 US advertiser has publicly deployed fully autonomous cross-platform budget reallocation across Google, Meta, and Amazon without human sign-off by the 2027 upfronts. Wrong if: at least one has.

Want AI Agents That Work? You've Gotta Do The 'Unsexy' Stuff First Listen to the episode →

IAB Europe's 2027 follow-up survey, expected around September 2027, will not report that a majority of respondents have autonomous AI agents executing live media buys without human sign-off.

Why The 58% figure measures expectation, not deployment, and the accompanying 86% "advanced human-AI collaboration" number describes humans working with AI, not agents buying alone. Real autonomous spend requires machine-readable deal pipes across DSPs and SSPs, buyer trust to remove the human trigger, and a clear owner when the agent spends wrong, and none of those are close on the open web. NLW's own reporting on agent risk points at exactly the control problem that keeps a human in the loop. The opposite outcome, majority autonomous execution within a year, would require the industry to hand budget authority to systems it does not yet trust and cannot yet cleanly reconcile, which is not how buyers behave with money.

Right if: IAB Europe's next agentic survey shows most respondents still keep a human approving agent-driven buys. Wrong if: a majority report agents executing live media buys without human sign-off.

MadTech Daily: 58% of Ad Execs See Agentic Buying Scaling; Apple & Amazon Face Renewed UK Consumer Case Listen to the episode →

Meta's checkout-driven agentic commerce will not have measurably shifted holding-company budget concentration toward Meta by the time the big agencies report second-half 2026 results in February 2027, and no top-five brand will publicly commit to routing its transactions through Meta's native checkout by then.

Why The whole funnel-collapse thesis depends on merchants handing Meta pricing, inventory, and fulfillment data, and brands have refused that deal since Instagram Checkout launched in 2019 for exactly the reason they ration data to Amazon: whoever owns the transaction record out-targets you next quarter. The signal in this story is a DSP CEO's quote, not a single named merchant or a shipped product at scale, and Vanderhook's incentive is to make Meta sound existential. For the opposite to happen, Meta would need to both ship native checkout broadly and win merchant trust it has never won, inside a few months, while EU regulators are already scrutinizing its commerce data use. That's a lot to clear on the timeline the "collapse" framing implies.

Right if: no top-five global brand has publicly committed to routing its checkout through Meta's native interface by the major agencies' second-half 2026 reporting. Wrong if: a top-five brand announces native Meta checkout as its primary transaction path, or a holding company reports a specific, quantified budget shift into Meta commerce inventory.

Viant CEO: Meta Collapsing Ad Funnel Into Single Commerce Interface Full Analysis → Read the source story →

Criteo's full-year 2026 revenue, reported in February 2027, will not show a distinct, disclosed OpenAI-driven revenue contribution material enough to lift its total above its 2025 revenue.

Why The Digiday reporting says buyers have already committed spend but can't find enough placements to deploy it, which means the ceiling on 2026 revenue through this channel is set by OpenAI's ad inventory, not by Criteo's ability to route demand into it. Criteo's core retargeting business has been roughly flat for years, so a genuine OpenAI contribution large enough to move the full-year total would have to overcome both that flat base and the scarcity ceiling in a single year. The opposite outcome, a visible OpenAI lift to Criteo's total, would require OpenAI to solve inventory, measurement, and privacy terms fast enough for real budgets to clear through Criteo's pipes in 2026, and the five hurdles named make that a 2027-plus story. The plumbing deal is real; the revenue timing is being pulled forward.

Right if: Criteo's reported full-year 2026 revenue is flat or lower than 2025 and no OpenAI-attributable revenue line materially changes that. Wrong if: Criteo reports 2026 revenue above 2025 with OpenAI-driven demand cited as a contributor.

OpenAI Ad Business at $1B Run Rate, Eyes $25B by Year-End Full Analysis → Read the source story →

Open web video CPMs cleared through DV360 will fall year-over-year when Magnite reports full-year 2026 results in February 2027, as Gemini's blended optimization pushes spend toward Google-owned Shorts inventory.

Why Gemini decides the split between YouTube Shorts, which Google owns and monetizes at full margin, and open web inventory, where Google earns a thinner slice. An AI that optimizes toward Google's performance signal has every reason to favor the inventory Google owns, and Scott Messer named exactly this CPM-compression risk at Programmatic IO. When a single automated engine arbitrages across a combined pool, the lower-margin supply, open web video, is where clearing prices get squeezed first. The opposite outcome, open web CPMs rising because walled-garden budget floods in, requires Google to voluntarily route meaningful spend off its own high-margin inventory, which its behavior under antitrust scrutiny gives no reason to expect.

Right if: Magnite's full-year 2026 CTV or video clearing rates and take-rate commentary show open web video pricing under pressure. Wrong if: open web video CPMs hold flat or rise year-over-year.

Google Unifies YouTube Shorts and Open Web Video Buying in DV360 Full Analysis → Read the source story →

Judge Leonie Brinkema's final remedy order in the DOJ ad-tech case will require Google to divest or open one of its two publisher-facing ad-tech assets (the AdX exchange or the DFP publisher ad server), rather than settling for behavioral-only conduct restrictions, when the order is issued.

Why Brinkema found Google unlawfully linked its ad exchange and its publisher ad server, and courtroom skepticism about the open web's future doesn't erase that liability finding. Courts that find illegal tying between two products routinely order the products separated, because a behavioral promise from a company that already broke the rules is hard to police. The Skeptic's argument holds: a shrinking market is still a monopolized one, and remedies address the harm proven at trial. The liability finding controls the remedy design; the market's projected size does not. The opposite outcome, a conduct-only order, is less likely because the DOJ won on the structural tie and Brinkema signaled during liability that behavioral fixes are weak against a repeat actor.

Right if: Brinkema's remedy order requires divestiture or mandatory open-sourcing/separation of AdX or DFP. Wrong if: the order imposes only conduct rules (interoperability, auction transparency, non-discrimination pledges) with no forced separation of either asset.

Google Antitrust Remedy Phase Questioned Whether Open Web Display Would Survive Full Analysis → Read the source story →

Lumen Research will be acquired by a measurement or verification company, most likely Integral Ad Science or DoubleVerify, with a deal announced on or before 2027-03-31.

Why Lumen struck three distribution deals at once (The Trade Desk, Amazon, IAS) without any evidence of scale in this story, which is the pattern of a company proving reach to raise its exit price rather than building an independent business. Attention is the obvious next bolt-on for verification vendors after brand safety and viewability, and IAS already holds the partnership that gives it first look, while DoubleVerify has been the more aggressive acquirer recently. The category is consolidating, and a subscale attention vendor with good signal is worth far more inside a giant's contract bundle than standing alone, so the independence path is the less likely one. The main way this call fails is timing: a buyer could wait for Lumen to prove premium pricing first, pushing the deal past Q1 2027.

Right if: Lumen Research announces an acquisition by a measurement, verification, or identity company on or before that date. Wrong if: Lumen remains independent with no acquisition announced, or raises a new independent funding round instead.

Lumen Research lands button in The Trade Desk, partners with Amazon and IAS Full Analysis → Read the source story →

By the 2027 agency budget cycle, at least one of the major buy-side platforms building AI inside its own walls (The Trade Desk, Google's DV360, or Meta) will ship or acquire a cross-platform workflow feature that reaches outside its own inventory, squeezing the standalone orchestration startups.

Why Mimica's whole positioning rests on the DSPs building AI only within their walls, leaving the space between platforms open. That gap is precisely what an incumbent closes once it sees demand, because a buyer who orchestrates across platforms from The Trade Desk's console has less reason to leave it. The Trade Desk already rebranded its AI push and Google and Meta are pouring money into buy-side automation, so extending one step outside their own reporting is a small product move with a large lock-in payoff. The opposite outcome, incumbents staying politely inside their walls while startups own the connective layer, is the less likely one because glue between rivals is worth more to the platform that captures the buyer than to a neutral third party.

Right if: The Trade Desk, DV360, or Meta ships or buys a workflow tool that pulls and acts on data from a competing platform. Wrong if: all three keep their AI automation confined to their own inventory and reporting through mid-2027.

Ep 151: Tanja Mimica of Kovva on What the Future Holds for Media Buyers Full Analysis → Listen to the episode →

By the Q4 2026 earnings calls of Publicis, WPP, and Omnicom (reported February to March 2027), at least one of the three holding companies will publicly report client contract terms shifting away from fixed headcount-based fees toward output- or outcome-based pricing.

Why Tobaccowala says clients are demanding roughly 10% annual fee cuts and cites FT reporting that banks are refusing standard law-firm associate billing rates, which means the pressure is already live in the same headcount-time business model agencies use. The mechanism is cost: Epoch AI's data shows the price to produce a given level of AI work is collapsing fast, so any buyer paying old rates for work that is now cheaper to make can see the margin and will push to reprice. Holding companies talk about pricing structure on earnings calls constantly, and Publicis in particular has built its whole outperformance story on being ahead of this shift, so it has a strong incentive to say so out loud. The opposite outcome, all three staying silent on pricing model change through an entire earnings cycle while the pressure is this public, is the less likely world.

Right if: Publicis, WPP, or Omnicom describes a concrete move toward outcome- or output-based client pricing on a Q4 2026 earnings call or in accompanying materials. Wrong if: all three continue to describe client fees only in traditional headcount, retainer, or commission terms.

Companies Don’t Transform, People Do: Rishad Tobaccowala on Work, Agencies, and Reinvention Full Analysis → Listen to the episode →

OpenAI will close a primary funding round at a post-money valuation of at least $800 billion before the end of Q1 2027 earnings season (by the mid-February 2027 close of the major-lab reporting window).

Why Swisher named an OpenAI round in progress that would roughly double the valuation to about $800 billion, and the pattern across the last two years is that these announced targets close, usually near or above the floated number, because the labs need the cash to pay for training and serving compute that no revenue line yet covers. The mechanism is straightforward: the compute bill is non-negotiable and rising, so the round has to happen, and investor demand for frontier AI exposure has consistently met these asks. The opposite outcome, a failed or sharply down round, would require AI investment appetite to crack in the next six months, and nothing in the current fundraising cadence points that way.

Right if: OpenAI announces or is confirmed to have closed a primary round at $800 billion post-money or higher by then. Wrong if: no such round closes at that level, or if it closes below $800 billion.

Sir Michael Moritz on ‘Auslander,’ Identity and Life as an Outsider Listen to the episode →

By the IAB Annual Leadership Meeting in early 2027, at least one major verification or measurement vendor among DoubleVerify, Integral Ad Science, and HUMAN Security will ship or announce a product that classifies whether ad creative is AI-generated.

Why DTC brands are already generating AI ad creative at volume, and the cost curve Mashrabov describes (open-weight models at 80% margins, cheap enough that video creative approaches free) means the feed fills with machine-made ads whether or not his CPM math is right. Brands will want to know what they're buying next to and whether their own creative reads as authentic, which is the exact adjacency DoubleVerify, IAS, and HUMAN already sell into: brand safety, fraud, made-for-advertising site detection. Adding "AI-generated creative" to that stack is a natural next SKU, and these vendors have a track record of shipping a classifier for every new content-quality panic. The opposite outcome (nobody productizes it) would require the verification category to sit out a fear its buyers are actively raising, which runs against how it has behaved on every prior scare.

Right if: DoubleVerify, IAS, or HUMAN publicly launches or announces an AI-generated-creative detection or classification product by then. Wrong if: none of the three has shipped or announced such a product.

20VC: $1BN ARR in 18 Months; The Untold Story of Higgsfield | Spending $4M Per Month on Models | Why Moats in AI are BS | Scaling a Content Team to 150 People with Alex Mashrabov Listen to the episode →

At least one more U.S. state attorney general will announce a teen-safety settlement or filed suit against TikTok or Meta by the end of Q2 2027.

Why Alabama just became the first state to settle a teen-safety case against TikTok, for $100 million with a defined set of controls (two-hour default limits, overnight lockouts). That gives every other state attorney general a proven playbook and a number to anchor to, and multistate coalitions have already been litigating against Meta on the same theory. When one AG establishes both the legal theory and the settlement price, others follow because the hard work of proving the case is done and the political upside of protecting kids is obvious. The less likely outcome is that no other state moves for over a year, which would require AGs to leave a proven, popular case on the table.

Right if: another U.S. state announces a settlement or files a new teen-safety suit against TikTok or Meta. Wrong if: no such state action appears by that date.

MadTech Daily: McDonald’s Bets on $1bn Ad Business; Temu Cuts $1bn Fake Influencer Network on Meta Full Analysis → Listen to the episode →

In its next full quarterly disclosure covering Q3 2026 (reported by early November 2026), Reddit will report US advertising revenue growth above 40% year-over-year, as publisher-diversification and search-driven referral traffic reallocates toward platforms advertisers can still buy at scale.

Why The Digiday summit named Reddit, Threads, and TikTok as the platforms publishers are steering audience toward as Google referrals collapse, which means engagement and time-on-platform are rising exactly where advertisers can still buy authenticated, high-intent audiences at scale. When audience attention moves to a platform, ad demand follows the attention, and Reddit sells the kind of interest-based, logged-in inventory that performance buyers reallocate into when open-web reach degrades. The opposite outcome, growth decelerating below 40%, would require the attention shift the publishers themselves are describing to somehow not translate into ad dollars, which runs against how budgets have chased engaged users on every prior platform migration.

Right if: Reddit's Q3 2026 earnings report shows US advertising revenue up more than 40% year-over-year. Wrong if: US ad revenue growth comes in at 40% or below.

Publishers Accept Google Search Traffic Is Gone for Good Full Analysis → Read the source story →

New Jersey will amend or formally narrow its data broker law to exempt or carve out first-party publishers before the April 1, 2027 registration deadline.

Why The summary reports an anonymous state official saying enforcement may be suspended pending legislative fixes, which means the people running the law already know publishers were caught by accident. Sweeping the New York Times into the same public registry as career data brokers is politically indefensible and legally shaky, so the pressure to narrow the "data collector" category is strong and comes from inside the state, not just from lobbyists. The opposite outcome, New Jersey holding the broad definition and letting publishers register as brokers on April 1, would require the legislature to defend a result it appears not to have intended, which is the less likely path. The risk to the call is timing: legislatures are slow, and the fix could slip past April even if everyone agrees it is coming.

Right if: New Jersey enacts an amendment, issues formal guidance, or passes a rule that exempts or materially narrows first-party publishers from the data-collector definition before the April 1, 2027 registration window opens. Wrong if: the broad definition stands unchanged and publishers are required to register as data collectors on April 1, 2027.

New Jersey Data Broker Law Unexpectedly Snares Publishers as "Data Collectors" Full Analysis → Read the source story →

No large agency holding company (Omnicom, WPP, Publicis, Dentsu, Havas, or Stagwell) will report GEO or "AI visibility optimization" as a distinct, quantified revenue line on any 2026 earnings call, through the Q4 2026 reports in February 2027.

Why Droesch's own data undercuts the category: the best lever is brand-owned websites, which is a content-and-SEO job the holdcos already sell, not a new product, and the visibility lift he cites (18% to 20%) is too small to survive a CFO review as a standalone fee. When a service is both hard to attribute and easy to fold into existing content and search practices, agencies bundle it rather than break it out, because a small, un-attributable line invites exactly the scrutiny that kills it. The opposite outcome, a holdco proudly quantifying GEO revenue, would require the category to be big enough to boast about, and nothing in this story says it is. This is a call about the business staying too small to name, not about reporting mechanics: the silence tracks the size.

Right if: none of the six named holding companies reports a quantified GEO or AI-visibility-optimization revenue figure through their Q4 2026 earnings calls. Wrong if: any one of them puts a specific GEO revenue number in front of investors in that window.

AI Recommendation Engines Compress Retail Product Visibility for Brands Full Analysis → Read the source story →

Neither Magnite nor PubMatic will attribute a measurable revenue lift to per-bidder GAM floors on their Q4 2026 earnings calls, reported in February 2027.

Why The 5% lift is measured at the publisher level on GAM-served inventory, and it redistributes auction surplus toward pre-bid paths, which is where Magnite and PubMatic sit. But that surplus is spread across thousands of publishers who adopt floor stratification at wildly different speeds, and most of the long tail won't tune floors carefully within a single quarter. Google also has every incentive to protect AdX margin by shifting budget to owned-and-operated supply the moment floors bite, which caps how much surplus actually reaches the independent SSPs. A per-bidder floor policy flipped on in the fall of 2026 does not move a public SSP's revenue line enough to call out by February, and management won't credit a lift they can't cleanly measure.

Right if: neither Magnite nor PubMatic cites per-bidder GAM floors as a revenue driver on their Q4 2026 calls. Wrong if: either one names it as a contributor to reported or guided revenue.

Google Restores Per-Bidder Price Floors in GAM, Reversing 2019 Decision Read the source story →

NVIDIA's Q3 fiscal 2027 data-center revenue, reported in its November 2026 earnings, will grow year-over-year, driven in part by inference (running trained models in production) rather than only model training.

Why This Amazon Bedrock architecture puts NVIDIA's Triton inference servers inside the live ad auction, which is one more example of AI compute moving from one-time model training toward always-on production use, the workload that scales with every bid request rather than every training run. NVIDIA management has repeatedly said inference is becoming the larger share of data-center demand, and ad-tech running deep-learning models per impression is exactly the kind of high-volume, low-latency workload that consumes GPUs continuously. The opposite outcome, a year-over-year decline, would require a broad collapse in cloud AI spend that nothing in this story or the wider market suggests. The interesting half is the composition: if NVIDIA calls out inference as a growing driver, it confirms the compute-layer thesis that value in programmatic is settling below the application, where Amazon and The Trade Desk fight, and into the chips underneath all of them.

Right if: NVIDIA's November 2026 data-center revenue is up versus the year-ago quarter and management attributes part of the growth to inference workloads. Wrong if: data-center revenue is flat or down year-over-year.

Amazon Quietly Builds Agentic Bidding Infrastructure on AWS Bedrock with NVIDIA GPUs Read the source story →

Publishers running per-bidder price floors in Google Ad Manager against Google's own AdX buyer will report a measurable single-digit-percent-or-better revenue lift, and Google will not roll back per-bidder floors before its next Google Ad Manager pricing policy update in the first half of 2027.

Why Google re-enabled per-bidder floors in Google Ad Manager after losing the antitrust case where its 2019 Unified Pricing Rules, forcing one floor for every buyer, were a central complaint. The early signal is a reported 5% lift from a publisher flooring Google's buyer higher, and the mechanism is simple: when a publisher can charge Google's AdX buyer more than rival supply paths, Google either pays up or loses the impression, and either way the publisher captures value the single-floor rule denied them. Google won't reverse this quickly because doing so re-creates exactly the behavior a court just ruled against, so the legal cost of undoing it now outweighs the revenue Google gives up. The opposite outcome, a fast rollback, would hand regulators fresh evidence Google can't help itself.

Right if: publishers report a measurable single-digit-or-better lift from per-bidder floors on Google's buyer and the feature remains live in Google Ad Manager. Wrong if: the lift fails to materialize in publisher reporting or Google removes per-bidder floors before then.

Coins in the Wishing Well Full Analysis → Listen to the episode →

The Trade Desk's take rate (the share of each ad dollar it keeps) will be lower in full-year 2026 than in 2025, visible when it reports platform spend against revenue on its Q4 2026 earnings call in February 2027.

Why The ANA gave buyers a quotable 29-cent cost-of-the-middle figure, and a reported Publicis audit fight gives the largest buyers both the motive and the template to negotiate the fee down. Once one holding company formalizes a DSP fee audit, the others have to follow or explain to clients why they didn't, so fee pressure compounds across the book rather than staying with one client. Guidance already came down $89 million and 15% of staff was cut, which says management is defending margin under pressure, not holding pricing power. The opposite outcome, take rate flat or up, would require buyers to stop pressing on a number they can now say out loud in a budget meeting, which runs against every incentive procurement has heading into 2027 renewals.

Right if: The Trade Desk's implied take rate (revenue divided by gross spend on the platform) for full-year 2026 comes in below the 2025 level. Wrong if: it holds flat or rises.

Opinion: The Trade Desk is a broker, not a platform Full Analysis → Read the source story →

PubMatic's and Magnite's combined revenue share of open-web programmatic display will not rise as a result of this ruling by their Q4 2027 earnings reports (February 2028), the last prints before the late-2027 implementation deadline this ruling sets.

Why The remedy only reaches Google's exchange and supply-side tools, and forces AdX to bid inside Prebid on "functionally equivalent" terms, but it leaves DV360, Google's buy-side platform, completely alone, so the advertiser demand that actually decides where money flows still routes through Google's own pipe. A cleaner auction gives the two independent SSPs a fairer shot at the same buyers, with no mechanism to attract new ones. On top of that, appeals and six years of source-code audits push real enforcement to late 2027, by which point AI-driven search is shrinking the open-web display market these auctions serve. The opposite outcome, a measurable share gain, would need fast enforcement, an enforceable definition of equivalence, and a stable market to gain share in, and none of those three is on the table.

Right if: PubMatic's and Magnite's combined share of open-web programmatic display revenue is flat or lower than their pre-ruling 2025 levels at Q4 2027 earnings. Wrong if: either shows a clear share gain attributable to Prebid parity by then.

Judge recommends Google relax ad-tech rules, rejects DOJ breakup Read the source story →

Apostia will be acquired by a larger ad-tech, holding-company, or platform buyer before its stated agentic transaction volume ever reaches an annualized run rate large enough to matter, with the deal announced by the end of Q1 2027 earnings season in February or March 2027.

Why O'Kelley built and sold AppNexus, so he plays the buy-and-sell game well, and a small team on its second rebrand in two years with an investor already talking it up in public is positioning, not scaling. Agentic buying is the loudest theme in ad-tech right now, which means larger players want the capability and the credibility of a named founder faster than they want to build it, and acqui-hiring the team is cheaper than waiting for the category to mature. The opposite outcome, Apostia growing $1.3 million into real independent liquidity before anyone buys it, requires premium publishers to trust an unknown counterparty at scale faster than any new ad-tech entrant has managed in the last decade, which is the slower and less likely path.

Right if: Apostia announces an acquisition or majority-stake sale to a larger ad-tech, agency holding company, or platform buyer by 2027-03-31. Wrong if: Apostia remains independent with no announced sale by that date.

Scope3 Rebrands as Apostia, Reports $1.3M in Agentic Ad Transactions Read the source story →

At least one other independent location or foot-traffic measurement vendor among Placer.ai, Near, and Samba TV will announce an acquisition, take-private, or down-round financing by the end of Q2 2027, sold or funded on terms visibly below its prior peak valuation.

Why When the natural strategic buyers (The Trade Desk, LiveRamp, AppLovin) walk past a scarce location corpus and it clears to a second-tier consolidator, they've set the market price for every peer, and that price is a discount. Foursquare's clearing number becomes the anchor any investor or acquirer uses on Placer.ai, Near, or Samba TV in their next raise. The mechanism is straightforward: location attribution is being commoditized from below by cheaper vendors and from the side by retail media's own first-party purchase data, so the category's pricing power is falling while consolidation pressure rises. The opposite outcome, an independent location vendor raising a clean up-round, would require a buyer to ignore the comp Infillion just set, which is not how term sheets get written.

Right if: Placer.ai, Near, or Samba TV announces an acquisition, take-private, or financing on terms below its prior peak by the end of Q2 2027. Wrong if: all three stay independent and unfinanced, or if any of them raises a clean up-round above its prior valuation.

Infillion Acquires Foursquare in Ad-Tech Consolidation Move Full Analysis → Read the source story →

At least one other major US card issuer among Chase, American Express, and Capital One will publicly launch or expand a first-party ad or commerce-media offering by the end of Q1 2027 earnings season (through May 2027).

Why Citi just told the market that transaction data plus a demand-side pipe equals a 15% pilot lift, and card issuers watch each other's monetization moves closely because they compete for the same affluent cardholders and the same interchange-adjacent revenue. Chase already runs a media business, so expanding it is a small step, and Amex and Capital One sit on comparable data with the same incentive to find revenue that does not depend on lending margins. The opposite outcome, everyone staying quiet, would require issuers to ignore a competitor publicly claiming strong performance in a category with retail-media-scale growth, which runs against how fast this crowd copied retail media itself.

Right if: Chase, Amex, or Capital One announces a new or materially expanded first-party advertising or commerce-media product by the end of Q1 2027 earnings season. Wrong if: none of the three makes such an announcement in that window.

Citi Launches Commerce Media Ad Business Targeting 70M Customers Read the source story →

Magnite's and PubMatic's combined full-year 2026 revenue growth rates, reported in their Q4 2026 earnings calls in February 2027, will show no acceleration attributable to the Google remedy versus their 2025 growth rates.

Why The final order isn't due until October 2, 2026, appeals are expected, and full implementation is realistically late 2027 at earliest, so no incremental AdX-in-Prebid demand can reach these SSPs during any 2026 reporting period. The mechanism that would help them, publishers credibly threatening to leave Google's DFP ad server, isn't even mandated in the recommendation as it stands. For 2026 growth to accelerate on the back of this ruling, Google would have to voluntarily open its auctions ahead of any binding order, which runs against its stated intent to appeal and its long track record of conceding narrowly and slowly. The opposite outcome, a measurable Google-driven bump, would require enforcement to arrive years ahead of schedule.

Right if: neither Magnite nor PubMatic attributes accelerated 2026 revenue growth to the Google remedy on their February 2027 Q4 calls, and their 2026 growth rates don't exceed 2025. Wrong if: either posts faster 2026 growth and credits the AdX/Prebid parity changes.

ExchangeWire on Google's Ad Tech Rules, OOH Advertising, and John Lewis' YouTube Chatshow Listen to the episode →

By Meta's Q4 2026 earnings call in late January 2027, Meta will have launched a paid placement product that lets brands influence what Muse recommends or buys, tied to advertiser spend.

Why Meta gave Muse away free while quietly metering tokens, which means the compute per user is expensive and the payoff has to come from somewhere else. Meta's entire business, 98% of revenue, is advertising, and every consumer surface it has ever built (Feed, Stories, Reels, Marketplace) eventually became an ad slot. Paparo named the exact mechanism on the podcast: a product becomes "eligible to be bought by Muse if you advertise." The Amazon block confirms the value at stake, because Amazon will not let a rival's agent decide purchases inside its catalog. The opposite outcome, Meta running a neutral agent that ignores advertiser money and quietly cannibalizes its own ad business, would break every pattern in the company's history and leave the compute bill unpaid.

Right if: Meta has announced or launched any advertiser-paid mechanism to influence Muse's product recommendations or purchases by the Q4 2026 earnings call. Wrong if: Muse's recommendations remain free of paid placement and Meta has shipped no such product by then.

Episode 192: The PE Guy and Cameo CEO Steven Galanis Close Out Marketecture Live Chicago Full Analysis → Listen to the episode →

At least one more large US bank or card network (Chase, Capital One, or American Express) will publicly launch or expand a commerce media / advertising business by the Q4 2026 earnings season in February 2027.

Why Citi entering advertising signals that banks now see their transaction data as an ad asset, and card issuers sit on the same cross-merchant spending map that Citi is pitching as its edge. American Express already runs Amex Offers and Capital One has been building an ad and shopping unit, so the muscle exists and the incremental-margin math on data you already own is hard for a CFO to ignore once a peer moves first. The opposite outcome, everyone sitting still, is the less likely one because banks copy each other's fee-adjacent revenue fast and none of them wants to concede a new high-margin line to Citi uncontested. The risk to the call is timing: financial-services legal and privacy review is slow, so a launch could slip past February.

Right if: Chase, Capital One, or American Express announces a new or expanded commerce/advertising media business by the Q4 2026 earnings season. Wrong if: none of the three makes such an announcement in that window.

MadTech Daily: Citi Enters Advertising With New Ad Business; IAB Europe Finds Growing Momentum for Agentic Ad Buying Listen to the episode →

By the AWS re:Invent 2026 keynote in December 2026, at least one of the three major cloud or ad platforms with a shipped MCP ad server (Google, Meta, Amazon) will add a mandatory human-approval or spend-cap gate on autonomous campaign actions, walking back full autonomy from the current implementations.

Why Meta's current server already lets agents set budgets and launch campaigns with no approval gate, which means the first unauthorized spend or mass audience misconfiguration is a matter of when, not if, once volume ramps through late 2026. Platforms have a consistent pattern: ship the flashy autonomous capability, then bolt on approval rails the moment a customer eats a real loss and complains, because the reputational cost of a runaway agent lands on the platform. The opposite outcome, all three keeping fully autonomous execution untouched through year end, would require zero costly incidents across a fast-growing base of agent-driven spend, which is the less likely world given how the segment itself flags governance as the unsolved problem.

Right if: Google, Meta, or Amazon publicly adds a required approval step, spend cap, or confirmation gate to autonomous MCP campaign actions by then. Wrong if: all three still allow agents to create audiences, set budgets, and launch campaigns with no mandatory human checkpoint.

MCP Servers Emerging as Standard Interface Layer for AI Ad Buying Full Analysis → Read the source story →

At least one publicly traded consent or clean-room identity vendor (LiveRamp the most likely, as the largest public pure-play) will name CIPA litigation exposure or private-right-of-action risk as a demand driver on an earnings call by the Q2 2027 reporting season, ending in August 2027.

Why SB-690 keeps a private right of action alive with over 4,000 suits already filed, which turns first-party data and provable consent from optional into a legal defense. Vendors that certify consent provenance sell directly against that fear, and management teams reliably fold a live regulatory threat into their demand narrative once it starts showing up in pipeline. The opposite outcome, silence, is less likely because a public vendor with a litigation-driven tailwind has every incentive to name it for investors, and CIPA is concrete enough to cite. The risk to the call is timing: if appellate courts narrow the wiretapping theory first, the sales story softens before it reaches a transcript.

Right if: a public consent or clean-room identity vendor cites CIPA or California private-right-of-action litigation as a demand or pipeline driver on an earnings call by the end of Q2 2027 reporting. Wrong if: none does by then.

California SB-690 Keeps Private Right of Action for Online Tracking Full Analysis → Read the source story →

A financial regulator or court in France or the Netherlands will cite the German Meta scam-ad ruling in its own platform-liability enforcement action against an ad platform before the end of Q4 2027 (December 31, 2027).

Why The ruling's core move, that algorithmic control over ad distribution strips the DSA's "we didn't know" defense, is written on EU-wide law, not German-specific rules, so any DSA regulator can borrow it. France and the Netherlands already run active platform-enforcement agendas and have shown appetite for testing DSA scope. Regulators cite each other's reasoning precisely because it lets them skip building the argument from scratch, which is why cross-border citation is the normal path for a novel DSA theory rather than each country reinventing it. The opposite outcome, every EU regulator ignoring a fresh liability theory that hands them leverage over the largest platforms, runs against how these agencies actually behave. The one thing that could stall it is a fast appellate reversal in Germany that makes the ruling toxic to cite, which is why this is Medium, not High.

Right if: a French or Dutch regulator or court references the German Meta scam-ad ruling in a platform ad-liability action by then. Wrong if: no French or Dutch enforcement action or ruling cites it by that date.

German Court Rules Meta Liable for Scam Ads Under DSA Framework Full Analysis → Read the source story →

Magnite's full-year 2027 revenue, reported in February 2028, will be higher than its 2026 revenue, because the WarnerMount integration will not have replaced the programmatic rails Magnite runs for both companies by then.

Why Merging two immature CTV ad stacks (Neo is barely at scale, Paramount's programmatic build is still maturing) while servicing heavy debt is a two-plus-year job, so both sides keep leaning on the programmatic pipes they already run rather than a proprietary buy point that doesn't exist yet. Magnite monetizes CTV inventory for both Paramount and WBD today, and during integration chaos that volume holds or grows as buyers take the guaranteed reach and keep biddable budget where it performs. The opposite outcome, a clean unified stack that starves the middlemen inside 15 months, requires two struggling companies to out-execute their own debt load and talent exodus, which is the less likely path. The long-term walled-garden risk is real, but it lands after 2027, not during it.

Right if: Magnite's reported full-year 2027 revenue exceeds its 2026 revenue. Wrong if: 2027 revenue comes in flat or lower.

Combined Paramount-WBD Ad Tech Stack Faces Integration Challenge Full Analysis → Read the source story →

By the end of Q3 2026 earnings season (November 2026), at least two of Magnite, PubMatic, and The Trade Desk will publicly announce or demo an "agentic" buying or selling capability positioned as native to their existing platform.

Why Brian O'Kelley rebranding to Apostra puts a credible AppNexus founder openly on record that the current stack is a removable tax, and credible-founder framing is exactly what forces incumbents to respond in their own investor narrative. The mechanism is defensive positioning: public ad-tech companies cannot let a "we make you obsolete" story sit unanswered on earnings calls, so they ship an agentic layer as a feature to neutralize the narrative. The opposite outcome, silence from all three, would require these companies to ignore a competitive story during a period when every ad-tech deck already has an AI slide, which cuts against how they have handled every prior platform-threat narrative from cookies to retail media.

Right if: at least two of Magnite, PubMatic, or The Trade Desk announce or demo a named agentic buying or selling feature on or before their Q3 2026 earnings calls. Wrong if: fewer than two do so by then.

Brian O'Kelley Rebrands Startup as Apostra, Pivots to Agentic AI Advertising Read the source story →

Meta will introduce a fee or revenue share on agent-driven sales completed through Shopify stores by the time Meta reports Q4 2026 earnings in late January 2027.

Why Meta is spending real infrastructure cost to run agentic checkout for every Shopify store and charging nothing, which no company sustains once volume proves the channel works. The stated position ("no additional fees") and Meta's revealed incentive point in opposite directions: Meta monetizes every surface it controls, and an agent that decides which products a shopper sees is the most valuable shelf placement in commerce. The mechanism is simple. Once the agent drives measurable GMV, Meta charges merchants to rank inside it, the same way it charges advertisers to reach users it already owns. The opposite outcome, Meta running this free indefinitely, only holds if agentic checkout volume stays too small to bother monetizing, which would itself prove the Skeptic right and make the whole moat story collapse. Either Meta charges because it worked, or it stays free because it didn't.

Right if: Meta or Shopify announces any fee, take rate, or revenue share on agent-completed transactions by the Q4 2026 earnings call in late January 2027. Wrong if: agent-driven checkout on Shopify remains free of any Meta charge as of that call.

Shopify Integrates Meta's AI Shopping Agent for Agentic Checkout Full Analysis → Read the source story →

The merged Paramount and Warner Bros. Discovery will not offer advertisers unified cross-app frequency capping spanning Paramount+ and HBO Max inventory by the 2027 upfront (May 2027).

Why Paramount runs its own ad-tech stack and WBD runs Neo, and the episode states the integration path is undetermined at the moment the deal closes. Unifying frequency capping across two apps means reconciling two ad servers, two identity systems, and two inventory catalogs, which is the hardest part of any CTV stack merger and the part that routinely slips past a year. The seller's whole pitch is scale and simpler buying, so if unified capping were close they would lead with it; the silence says it isn't ready. The opposite outcome, a clean unified package inside roughly 18 months of close, would require an integration timeline faster than any comparable media-company ad-stack merger has hit.

Right if: We're right if, going into the 2027 upfront, buyers still purchase Paramount+ and HBO Max inventory as separate deals without a single cross-app frequency cap. Wrong if: the merged company sells one package with unified frequency management across both apps by then.

Buying CTV In The WarnerMount World Listen to the episode →

No major retail media network (Walmart Connect, Amazon Ads, Kroger Precision Marketing, or Albertsons Media Collective) will replace CPM or CPC pricing with a cost-per-incremental-dollar model as its standard buying unit by the Q4 2026 earnings season in February 2027.

Why Leach predicts retail media networks face "repricing pressure" once true incrementality is exposed, and he's got an obvious reason to say so, because Ibotta sells the incrementality product that would replace them. The mechanism that kills his prediction is money: CPM and CPC pricing charges the advertiser for impressions and clicks that would have converted anyway, and a cost-per-incremental-dollar model charges only for sales the ad genuinely caused, which is a much smaller, much less flattering number. No retailer voluntarily switches to the pricing model that shrinks its own revenue and admits most of its ad load was padding. The opposite outcome, a major RMN adopting incremental pricing as standard, would require Walmart or Amazon to reprice their fastest-growing profit line downward on purpose, and nothing in their incentives points that way.

Right if: none of Walmart Connect, Amazon Ads, Kroger Precision Marketing, or Albertsons Media Collective has made cost-per-incremental-dollar its standard buying unit by the Q4 2026 earnings season. Wrong if: any one of them has.

From Cash Back to Outcomes: Bryan Leach on Building Ibotta, Going Public, and Betting on Denver Full Analysis → Listen to the episode →

No AI chatbot maker (OpenAI, Anthropic, Google, Meta, xAI, or Character.AI) will ship a product that removes first-person "I/we" speech from its consumer-facing model as a safety measure before OpenAI's next flagship model release after GPT-5.

Why Turkle names first-person speech as the manipulation, and a reader might expect safety-conscious labs to respond. They won't, because the "I" is what makes a chatbot feel like a companion, and companion-style engagement is what drives retention and, for Meta and Character.AI, the entire product thesis. Zuckerberg is on record wanting AI avatar friends to fill the gap between the 2-3 friends people keep and the 15 they want. Removing first-person voice would gut that. Labs will respond to this pressure with guardrails, disclaimers, and age-gating, which cost nothing structural, not by changing the voice that makes the product work. The opposite outcome would require a lab to sacrifice its main retention lever ahead of any court forcing it to, and no verdict has landed yet to force the choice.

Right if: every major consumer AI chatbot still speaks in the first person by that date. Wrong if: any of the named labs ships a consumer model that deliberately drops "I/we" self-reference citing psychological safety.

Sherry Turkle Doesn't Buy the Promise of a Perfect Machine Listen to the episode →

Apostra will not have a paying publisher or buy-side customer running live agent-to-agent ad negotiation against real budgets by the Q1 2027 programmatic earnings season (February 2027).

Why The episode can't even confirm what company Apostra is, which tells you how early this is. Agent-to-agent negotiation fails on the same thing every programmatic disruption fails on first: nobody hands price-setting to a counterparty's bot without an audited floor and a way to check the deal, and none of that exists here. Publishers guard their rates precisely inside the manual, fragmented process O'Kelley wants to remove, so the incentive to adopt runs backwards for the side he needs first. The opposite outcome, a real live deal in under six months, would require a publisher to trust an unbranded startup's agent with live rate negotiation faster than any prior programmatic tool ever earned that trust, and that is not how sell-side adoption moves.

Right if: no named publisher or advertiser is publicly running live agent-negotiated buys through Apostra by the February 2027 earnings cycle. Wrong if: Apostra names a real customer transacting real budgets through agent-to-agent negotiation before then.

MadTech Daily: Shopify Taps Meta’s Muse for Agentic Checkout; Brian O’Kelley’s Startup Makes AI Agent Pivot Listen to the episode →

Google's referral traffic to third-party publishers, as measured by Similarweb's aggregate publisher-traffic data, will show a year-over-year decline for the fourth quarter of 2026 when that data publishes in January 2027.

Why Google earns nothing on an outbound click to a publisher; that click was always a cost it absorbed to keep search useful. AI Overviews already suppress click-through by answering the query on-page, and now the follow-up links route into AI Mode instead of out to sites, so each new AI surface removes another reason to leave Google. The mechanism is one-directional: once a user gets the answer plus the next question without clicking through, that referral does not come back. The opposite outcome, referral traffic rising, would require Google to reverse a change that costs it nothing and earns it retained engagement, which runs against its own economics.

Right if: Similarweb's Q4 2026 publisher-referral data shows Google-sourced traffic down versus Q4 2025. Wrong if: it's flat or higher.

Google routes AI Overview links into AI Mode, bypassing publisher sites Full Analysis → Read the source story →

The Trade Desk will not announce a live, buyable ChatGPT Ads integration by its Q4 2026 earnings call in February 2027.

Why Every mid-tier partner has rushed in because the cost of joining is an NDA, but The Trade Desk, the largest independent buying platform, has stayed out even after hiring OpenAI's own new partnerships lead away from itself, which means the barrier isn't relationships, it's commercial. Jeff Green doesn't skip addressable audience at this scale unless the take-rate terms squeeze his margin or the audience data access is too restricted to run the performance campaigns his buyers demand. OpenAI is optimizing for partner count right now, not for handing a rival buying platform favorable data terms, so the two sides' incentives point apart. The opposite outcome, a signed deal within months, would require OpenAI to give TTD data access it hasn't given anyone, and nothing in this announcement wave suggests that's on the table.

Right if: The Trade Desk has no live, buyable ChatGPT Ads integration announced by its Q4 2026 earnings call. Wrong if: TTD announces a live ChatGPT buying integration before then.

ChatGPT Ads Rapidly Expands Programmatic Partner Ecosystem Full Analysis → Read the source story →

By the close of Q2 2027 earnings (August 2027), at least two additional US retail media networks beyond Dollar General will publicly announce unified onsite-plus-offsite buying built on an owned ad server paired with The Trade Desk.

Why The signal in this story is that the split-stack design, own the on-site server through a vendor like Kevel and route off-site through The Trade Desk, is now live and being marketed as a repeatable pattern rather than a bespoke project. The mechanism is straightforward: building an in-house buying platform to rival The Trade Desk is a capital sink no tier-two retailer wins, while an owned on-site server protects the margin retail media contributes to the P&L, so every retailer running fragmented onsite and offsite tools faces the same board question Dollar General just answered. The opposite outcome, no retailer follows, would require mid-market networks to keep eating reconciliation overhead they've already admitted is a "dirty secret," which cuts against their own cost pressure.

Right if: two or more US retail media networks announce unified onsite/offsite buying pairing an owned ad server with The Trade Desk. Wrong if: fewer than two do, or if the copycat deals route off-site through Amazon's platform or an in-house DSP instead.

Dollar General Unifies Onsite and Offsite Retail Media via Trade Desk and Kevel Full Analysis → Read the source story →

At least one major retail media network among Walmart Connect, Amazon Ads, and Kroger Precision Marketing will report full-year 2026 ad revenue growth above 20% when 2026 results are published by the end of Q1 2027 earnings season.

Why Mallard's public arc from feeling "forced to allocate" to calling retail media efficient and measurable is the mid-market advertiser most likely to have resisted, and he has stopped resisting. When the reluctant buyers convert and defend the spend themselves, the category stops relying on new-logo persuasion and grows on renewal plus expansion, which is more durable than a hype cycle. Amazon's ad business and Walmart Connect have printed 20%-plus growth on far larger bases than the snack-brand budgets now flowing in, so the incremental mid-market dollars land on top of an already-fast line. The opposite, growth dropping below 20% across all three, would require the reluctant buyers to reverse the exact conversion Mallard just described, and nothing in the demand picture points that way.

Right if: at least one of Walmart Connect, Amazon Ads, or Kroger Precision Marketing posts full-year 2026 advertising revenue growth above 20%. Wrong if: all three come in at or below 20%.

J&J Snack Foods CMO Lynwood Mallard on the art of keeping the 60-year-old ICEE brand cool Full Analysis → Listen to the episode →

At least one channel-neutral independent DSP or ad-tech platform (The Trade Desk, Criteo, or Viant) will publicly announce a purpose-built Amazon-DSP or retail-media agentic buying feature by the Q3 2027 earnings season in October 2027.

Why Amazon's ad business at almost $80 billion growing 20-plus percent a year is the fastest-scaling programmatic channel, and Epstein's whole pitch rests on agencies being unable to staff it with humans. That staffing gap is not unique to Gigi's customers; it hits every buyer, which means the channel-neutral platforms that sell to those same agencies will get the same "help me run Amazon" request. The independents already build channel-specific tooling to stay relevant, so adding an Amazon-DSP agent is the obvious defensive move once a well-funded specialist like Gigi shows demand exists. The opposite outcome, everyone staying purely horizontal, loses because horizontal breadth without depth in the channel where the money is growing fastest is exactly the weakness Epstein identified, and buyers will vote with their spend.

Right if: The Trade Desk, Criteo, or Viant announces a named Amazon-DSP or retail-media agentic buying tool by the Q3 2027 earnings season. Wrong if: none of the three ships such a feature by then.

Hiring Gigi, Your AI Media Manager Full Analysis → Listen to the episode →

At least one other major US retail media network among Walmart Connect, Target Roundel, and Kroger Precision Marketing will publicly restrict or set formal terms for third-party AI shopping agents accessing its platform before the end of 2026 holiday-season reporting in February 2027.

Why Amazon blocked Meta's Muse because an autonomous agent that filters straight to a purchase skips the sponsored-product search results that Amazon sells to advertisers, which is the core of its retail media revenue. Every other retail media network monetizes the same way, so every one of them faces the same threat the moment shopping agents reach real volume, and Amazon has just handed them the template and the cover story ("credential security"). The opposite outcome, everyone staying silent, requires believing that agents stay too small to matter through the holidays and that no rival wants to plant a flag on the issue Amazon just made public, which is the less likely bet once a market leader has acted first.

Right if: Walmart, Target, or Kroger's retail media arm publicly restricts or publishes formal access terms for third-party AI shopping agents by then. Wrong if: all three stay silent on agent access through that date.

MadTech Daily: Paramount Clears Legal Hurdle Ahead of Warner Merger; Amazon Bars Meta’s Muse From Agentic Shopping Listen to the episode →

By the end of Q2 2027, no top-tier SSP (Magnite, PubMatic, or Index Exchange) will publicly market asymmetric AdX floor-pricing as a publisher feature, because none will accept the cost to their Trade Desk relationship.

Why The exchanges make their money on volume flowing through their pipes, and The Trade Desk is one of the largest independent sources of that volume. TTD has already said, per Scott Messer, that it dislikes floor differentiation across and within exchanges. An SSP that builds and markets tooling explicitly designed to let publishers floor AdX higher is picking a fight with the buyer whose spend it depends on, for the benefit of publishers who mostly lack the tooling or spend concentration to use it well. The safe commercial move is to quietly enable the capability for the handful of publishers who ask while never making it a public selling point that TTD would read as a declaration. The opposite outcome, an SSP planting a flag on publisher pricing autonomy, would require betting that publisher loyalty outweighs TTD spend, and the volume math doesn't support that bet yet.

Right if: none of Magnite, PubMatic, or Index Exchange has publicly launched or marketed a floor-differentiation product framed around charging AdX a higher floor. Wrong if: any of the three publicly ships or promotes such a feature as a publisher benefit.

The Trade Desk Signals Discomfort With Asymmetric Floor Pricing Under New Rules Full Analysis → Read the source story →

Neither Magnite nor PubMatic will report a full-year 2026 revenue growth rate above 15% when they release 2026 results in February 2027, because the publisher ad-server fragmentation their growth story assumes will not materialize from these remedies.

Why Investors have priced mid-tier exchanges Magnite and PubMatic partly on the bet that a cracked-open Google forces publisher ad-server fragmentation, which opens new paths for ad demand to reach inventory and flows to independent exchanges. These 4/10 remedy scores say that fragmentation isn't coming: data portability has no rival ad server to receive the data, and unbundling adds no price competition because no competitor exists. Without that structural tailwind, both companies are left growing on CTV and existing supply-path deals alone, which is real but not enough to clear 15%. The opposite outcome, a growth surge above 15%, would require a demand-path windfall the remedy structurally cannot deliver in this window.

Right if: both Magnite and PubMatic report full-year 2026 revenue growth of 15% or less on their Q4 2026 earnings calls. Wrong if: either one reports full-year 2026 revenue growth above 15%.

Data Portability and AdX/DFP Unbundling Both Score Low at 4/10 Full Analysis → Read the source story →

The final Google ad-tech remedy that opens HBT to all publishers will not grant rival SSPs or publishers a court-enforced right to audit bid-level auction logs inside Google's Prebid execution, when Judge Leonie Brinkema issues her remedies ruling.

Why The remedy as described and graded is about access to the trafficking tool, and the executives themselves flagged the black-box Prebid execution as the unaddressed gap, which means it isn't currently in the proposed remedy. Google's entire history is protecting its informational edge in auctions it runs, so it will not volunteer transparency it wasn't ordered to give, and courts crafting behavioral remedies tend to mandate the headline structural fix (open access) rather than the granular operational plumbing (per-auction log rights) that requires ongoing technical supervision. The opposite outcome, a court writing bid-level audit rights into the order, would require a party to have pushed hard for it in the record, and the framing here is a 7/10 celebration of access, not a fight over logs.

Right if: Brinkema's ad-tech remedies ruling opens HBT access without a court-enforced bid-level auction-log audit right for rival exchanges or publishers. Wrong if: the ruling mandates auditable per-auction bid logging inside Google's Prebid execution.

Header Bidding Access for All Publishers Rates 7/10, With Transparency Caveats Full Analysis → Read the source story →

At least one of Comscore, VideoAmp, or iSpot.tv will announce an acquisition, take-private, or majority-stake sale by the end of Q1 2027 earnings season (roughly May 2027).

Why The JIC just validated three subscale measurement companies competing against a weakened incumbent, which is an unstable market structure that resolves through consolidation, not through three of them thriving independently. VideoAmp is private and cash-dependent, Comscore is the most acquirable public name at its current size, and iSpot is a clean tuck-in for a larger data or ad platform wanting certified measurement it can own. The buyer logic is straightforward: certification plus publisher relationships bought in one move beats building either from scratch, and a strategic acquirer reconciling three schemas would rather own one. The less likely outcome is all three staying independent and fully funded through a full year of upfront competition, because the money to sustain that many neutral-currency players against Nielsen simply isn't there for the smallest of them.

Right if: any of the three announces an acquisition, take-private, or majority-stake sale by then. Wrong if: all three remain independent with no controlling-stake transaction announced.

US JIC Recertifies Comscore, VideoAmp, and iSpot as Nielsen Alternatives Full Analysis → Read the source story →

No AI creative-generation vendor selling to brand advertisers will reposition its core product pitch from "scale and volume" to "authenticity and human finishing" before the 2027 upfront presentations in May 2027.

Why The only evidence here is one panelist's line, "they just know," with zero brand-lift or conversion data behind it, and vendors do not rebuild a pitch on a conference quote. The economic pull runs the other way: generative vendors win deals today on cost-per-asset and output speed, and that message is landing with mid-market buyers who have budget pressure, not authenticity budgets. A repositioning to "authentic scale" is a harder, slower product problem that undercuts the vendor's own price advantage, so the rational move is to keep selling volume until a customer shows churn data that forces the change. The opposite outcome, a real repositioning, would require someone to publish performance splits proving younger audiences convert worse on AI creative, and that number does not exist in this story.

Right if: the major AI creative-gen vendors (Adobe Firefly, Typeface, Jasper, Runway) still lead their brand-advertiser marketing with volume, speed, or cost-per-asset going into the 2027 upfronts. Wrong if: any of them makes "human-finished" or "authenticity" the headline claim of its brand-advertiser product before May 2027.

Younger Consumers Already Distrust AI-Generated Content, Panelist Warns Full Analysis → Read the source story →

No US antitrust complaint naming Google's connected-TV ad business will be filed by the Department of Justice or the FTC before the end of 2027.

Why The only thing behind this "next case" story is an anonymous source extrapolating from the display ruling, with no filing, no named theory of harm, and no regulator on record. CTV supply is split across Roku, Samsung, Amazon, and the streamers, so the single-dominant-pipe argument that carried the display case is much harder to make here. Antitrust cases against Google's display stack took the better part of a decade from investigation to remedy; starting a CTV case from a standing start puts any complaint well past 2027. The opposite outcome would need a regulator to open, build, and file a novel monopolization case in under 18 months, which is faster than any recent ad-tech action has moved.

Right if: no US federal antitrust complaint specifically targeting Google's CTV or streaming-video advertising business has been filed by 2027-12-31. Wrong if: the DOJ or FTC files such a complaint before that date.

Expert Warns CTV Is 'The Next Case Waiting to Happen' Post-Google Remedies Full Analysis → Read the source story →

By the 2027 upfront season, at least one major measurement or verification vendor (DoubleVerify, IAS, or iSpot) will launch and market a productized "agent-ready" or agent-verification data offering aimed at feeding autonomous buying tools auditable inputs.

Why This story shows eight practitioners agreeing that agents act confidently on flawed inputs, which turns "verified, auditable signal" into the thing that makes autonomy safe to deploy. Measurement vendors already sell exactly that raw material and are hunting for growth stories as their core verification businesses mature. Repositioning existing signal as "agent-ready" costs them almost nothing and lets them ride the agent hype instead of being disintermediated by it. The opposite outcome, all three staying quiet, would require them to watch a new buying layer form on top of their data without planting a flag, which cuts against how aggressively this category chases every adjacent narrative.

Right if: DoubleVerify, IAS, or iSpot has publicly launched and named a product or data offering specifically positioned for AI buying agents or agent verification by then. Wrong if: none of the three has shipped or marketed such an offering by that date.

AI Buying Agents Seen as Confidently Wrong 'Regularly,' Experts Warn Full Analysis → Read the source story →

Viant will keep the "measures the market" positioning in its measurement marketing through its Q4 2026 earnings call in February 2027, rather than soften it in response to the independence criticism.

Why Viant paid for TVision to sell an industry-standard measurement story, and that story is what carries a richer valuation than a buying-tools story. Backing off "measures the market" would concede the strategic reason the deal existed, so the incentive runs hard toward keeping it even after Nardone named the conflict. A single trade-press quote, however clean the logic, does not outweigh the positioning the acquisition was built to support. The opposite outcome, a quiet retreat, would mean Viant paid up for a claim and abandoned it within a quarter under LinkedIn pressure, which management almost never does. The interesting consequence is that keeping the language leaves the "certified independent" wedge sitting on the table for iSpot, VideoAmp, and Comscore to pick up through the H2 planning cycle.

Right if: Viant's measurement marketing and executive commentary through the February 2027 earnings call still frames the company as measuring the whole market rather than just its own campaigns. Wrong if: Viant drops or materially softens the "measures the market" positioning in that window.

John Nardone of JWX: Viant Measurement 'Absolutely Not' Independent Full Analysis → Read the source story →

At least one of the six largest agency holding companies (Omnicom, WPP, Publicis, Dentsu, Havas, or the merged Omnicom-IPG entity) will introduce a contractual requirement for third-party verification of AI buying-agent CPA or ROAS outcomes before its next major autonomous-buying renewal cycle, disclosed in trade press or an agency statement by 2027-06-30.

Why Vanderhook wrote the rule and the buy-side now has public language to enforce it, which is exactly the kind of ammunition procurement uses to add clauses at renewal. Agency trading desks are under fresh P&L scrutiny after the holdco consolidation, and self-reported agent outcomes are the newest unaudited liability on their books, the same category as viewability, which they already require third-party checks on. The opposite outcome, nobody codifying it, is plausible only if buyers keep renewing on inertia, but the incremental-sales problem Mike Follett named gets more expensive to ignore as more spend flows through agents that grade themselves.

Right if: a top-six holding company publicly requires third-party verification of AI buying-agent outcomes in its trading-desk contracts. Wrong if: no holdco has introduced such a requirement by that date.

Viant's AI Agent Accused of Grading Its Own Homework Read the source story →

No network selling NFL inventory will sign a guaranteed-audience ad deal denominated in a non-Nielsen currency for the 2027 season, as visible in the reporting around the 2027 upfront cycle (May to June 2027).

Why The NFL publicly endorsed a two-year timeline for fixing Nielsen and said it's expanding work with rivals, which reads as a migration but functions as renewal leverage. Guarantees are where measurement currency becomes real, and writing one against VideoAmp, iSpot, or Comscore forces buyers to transact on a number the seller prefers, which agencies resist because a favorable number from the seller is a number they can't defend to a client. Every prior Nielsen methodology change got absorbed with a restatement and shrugs rather than a currency switch, and nothing in the season's chaos changes the buyer's need for one defensible number. The opposite outcome requires the NFL to accept the operational cost of dual-currency guarantees mid-crisis, which the league itself says is two years out.

Right if: NFL networks sell 2027 guarantees on Nielsen, with rival measurement present only as supplemental comps in decks. Wrong if: any network signs a guaranteed-audience NFL deal transacted on a non-Nielsen currency.

Nielsen Methodology Chaos Makes NFL Audience Comparisons Unreliable Read the source story →

No top-five global advertiser or major holding company will publicly run live cross-platform media budgets shifted autonomously by an AI agent across Google, Meta, and Amazon, with no human approving each shift, by the end of the 2027 upfront negotiating season in June 2027.

Why Every major platform has shipped an MCP server and Meta already lets agents set budgets in a controlled setting, so the technical path exists. But Basis's own Mark MacAchren says autonomous cross-platform budget shifting is "not the standard practice yet," and the episode itself documents why the trust isn't there: creator pricing runs on follower count, which doesn't predict outcomes, and buyers get no post-campaign feedback to check the machine's work. Agencies get paid to be accountable for where money goes; handing an auditless agent the authority to move seven-figure budgets across three rival platforms removes the human they can point to when a campaign misses. The interface consolidation will race ahead because it cuts cost with little risk, but the money-moving autonomy stays a demo until measurement can attribute an agent's decision, and nobody in this episode claims it can.

Right if: no top-five advertiser or major holding company has gone on record running live, human-out-of-the-loop cross-platform budget reallocation via AI agents by then. Wrong if: any of them publicly confirms live autonomous budget shifting across at least two of Google, Meta, and Amazon.

Creator Marketing Is Growing Up Full Analysis → Listen to the episode →

Infillion will narrow or end Foursquare's open data-selling to direct competitors by the time of the 2027 upfront selling season (May 2027), replacing the "sells to competitors" posture with preferred or exclusive access inside Infillion's own stack.

Why Infillion paid for Foursquare to differentiate its own measurement product, and the only durable differentiation is data a rival can't get on the same terms. Right now keeping the data open costs nothing because integration isn't done, so the pledge is free to make. Once the location-plus-purchase product is actually stitched together and Infillion needs it to command a premium, selling the same signal to competitors undercuts the exact advantage they bought. The opposite outcome, staying fully open, only wins if Foursquare's network effects generate more revenue from rivals than exclusivity would from Infillion's own margin, and mid-tier data businesses under signal pressure from Apple and Google rarely have that luxury.

Right if: by the 2027 upfront season Infillion has restricted, tiered, or ended Foursquare's data sales to named ad-tech or measurement competitors, or moved its best signal into an Infillion-only tier. Wrong if: Foursquare's data remains sold to direct competitors on the same open terms it does today.

MadTech Daily: Infillion Acquires Foursquare; McCann Merges FP7 & MullenLowe in MENA Infillion Acquires Foursquare; McCann Merges FP7 & MullenLowe in MENA Full Analysis → Listen to the episode →

By Disney's fiscal Q3 2027 earnings call (reported August 2027), at least one other major US streaming service among Peacock, Paramount+, HBO Max, and Netflix will update its terms or product to place ads on a tier it currently markets as ad-free.

Why Disney just wrote into a legal document that "ad-free" can carry pre and post-roll ads, and once one major redefines the phrase, the competitive pressure to match runs one direction only, because ad-free subscribers are the highest-income, most valuable audience any streamer owns and leaving that inventory unsold is money on the floor while rivals grab it. Streaming ad revenue is the growth line every one of these services is under board pressure to expand, and adding supply from an existing subscriber base costs nothing to acquire. The opposite outcome, everyone holding a pure ad-free tier, only happens if Disney's UK subscribers cancel hard enough to scare the rest off, and price-inelastic premium subscribers historically absorb small ad loads rather than churn. The reason to doubt is timing, not direction: a bad UK reaction delays the copycats past the revisit window.

Right if: Peacock, Paramount+, HBO Max, or Netflix updates terms or product to run ads on a currently ad-free tier by then. Wrong if: all four keep their premium tiers genuinely ad-free through that date.

Disney+ Updates Terms to Allow Ads on Premium Ad-Free Plans Full Analysis → Read the source story →

Within 12 months of the Paramount-Skydance and Warner Bros. Discovery deal closing, the combined company will make Max and Paramount+ streaming inventory available through at least one major programmatic pipe (The Trade Desk, Magnite, or FreeWheel) on materially more open terms than either offered independently before the merger.

Why The combined entity closes with over $80 billion in debt and interest due immediately, and ad revenue is the fastest cash lever it has. A company that needs cash now opens inventory it previously kept scarce or direct-sold, because programmatic access widens the buyer pool overnight and fills unsold slots. Both Paramount and WBD have historically guarded premium streaming supply to protect direct-sold prices; the debt schedule removes the luxury of that discipline. The opposite outcome, keeping supply tight to protect pricing, requires a patience that $80 billion in debt does not permit.

Right if: We're right if, within a year of close, Max or Paramount+ premium video inventory is transacting through a major CTV programmatic platform on broader terms than pre-merger. Wrong if: the combined company keeps that inventory direct-sold or restricted at the same or tighter access levels than before.

Update: Paramount-Skydance Settles Antitrust Lawsuit, Clears Path to Warner Bros. Discovery Acquisition Full Analysis → Read the source story →

Google's Prebid integration required by the antitrust remedy will ship covering only open web display, with video, programmatic deals, and retail media inventory excluded, and no DV360 Prebid adapter will be live by the time Google reports Q4 2026 earnings in early February 2027.

Why The ruling targets open web display, which gives Google a written basis to exclude video, deals, and retail media, and Garett McGrath's "very light lift" language signals a minimal-scope build, not an expansive one. DV360, Google's demand-side buying platform, was not covered by the ruling at all, so there is no legal force compelling a DV360 adapter and no commercial reason for Google to build one voluntarily against its own interest. The opposite outcome, Google voluntarily including its highest-value inventory formats and wiring up DV360 when neither is required, asks a company to surrender auction leverage it fought years to keep, which is the least likely thing it does.

Right if: Google's shipped Prebid integration excludes at least one of video, deals, or retail media, and no DV360 Prebid adapter is live as of the Q4 2026 earnings report. Wrong if: the integration covers those formats or a DV360 adapter is live by then.

Google Compelled to Integrate with Prebid for First Time Read the source story →

Magnite's curation revenue line, reported on its Q1 2027 earnings call in February 2027, will be higher year-over-year, as sell-side curated deals absorb budget fleeing cookieless inventory.

Why Publishers are losing 41% on unaddressed CPMs today, which is a bleeding-now problem that yield teams have to answer this quarter, and curation is the sell-side tool built to answer it. The Permutive-TransUnion deal is one instance of a broader move: identity graphs are being wired directly into sell-side curation so audiences stay targetable without cookies, and that budget flows through SSPs' curation lines. Magnite runs the largest independent curation business among public SSPs, so if the category grows the money lands there first. The opposite outcome, curation revenue falling, would require buyers to abandon curated deals and retreat to walled gardens even as open-web CPMs stay depressed, which cuts against the revenue math driving the whole shift.

Right if: Magnite reports year-over-year growth in its curation-related revenue on the Q1 2027 call. Wrong if: that line is flat or lower than the prior year.

Permutive-TransUnion Integration Targets Cookieless Audience Gap Read the source story →

By the end of Q1 2027 earnings season (the February 2027 reporting calls), no major independent supply-side platform, meaning Magnite, PubMatic, or Index Exchange, will announce a direct supply partnership to sell ads inside a top-five consumer AI chat product (ChatGPT, Gemini, Copilot, Perplexity, or Grok).

Why OpenAI, the largest consumer AI product, chose Amazon's closed buying platform over the open market for its first ad inventory, and Amazon paired it with proprietary commerce data no independent can match. AI publishers picking a partner want two things independents can't offer: buyer demand already in hand and purchase data to close the loop, which is exactly what Amazon, Google, and Microsoft bring and Magnite or PubMatic do not. The opposite outcome, an AI publisher handing supply to a neutral SSP, would require it to walk away from the walled gardens' demand and data advantages in exchange for pipes that are commodity by design, and nothing in how OpenAI just chose suggests the next one thinks differently.

Right if: no top-five consumer AI chat product has publicly named Magnite, PubMatic, or Index Exchange as an ad supply partner by the close of February 2027 earnings calls. Wrong if: any one of the three announces such a deal before then.

Amazon Opens ChatGPT Ad Inventory via Its DSP Full Analysis → Read the source story →

At least two more independent ad-tech vendors with under $50M in annual revenue and a native, contextual, or content-recommendation core will announce a sale or take-private to a larger operator or PE roll-up by the end of Q2 2027 earnings season in August 2027.

Why Dianomi told the market directly that AI summaries and zero-click search are draining the publisher traffic its widgets need, and it still cleared at a distressed £19M with most of the value pushed into an earn-out, which is how buyers price revenue they don't trust to hold. Foursquare going to Infillion in the same window shows the pattern isn't a one-off. When a category's demand source shrinks faster than its players can grow out of it, the mid-tier gets bought by scaled operators and PE shops chasing earnings, because staying independent means watching the multiple fall every quarter. Vendors holding out for better prices face a simple problem: waiting makes the number worse, so the incentive runs toward selling sooner.

Right if: at least two ad-tech vendors under $50M in revenue with a native, contextual, or content-recommendation business announce a sale or take-private between now and the end of Q2 2027 earnings season. Wrong if: fewer than two such deals are announced in that window.

Taboola Acquires Dianomi for up to £27M Amid AI Traffic Pressure Full Analysis → Read the source story →

Neither Meta nor Google will add a contract term, by their next major advertiser terms update, that promises a brand's creative-performance patterns won't inform other advertisers' automated results in Advantage+ or PMax.

Why Advantage+ and PMax work by pooling signals across advertisers to find what performs, so a promise not to use one brand's patterns for another would weaken the exact machinery that makes those tools convert. Collmer, who sits on Meta's own Global Creative Council, still can't point to such a guarantee, which suggests it doesn't exist and isn't coming. Both platforms compete on automated performance, and voluntarily narrowing what their models can learn from cuts against that. The opposite outcome, a public per-advertiser data-isolation pledge, would mean Meta or Google giving up a measurable performance edge with no regulator forcing them to, and nothing here shows that pressure.

Right if: neither Meta's nor Google's advertiser terms add a clause isolating one advertiser's creative-performance learnings from others' automated campaigns. Wrong if: either publishes such a clause.

When Content Becomes Infinite: Alex Collmer on Creative Intelligence in the AI Advertising Era Full Analysis → Listen to the episode →

By September 30, 2027, at least one of the 50 largest US publishers by audience will publicly announce that it has moved its primary ad server off Google Ad Manager.

Why The court-ordered remedy requires Google's ad exchange to bid into rival auctions on the same terms it gives its own ad server, which removes the main reason publishers stayed on Google Ad Manager: fear of losing Google's demand. Rival ad servers from Kevel, Equativ, and Magnite have been waiting for exactly that opening, and a large publisher that switches gets a public story about independence and lower fees at the moment the trade press is watching for one. The opposite outcome, every large publisher staying put through 2027, would require the remedy to change nothing about where Google's demand flows, and the whole point of the order is to change that.

Right if: a top-50 US publisher by Comscore audience publicly announces a primary ad-server move off Google Ad Manager on or before September 30, 2027. Wrong if: no such announcement by that date.

The Remedies Episode Full Analysis → Listen to the episode →

Pinterest will report Q1 2027 revenue at or above the top of its own guidance range when it reports in late April or early May 2027.

Why Companies raise guidance when the booked pipeline already supports it, and Pinterest just lifted its range on the back of 18% growth and rising Performance+ adoption. Advertisers fund a new test by moving budget from the weakest existing line, and Pinterest is currently the test that keeps getting funded. The opposite outcome, Pinterest missing a number it just raised, would require demand to fall off inside a single quarter right after management signaled confidence.

Right if: Pinterest's reported Q1 2027 revenue is at or above the top of the guidance range it gave for that quarter. Wrong if: Pinterest's reported Q1 2027 revenue is below the top of that range.

Pinterest Reports 18% Revenue Growth, Raises Q3 Guidance Full Analysis → Read the source story →

Snap's North America average revenue per user for Q1 2027, reported in late April or early May 2027, will be lower than its Q4 2026 figure.

Why Pinterest and Snap compete for the same social budget that is not going to Meta or Google, and agencies fund a new test by cutting the weakest existing prospecting line first. Snap's North America monetization has been the soft spot in its recovery story, which makes it the natural donor account when budgets shift toward Pinterest's Performance+ products. The opposite outcome, a clear sequential gain in North America revenue per user, would require Snap to win back prospecting budget from the platform advertisers are currently expanding.

Right if: Snap's reported North America average revenue per user for Q1 2027 is below its reported Q4 2026 figure. Wrong if: the Q1 2027 figure is equal to or above the Q4 2026 figure.

Pinterest Reports 18% Revenue Growth, Raises Q3 Guidance

Pinterest's Q3 2026 revenue, reported in late October 2026, will grow at least 18% year over year, matching or beating its Q2 2026 growth rate.

Why Pinterest grew revenue 18% in Q2 2026 and raised guidance, with management crediting its Performance+ automated ad products for pulling in performance budget. Advertisers looking for growth outside Meta and Google are spreading commerce and search-adjacent spend across second-tier platforms, and Pinterest is the one with a shopping intent signal, so it collects a disproportionate share of that money. Adoption of an automated product compounds quarter to quarter as more advertisers switch it on, so growth holds or accelerates before it slows. The opposite outcome, growth slipping below 18% one quarter after a guidance raise, would require the budget shift to reverse inside three months.

Right if: Pinterest reports Q3 2026 year-over-year revenue growth of 18% or more. Wrong if: Pinterest reports Q3 2026 year-over-year revenue growth below 18%.

Pinterest Launches Visual Search Ads and Expands AI Ad Suite Full Analysis → Read the source story →

By 31 May 2027, one of the 25 largest US advertisers will be named as a defendant in a US class action claiming that its own tracking or measurement code, running on another company's video pages, illegally shared what viewers watched.

Why US privacy suits have produced roughly $7 billion in settlements since 2022, with 3,414 cases filed in 2025 alone, almost all of them built on the 1988 Video Privacy Protection Act, which bars sharing what someone watched without separate written consent and carries $2,500 per violation. Courts have started treating a website's analytics and advertising vendors as extensions of the website itself, which points the same logic back up the chain at the brand whose pixel fired. The plaintiff bar follows the deepest pocket, and a consumer-goods or retail advertiser with a national brand is a far richer target than a mid-tier publisher paying $50 a head to settle. The opposite outcome requires plaintiff firms to keep suing only publishers and pixel vendors even as their own theory of liability now reaches the advertiser, and to leave the largest available defendants untouched for another eight months.

Right if: A complaint on file in a US federal or state court by 31 May 2027 names a company among the 25 largest US advertisers by measured spend as a defendant over video-watch data collected by its own tracking, analytics, or measurement code on a third party's site or app. Wrong if: As of 31 May 2027, every such video-privacy class action still names only publishers, platforms, or ad-tech vendors as defendants, with no top-25 US advertiser sued over its own code.

Ad-Tech Privacy Litigation Tops $7B in Settlements Since 2022 Full Analysis → Read the source story →

The combined US ad-revenue share of Google, Meta, and Amazon will be higher in 2026 than in 2025.

Why The concentration figure moved from 53 to 56 cents of every US ad dollar in a single year while open-web infrastructure spending rose, which shows interoperability plumbing does not redirect budget. Advertisers buy the walled gardens for reach and measurable outcomes that live inside those platforms, and no court remedy touches either. Amazon's retail media is still the fastest-growing piece of the three and keeps adding shopping surfaces only Amazon can sell ads on. The opposite outcome, share falling because of court-ordered openness, would require budget to chase publisher-ad-server integrations that most buyers never see, which is not how spend moves.

Right if: Madison and Wall's first full-year 2026 estimate of US ad revenue puts the three-platform share above its 2025 figure of about 56%. Wrong if: that estimate puts the share at or below the 2025 figure.

S2E15: The Open Exchange Myth Full Analysis → Listen to the episode →

By 30 September 2027, OpenAI will publish list pricing on its public pricing page for at least one developer product billed per completed task or per successful outcome.

Why Storing and reusing the intermediate math behind repeated prompts means a cached read costs roughly one-thousandth of fresh computation, yet buyers are billed only a modest discount for it, which is a large part of why frontier model providers are reported to run gross margins near 80 points on their interfaces for developers. Selling by the token puts that margin under permanent attack: prices per million tokens have fallen from about $60 five years ago to under $1, custom in-house chips are cutting the labs' own costs further, and every efficiency gain has to be handed back as a price cut to hold volume. Charging for a finished task instead lets the provider keep the savings, and OpenAI's president Greg Brockman has floated charging for useful results in place of raw tokens. The alternative, OpenAI staying purely on per-token and per-seat rates through 2027, means accepting that its price and its margin fall together every time inference gets cheaper. Agent products, where a single job burns an unpredictable number of tokens, give it the cleanest excuse to reprice.

Right if: On 30 September 2027, OpenAI's public pricing page lists at least one developer product with a published rate charged per completed task, per run, or per successful outcome. Wrong if: On 30 September 2027, every developer product on OpenAI's public pricing page is charged by tokens, by seat, or by subscription.

20VC: "Anti-Data Centres is a Chinese Psyop" | How Many Planned Data Centers Will Actually Get Built? | Is Energy AI's Biggest Bottleneck? With Thomas Sohmers, Co-Founder @ Positron Listen to the episode →

The remedies Judge Brinkema ordered against Google in the US ad-tech antitrust case will not take binding effect before the 2027 upfront selling season concludes.

Why Google has already announced it will appeal to the Fourth Circuit, and an appeal of a landmark antitrust remedy routinely delays enforcement by years, so no operator will see a functionally different AdX before mid-2027. The remedy also touches neither the reach nor the in-platform measurement advertisers actually buy, so nobody with budget is pushing the court for faster enforcement.

Right if: Google's AdX interoperability obligations remain unenforced or stayed through the 2027 upfronts. Wrong if: the remedy takes binding effect on Google before the 2027 upfront season ends.

S2E15: The Open Exchange Myth

At least two more independent location or purchase data companies, from a list that includes Placer.ai, Cuebiq, Blis, InMarket, and Cardlytics, will be acquired by the Q2 2027 earnings season.

Why The larger trend is that Google, Meta, and Amazon can only measure what happens inside their own walls, so anyone who wants to prove an ad drove a store visit or a purchase needs real-world signal the platforms do not own. That signal is now cheap to buy: location and purchase data companies lost much of their app supply after Apple's tracking changes, several have been through distress or fire sales, and the ones still independent are subscale. Infillion buying Foursquare shows the pattern, a buyer taking a data asset outright so it can lock the signal behind exclusivity and sell measurement the open web cannot otherwise offer, and the same logic applies to any holdco, DSP, or retail media network that wants a measurement story. The opposite outcome, the remaining independents staying independent, would require buyers to keep renting a signal a rival can lock up at any time.

Right if: at least two acquisitions of independent location or purchase data companies are announced by the end of the Q2 2027 earnings season, whether or not the target is on the named list. Wrong if: one or zero such acquisitions are announced by then.

The Future of the Open Web With Aly Nurmohamed and Christer Ljones Full Analysis → Listen to the episode →

No premium open-web publisher outside login-first players like Schibsted will launch a Meta-style outcome-based direct ad product at scale before the Q2 2027 earnings season.

Why Schibsted's path required a decade of login-first product investment and a deliberate vendor cull that most mid-tier publishers have not started. Without authenticated scale already in place, no publisher can close the outcome-budget gap against Meta before mid-2027, regardless of how cheap the model layer gets.

Right if: no non-Schibsted-tier open-web publisher publicly reports outcome-based direct budgets at material scale by the Q2 2027 earnings season. Wrong if: a mid-tier publisher outside the Nordic login-first model launches an audited outcome product that attracts meaningful direct advertiser spend.

The Future of the Open Web With Aly Nurmohamed and Christer Ljones

By the end of the Q1 2027 earnings season, Meta will offer advertisers a tool that decides how their budget is split across channels, including channels Meta does not own.

Why Meta has already taken away every advertiser control except creative so its own system runs the buy, and analysts credit that automation for much of the 2026 spend surge. The next decision up the chain is how much of the total budget belongs on Meta at all. A company that removed the dials to own the buying decision will not leave the budget decision to a startup, because whoever holds it steers the money. Doing nothing hands that position to a third party for free.

Right if: Meta launches or expands a tool that recommends or sets an advertiser's budget split across channels, including non-Meta channels. Wrong if: Meta ships nothing of the kind by the end of the Q1 2027 earnings season.

Episode 191: Corey Ferengul on the Newest Start-Up Category -- UMPs Full Analysis → Listen to the episode →

By the end of the Q1 2027 earnings season, Google will offer advertisers a tool that decides how their budget is split across channels, including channels Google does not own.

Why Google already decides the split inside its own properties through Performance Max, and analysts credit that automation for pulling in extra 2026 spend. Extending the same decision to the advertiser's whole budget is a small product step with a large prize: whoever sets the split steers the money. Startups are selling exactly that layer today, and Google has no reason to let a middleman sit between it and the budget.

Right if: Google launches or expands a tool that recommends or sets an advertiser's budget split across channels, including non-Google channels. Wrong if: Google ships nothing of the kind by the end of the Q1 2027 earnings season.

Episode 191: Corey Ferengul on the Newest Start-Up Category -- UMPs

By the end of the Q1 2027 earnings season, Amazon will offer advertisers a tool that decides how their budget is split across channels, including channels Amazon does not own.

Why Amazon is the one of the three that already buys inventory it does not own, through Amazon DSP, so a budget tool that spans other channels fits its existing product. Its retail data gives it the strongest claim to know which spend actually produced a sale. Startups are selling that budget-splitting layer today, and Amazon gains more from owning the decision than from letting a middleman hold it.

Right if: Amazon launches or expands a tool that recommends or sets an advertiser's budget split across channels, including non-Amazon channels. Wrong if: Amazon ships nothing of the kind by the end of the Q1 2027 earnings season.

Episode 191: Corey Ferengul on the Newest Start-Up Category -- UMPs

The Trade Desk's full-year 2026 revenue growth, reported on its Q4 2026 earnings call in February 2027, will come in below 15%.

Why The damage is on the demand side and it compounds: Publicis warning clients off, Omnicom moving to Amazon DSP, Dentsu and WPP leaving Open Path, and agencies building their own identity infrastructure so they stop paying The Trade Desk to resell it. Those are contract and workflow decisions that don't reverse inside a year, and they land on a leadership team that's still rebuilding after near-total C-suite turnover and a 15% staff cut. The bull case requires the CTV growth story to outrun holdco defection and a newly unleashed DV360, and nothing in the current picture says it will. The opposite outcome, a snap back above 20% growth, would require the same four holdcos that just walked to walk back, and no one has offered a reason to expect that.

Right if: The Trade Desk reports full-year 2026 revenue growth under 15% on its Q4 2026 call. Wrong if: it reports full-year 2026 revenue growth of 15% or higher.

The Trade Desk's Bumpy Ride Down Full Analysis → Listen to the episode →

Through mid-2027, AI licensing will still bring the New York Times less than 3% of its revenue, roughly triple the Amazon deal and still a rounding error beside advertising and subscriptions.

Why The Amazon deal is worth about $20 to 25 million a year against $2.6 billion in revenue. The AI labs have shown they will pay for a handful of marquee brands and scrape the rest, and every deal since has been smaller, not larger. The people arguing licensing is the industry's ace card are the vendors who sell the licensing plumbing, and they cite counts of sources and countries, never revenue. The opposite outcome would require a lab to pay the Times something like $80 million a year, more than triple any published deal.

Right if: the Times' disclosed or reported AI licensing income stays under 3% of revenue through the Q2 2027 report. Wrong if: a new deal or the reported total puts it at 3% or more.

Dow Jones Factiva's Emma O'Brian on why publishers hold 'the ace card' in the AI era Full Analysis → Listen to the episode →

Before the 2027 upfront season ends, at least one top-20 US consumer-goods advertiser will publicly cut or freeze its retail media spend at Amazon or Walmart and say measurement is the reason.

Why Brands have merged in-store, shopper, and retail media under single leaders so one person can compare networks on a common number, and that person cannot compare networks that each grade their own homework. Amazon and Walmart own the closed-loop sales data, so their reporting always looks good inside their own walls. A buyer who now holds the whole budget has leverage the separate teams never had, and the way leverage gets used in this industry is a public spending pause that forces the seller to the table. The opposite outcome, brands quietly accepting platform-supplied returns, would require the buy side to give up the leverage it just reorganized to get, and nobody reorganizes twice in eighteen months to do that.

Right if: a top-20 US consumer-goods advertiser, or its CFO or CMO on the record, announces a cut or freeze in Amazon or Walmart retail media spend and names measurement or incrementality as the reason. Wrong if: no such public cut or freeze happens by the end of the 2027 upfront season, or the only cuts cited are budget or macro reasons.

The Operating System for Commerce Media: Jeffrey Cohen on Amazon Ads’ Rise, Retail Media’s Measurement Reckoning, and the Shift from Dashboards to Agents Listen to the episode →

Ozone's Arc, the UK publisher group's new product that sells pooled reader data for performance campaigns, will not have a single named advertiser saying it moved performance budget to Arc by June 2027.

Why Performance money left the open web because publishers cannot prove an ad led to a sale. Pooling reader data across publishers makes the audience bigger, but it does not fix that proof problem, because the sale still happens somewhere the publishers cannot see. Meta, TikTok, and Amazon can see it. Arc will win test budgets and trade-press coverage, and the copycats in other markets will follow, but a brand only moves real performance money when the numbers beat what it gets from the platforms.

Right if: no advertiser is publicly named, by Ozone or by itself, as having shifted performance or conversion budget to Arc by June 30, 2027. Wrong if: a named advertiser says on the record that it moved performance budget to Arc.

MadTech Daily: Programmatic DOOH Reaches $1.34bn Globally; Shein Stock Hits New Low Since IPO Full Analysis → Listen to the episode →

Layers, the startup pitching itself as the first mobile measurement agent, will be acquired, shut down, or drop that pitch by June 2027.

Why Layers' whole product is letting AI agents pull mobile attribution data, and the data belongs to AppsFlyer and Branch. Giving agents direct access to that data through an MCP connector is a quarter of engineering work for the incumbents, and every software vendor is shipping one right now. Once the data owners offer the connector themselves, a startup whose pitch is access to someone else's data has nothing left to sell, and the usual exits are a small acquisition for the team or a pivot to a different problem. The opposite outcome, Layers still selling agent access to AppsFlyer and Branch data as an independent company a year from now, would require both incumbents to leave that door open, and vendors defending a data business do not do that.

Right if: Layers announces an acquisition, closes, or publicly repositions away from the mobile measurement agent pitch by June 30, 2027. Wrong if: Layers is still independent and still selling agent access to mobile attribution data on that date.

Building Apps Is Solved. Distribution Isn't. Full Analysis → Listen to the episode →

Another location- or purchase-data asset outside Google and Meta will be acquired by a holdco, DSP, or larger ad-tech platform by the 2027 upfront (May 2027), as buyers move to own real-world signal rather than license it.

Why Infillion just did this deal twice in seven months and said it will keep buying, which tells you the buy-side reads location and purchase data as scarce currency now that Google's stack is under antitrust pressure and the cookie keeps dying. The mechanism is straightforward: walled gardens can only measure inside their own apps, so any advertiser or agency that wants to prove ads drive store visits needs data they can't get from Google or Meta, and owning it beats renting it once a rival locks a source behind exclusivity. Omnicom, WPP, Publicis, and The Trade Desk all have both the incentive and the checkbook. The opposite outcome, everyone keeps licensing and nobody buys, is less likely precisely because Infillion just showed that exclusivity turns a shared data source into a competitive weapon, and no large buyer wants to be the one still renting when the source goes dark.

Right if: a holdco, DSP, or ad-tech platform announces the acquisition of a location or offline-purchase data company (Placer.ai-class, a retail-data firm, or similar) before the 2027 upfront. Wrong if: no such deal is announced and the category's data continues to trade primarily through licensing.

Infillion Acquires Foursquare to Bolster Location-Data Stack Read the source story →

By Take-Two's fiscal-year-end earnings in May 2027, Zynga will have shut down Ads+ as a third-party ad network or folded it back into selling only its own games' inventory.

Why The launch leans on a DoubleVerify stamp and a 4.5-star app-rating door check because the audience case is weak. Zynga's players are casual mobile gamers, an audience buyers already reach cheaply through DV360 and every DSP, so there is no reason to route budget through a new middleman unless the price or the data is better, and nothing in the launch shows either. AppLovin's third-party network worked because of a machine-learning bidding engine and massive scale, and Zynga has neither at that level. A network that cannot win budget on price or data becomes a cost line, and Take-Two has shown it will cut those. For this to be wrong, Zynga would have to prove its game-content signal drives outcomes generic in-app supply cannot, and land a brand advertiser willing to say so publicly.

Right if: Zynga or Take-Two announces, or the trade press reports, that Ads+ has stopped selling third-party inventory or has been shut down by May 31, 2027. Wrong if: Ads+ is still operating as a third-party network on that date.

Zynga Launches Ads+ Third-Party Publisher Network with DoubleVerify Read the source story →

Through the 2027 upfront negotiations, no top-five US media agency will accept a CTV publisher's own data as the sole measurement for buys on that publisher's inventory.

Why The whole value of an agency to a brand is independent judgment about where the money goes, and that judgment evaporates the moment the seller supplies the grade. Every large advertiser's procurement team knows this, which is why cross-publisher measurement and clean rooms exist in the first place. Horizon's Roku data deal is real, but it has no disclosed audit and no confirmation the data blends with other publishers, and those are exactly the pieces a CMO's finance team demands before a seller's numbers become the official scorecard. The opposite outcome, agencies accepting Roku's self-reported measurement as the currency, would require the buy side to surrender the one thing that justifies its fee, and holding companies do not give that away for a data pipe.

Right if: through the 2027 upfronts, no top-five agency publicly adopts a single CTV publisher's own data as the sole measurement for that publisher's inventory. Wrong if: a top-five agency announces it is grading a major CTV publisher's buys solely on that publisher's own feed, with no independent measurer involved.

Roku Supplies Data That Plans, Targets, and Measures Its Own Inventory Full Analysis → Read the source story →

By June 2027, at least one of the six largest agency holding companies will announce a spending commitment with OpenAI for ads inside ChatGPT.

Why Every holding company says it will not move real budget into inventory that DoubleVerify, IAS, or Comscore cannot audit, and no accredited measurement exists for ads inside a chat agent today. The same companies said the same about TikTok and Amazon and signed spending deals anyway once the reach was there, because a holding company's real product is access to the next big audience, and being first to a deal with OpenAI is worth more to its new-business pitch than waiting for an audit. OpenAI needs the revenue and a marquee agency name, so it will offer the pricing and the co-marketing that make a commitment easy to announce. The opposite outcome, every holding company holding out until third-party measurement is accredited, would require the industry to behave differently from every prior platform launch.

Right if: one of WPP, Omnicom, Publicis, IPG, Dentsu, or Havas (or their successor entities) publicly announces a spending commitment, partnership, or upfront-style deal for advertising inside ChatGPT by June 30, 2027. Wrong if: no such announcement from any of the six by that date.

OpenAI Announces 'Sponsored Agents' Ad Product for Conversational AI Read the source story →

PubMatic's full-year 2026 revenue, reported in February 2027, will be lower than its 2025 revenue.

Why The forecasters raising their 2026 ad-spend estimates say the new money is going to Google, Meta, and Amazon on the strength of their automated buying tools, which means the independents face a market where headline growth never reaches them. PubMatic sells open-web inventory to buyers who are steadily automating that budget into the walled gardens and into Amazon's DSP, and it has already lost volume when a large DSP partner changed how it bids. When the addressable budget stops growing, an independent's revenue falls before its strategy changes, and the strategy change follows: expect PubMatic to spend 2027 repositioning as plumbing that supplies the walled gardens rather than as an alternative to them. The opposite outcome, PubMatic growing in 2026, would require open-web spend to outgrow the platforms for the first time in a decade.

Right if: PubMatic reports full-year 2026 revenue below its reported full-year 2025 revenue. Wrong if: full-year 2026 revenue is flat or up on 2025.

AI Driving 2026 Ad Spend Above Forecast; Growth Concentrated in Google, Meta, Amazon Read the source story →

By the end of September 2027, no independent, audited standard for measuring purchases made through AI agents will be in market.

Why Every new ad channel arrives with the platform keeping its own scoreboard first, and neutral audited measurement shows up only after buyers have enough spend at stake to demand it. That took years for CTV and retail media and is still incomplete there. The measurement vendors who would build a neutral standard are the ones the platforms least want holding the ruler, so the incentive runs against one emerging fast. The opposite outcome, a neutral audited currency for agent purchases inside a year, would require the platforms to invite outside grading of a channel they currently own end to end, which none of them did with the walled inventory they already have.

Right if: no MRC-accredited or comparably independent, audited cross-platform standard for measuring AI-agent-driven purchases is live by September 30, 2027. Wrong if: such a standard is live and adopted by at least one major holding company by then.

Information Quality Is What Wins In The AI Era Full Analysis → Listen to the episode →

Through September 2027, none of Microsoft, Amazon, or Google will give an outside measurement company access to verify purchases made through its AI shopping agent.

Why Each platform is shipping its own reporting for purchases its agent drives, which lets it grade its own homework on a channel nobody else can see into. Letting DoubleVerify, IAS, or Comscore verify those purchases would surrender the one advantage of owning the agent, and the platforms have never opened comparable data on their existing walled inventory until forced. Buyers will complain, but they will keep spending, because the reach is there and there is no other place to buy it. The opposite outcome, a platform volunteering outside verification of agent purchases, would require pressure from advertisers that does not exist while the channel is this new.

Right if: none of Microsoft, Amazon, or Google announces an independent measurement partner with access to verify AI-agent-driven purchases on its platform by September 30, 2027. Wrong if: any of the three announces such access for an outside measurement company.

Information Quality Is What Wins In The AI Era

Madison and Wall's first estimate of full-year 2026 US ad revenue, due by March 2027, will put Amazon's share at 10% or higher, up from about 9% for 2025.

Why Amazon's retail media sells ads against its own shopping data, and it is the fastest-growing piece of the three big platforms. Amazon is still pouring capital into the shopping surfaces that feed that ad business, including its quick-commerce push in India, and every new shopping surface becomes new ad inventory only Amazon can sell. Google and Meta grow with the market; Amazon is taking share from it. The opposite outcome, Amazon stalling at 9%, would require retail media spend to stop growing faster than total ad spend, which has not happened in any year since Amazon began reporting ad revenue.

Right if: Madison and Wall's first full-year 2026 US ad revenue estimate shows Amazon at 10% or more of the total. Wrong if: that estimate shows Amazon below 10%.

MadTech Daily: AI Strengthens Big Tech’s Hold on Ad Growth; China Pulls Ahead of US in Consumer AI Adoption Full Analysis → Listen to the episode →

By 31 May 2027, no independent media agency will have publicly said it is running live advertiser budget through Quantum Path, the JWX product that moves campaign buying decisions into publisher-side software.

Why Ken Rona, Chief AI Officer at JWX, argues the decision-making layer of ad buying will drain out of the platforms agencies use and pool inside publisher-side software, and Quantum Path is the product built on that argument; it went live with one client, with a feature-complete version promised for January 2027. The obstacle is who pays whom: publisher-side software is paid by the seller, so an advertiser handing it budget allocation is asking the party that profits from higher prices to decide what to pay. Agencies have spent a decade keeping that decision on their own side of the table, and a smarter seller-side algorithm sharpens the reason to keep it there. The likelier path is that routing and campaign setup drift to the seller's software while budget allocation stays with the buyer, which would leave Quantum Path with publisher-owned demand and pilots rather than an agency willing to put its name on it.

Right if: As of 31 May 2027, no independent media agency has publicly stated that it runs live advertiser budget through Quantum Path. Wrong if: On or before 31 May 2027, at least one independent media agency publicly states that it runs live advertiser budget through Quantum Path.

S2E14: AI Is the Easy Part with JWX's Ken Rona Full Analysis → Listen to the episode →

Walmart will open a tool to all of its suppliers that shows whether a streaming TV ad led to a purchase in its stores or on Walmart.com, by 15 May 2027.

Why Walmart now sells streaming ads from three sources at once: the Vizio TV sets it owns, a supply deal with Roku, and Vibe, the reseller of streaming ad space it just bought. Brands that already pay Walmart for shelf space, in-store promotion and search ads have no reason to add a television line item unless it comes with proof the ad moved product, which is the one thing no rival retailer or independent buying platform can offer, and cheap reach is already available elsewhere. Consumer-goods budgets for 2027 are set in the back half of 2026 and locked around the upfront, so Walmart either has a measurable video product in market by then or waits a full cycle while Amazon extends its lead. That commercial clock is a stronger force than the engineering difficulty, and Walmart owns both the purchase records and the viewing data, so the shortcut of shipping a rough version to all suppliers is available in a way it is not to any competitor. The case against is that Walmart is assembling its audience data and its measurement at the same time, unlike Amazon, which had Prime viewers in hand before it built the ad product.

Right if: By 15 May 2027, Walmart Connect has announced or listed as generally available a purchase-measurement tool for streaming ads that any supplier can buy against without a Walmart sales team running the campaign. Wrong if: As of 15 May 2027, no such tool is generally available, with purchase measurement for Walmart's streaming ads still confined to pilots, betas, or managed-service arrangements.

Walmart buys Vibe CTV inventory; accelerates retail media video strategy Full Analysis → Read the source story →

Before the end of 2027, at least one of the 20 largest US consumer-goods advertisers will publicly cut or freeze its spending with a retailer's own ad network other than Amazon's or Walmart's, and say the quality of the measurement it was given is the reason.

Why Retailers currently grade their own homework: proving an ad caused a sale requires holding a group of shoppers back from seeing it, and the retailer decides who sits in that group. Early Amazon search campaigns posted returns of 40 to 50 percent on ad spend because last-click credit had no competition for the sale, and that same flattery now lives one layer down, inside retailer-run tests. A large packaged-goods buyer running across eight retailers cannot reconcile eight sets of results that each claim the same purchase, and the biggest buyers already have in-house analysts running rival tests. The quiet alternative, where everyone renews on familiar numbers and nobody says anything in public, is exactly what the past decade looked like, but retail media budgets are now large enough that one visible walk-away buys an advertiser pricing leverage in every other retailer negotiation it has.

Right if: Right if, by 31 December 2027, AdExchanger reports a named advertiser from the 20 largest US consumer-goods ad spenders stating publicly that it has cut, paused, or frozen spending with a retail media network other than Amazon Ads or Walmart Connect because of how that network measures results. Wrong if: Wrong if no such named advertiser statement has been reported by AdExchanger on or before 31 December 2027.

Retail media measurement shifting from ROAS to incrementality Full Analysis → Read the source story →

By Omnicom's second-quarter 2027 earnings report in July 2027, Omnicom will have publicly named at least one client outside retail media and online commerce whose fee is mostly tied to agreed campaign results.

Why Flywheel Digital, Omnicom's e-commerce arm, already prices a reported 80 to 90% of its work on results, because retail platforms hand the agency the sale and make the scoreboard easy to agree on. Omnicom is using that proof point to tell clients and investors it can grow revenue without adding staff, and a commerce-only example does not carry that argument into the far larger part of the group that sells brand advertising. The standard objection is that brand outcomes cannot be measured cleanly, but the softness works in Omnicom's favour: when the agreed result is a brand-lift score or a modelled sales lift, the agency drafts the definition, which makes such a contract easier to sign than a hard cost-per-sale deal. One cooperative client and a friendly metric is all a showcase contract requires, and Omnicom has strong commercial reason to put one on stage while the story about artificial intelligence replacing agency labour is still live.

Right if: Right if, on or before its second-quarter 2027 earnings report in July 2027, Omnicom has publicly named a client outside retail media and online commerce whose fee is majority results-linked. Wrong if: Wrong if no such named client has been made public by Omnicom on or before that report.

MadTech Daily: Omnicom Eyes New Pricing Model; Anthropic Forecasts Second Straight Quarter of Profit Listen to the episode →

OpenAI will have a paid advertising product available to US advertisers that places brand messages inside ChatGPT's answers by March 31, 2027.

Why Publishers such as Time have started building pages aimed at the crawlers that feed chatbots and selling brand mentions on them, with undisclosed mentions being solicited for as little as $200, and Perplexity has already blocked those pages from influencing its answers. That block shows the placement has value and that the company controlling the answer will not let outsiders sell it cheaply underneath them. The precedent is Google, which turned the search results page into the most profitable ad surface ever built rather than leaving the ranking to be gamed by whoever paid a publisher. OpenAI has hundreds of millions of weekly users, costs that subscriptions do not cover, and the same reason to capture this demand directly. Leaving paid mentions to leak in through publishers is the less likely path because it corrupts the trustworthy answer that is the only product OpenAI sells, while earning OpenAI nothing.

Right if: OpenAI publicly announces and makes available to US advertisers, on or before March 31, 2027, a paid product that places brand messages inside ChatGPT responses, per OpenAI's own announcement or advertiser-facing product documentation. Wrong if: No such product is publicly available to US advertisers as of March 31, 2027.

As more surfaces become ad inventory, the economics get complicated Full Analysis → Listen to the episode →

By 30 June 2027, advertisers will not be able to buy ad placements inside ChatGPT through the Amazon DSP as a product open to any Amazon DSP advertiser.

Why The claim that Amazon's demand-side buying platform now reaches ChatGPT placements has circulated through vendor briefings with a commercial interest in more Amazon DSP spend, without a dated confirmation from either Amazon or OpenAI. OpenAI is building its own advertising business and its own shopping flow, and routing that inventory through Amazon would hand a retail competitor the pricing and the conversion data on the exact moment a shopper decides what to buy. On the buy side the demand is soft: Amazon Marketing Cloud, the most capable clean room in retail media, sits underused on Amazon's own mature surfaces, so advertisers already cannot tell whether that spend is incremental, and a chat surface with no shared identifiers and no agreed conversion definition is a harder sell, not an easier one. For the opposite to happen, two companies racing each other for the same commerce transaction would have to ship a joint, generally available buying path inside nine months.

Right if: On 30 June 2027, Amazon Ads' public product documentation does not list ChatGPT or OpenAI inventory as a placement any Amazon DSP advertiser can buy. Wrong if: On 30 June 2027, Amazon Ads' public product documentation lists ChatGPT or OpenAI inventory as a placement any Amazon DSP advertiser can buy.

Beyond Retail Media: Connected Commerce with Pacvue’s Tommy Burton Full Analysis → Listen to the episode →

At least one of the six largest advertising agency holding companies will publicly require its media vendors to complete the IAB Diligence Platform privacy attestation in order to stay on its buys, by 31 December 2027.

Why A compliance form turns into a gatekeeper when the buyer carries the legal exposure, and under US state privacy law an advertiser can be held responsible for what vendors several tiers down its supply chain do with consumer data. The IAB built its Diligence Platform in 2023 with roughly 80 member companies and 10 law firms behind it, running on SafeGuard Privacy's engine, so a buyer can point to one shared questionnaire instead of hundreds of bespoke ones — the cheapest paperwork a holding company can produce to show it checked. California enforcement so far tops out around a $130,000 fine against a single data broker, small enough that spreadsheets still technically work, but the penalties are statutory and stack per violation, and a single plaintiff-side lawsuit makes a documented industry standard worth far more than its license fee. The alternative, every holding company keeping vendor diligence private and bespoke, gives up the one defense that is cheap to generate and forces vendors to keep answering forms they have already answered once.

Right if: Right if, on or before 31 December 2027, one of Omnicom, Publicis, WPP, Dentsu, Havas or Accenture Song has stated publicly, or published a supplier policy showing, that IAB Diligence Platform attestation is required of media vendors it buys from. Wrong if: Wrong if no such public statement or published supplier policy exists from any of those six companies by 31 December 2027.

The Permission Layer: Who Told the Al It Could Do That? Richy Glassberg on Data Privacy, Consent, and Al Innovation Full Analysis → Listen to the episode →

Meta will make a face-identification feature that names people through its Ray-Ban smart glasses available to consumers in at least one US state before 30 September 2027.

Why Meta quietly pushed the code for a face-identification feature called Nametag to Ray-Ban glasses and pulled it only after New York Times and Wired coverage, which means the capability was finished and shippable and the block was public attention rather than engineering. Internal planning reportedly discussed timing the rollout for a moment when privacy advocates were busy elsewhere, and the legal constraint is concentrated in a handful of states: Illinois's biometric law lets individuals sue over face data collected without consent, and Meta already paid $650 million to settle one such claim over photo tagging. That combination points to a carved-up release with consent screens and state-level blocking, not permanent shelving, because face identification is the main thing that would separate Meta's glasses from cheaper camera eyewear. Permanent abandonment would require Meta to write off a completed feature in its most strategically important new hardware line, which it has rarely done when the liability can be geofenced away.

Right if: Meta has made a feature that identifies people by name through Ray-Ban glasses available to ordinary buyers in at least one US state, per Meta's own product announcements, on or before 30 September 2027. Wrong if: No such feature is available to ordinary buyers in any US state as of 30 September 2027, per Meta's own product announcements.

Flock Cameras, Meta Glasses and the Surveillance Backlash Listen to the episode →

By June 30, 2027, advertisers will be able to buy ads inside ChatGPT directly from OpenAI, through an OpenAI sales team or an OpenAI self-serve tool, without going through Amazon or any other outside ad partner.

Why Amazon is currently placing advertiser campaigns inside ChatGPT through a US pilot in which OpenAI supplies the users and Amazon supplies the advertisers, the pricing power and the shopper data. That arrangement gives OpenAI revenue without building an ad stack, but it also hands the advertiser relationship and most of the margin on the highest-intent new ad surface in a decade to a company that competes for the same shopping budgets. OpenAI's applications chief, Fidji Simo, built Instacart's advertising business before joining, and OpenAI's compute bills give it a strong reason to own that margin rather than rent it out. The opposite outcome, OpenAI staying a pure supply partner past mid-2027, would mean it permanently accepting the thinner slice of its own inventory, which platforms with their own logged-in users have historically refused to do once the demand is proven.

Right if: Right if, by June 30, 2027, OpenAI has publicly launched a way for advertisers to buy placements in ChatGPT directly from OpenAI, confirmed by OpenAI's own published advertising product announcements. Wrong if: Wrong if, on June 30, 2027, the only way to place ads in ChatGPT is through Amazon or another third-party partner, or ChatGPT carries no advertiser placements at all.

MadTech Daily: Amazon Ads Expands Into ChatGPT; Sony Proactively Shared WPP Rebate Probe Full Analysis → Listen to the episode →

The price Microsoft publishes for one million input tokens (the billing unit for text sent to an AI model) on the most capable OpenAI model generally available through Azure will be the same or higher on 30 June 2027 than it is today.

Why David Morehead, chief investment officer of Baylor University's $2.6 billion endowment and a direct investor in data center infrastructure, says the stock of sites that already hold power and building approvals rose roughly 50% in six months and that local permitting, not electricity or chips, is now the binding limit on new capacity. Cloud providers have spent the past three years cutting per-token prices roughly annually as chip supply improved, so buyers and product roadmaps assume the cuts continue. If the constraint has moved from silicon to zoning boards and grid interconnection queues, supply stops catching up with demand and the providers have no commercial reason to discount capacity they cannot expand. A price cut over this window would mean new buildings coming online faster than planned, and municipal approval fights do not resolve in three quarters. That makes flat-or-higher pricing the harder but more likely outcome, and it is the number every ad-tech firm budgeting AI features into 2027 is quietly betting against.

Right if: Microsoft's published Azure OpenAI pricing page on 30 June 2027 lists a per-million-input-token price for its most capable generally available OpenAI model that is equal to or higher than the price listed for the equivalent top model on 20 September 2026. Wrong if: That published price on 30 June 2027 is lower than the price listed for the equivalent top model on 20 September 2026.

20VC: How LPs Allocate to Venture in 2026: What They Want, What They Do Not Want | Why Fund Multiple Does Not Matter Without a Timeline | Why Velocity of Cashback is the Most Important Thing with David Morehead, CIO @ Baylor Listen to the episode →

By June 30, 2027, at least one of the five largest agency holding companies (WPP, Omnicom, Publicis, Dentsu, Havas) will publicly announce a committed spending agreement covering in-game advertising inventory at Electronic Arts.

Why Electronic Arts has wired its player accounts into LiveRamp, the company that matches a brand's own customer list to a publisher's audience, and is standardizing its ad formats with the IAB while having Integral Ad Science measure whether ads are actually seen. The point of all three moves is to remove the six-to-twelve-month engineering build that today stands between an advertiser and a single in-game impression, which is the reason agency planners lose gaming budget to Meta and Amazon. EA brings roughly 500 million player accounts, and agency groups have a habit of signing volume commitments early in a new channel to lock in pricing and first access before measurement is settled, exactly as they did with connected TV years before viewability norms were agreed. The case against is that finance teams will not approve gaming money without a defensible viewability number, but a publicly announced commitment costs a holding company little and buys it positioning, which is why the announcement usually arrives ahead of the audited measurement rather than after it.

Right if: WPP, Omnicom, Publicis, Dentsu or Havas publicly announces a committed spending agreement covering Electronic Arts in-game ad inventory, in a statement from the holding company or Electronic Arts, on or before June 30, 2027. Wrong if: No such public announcement from any of those five holding companies or Electronic Arts exists on or before June 30, 2027.

EA Partners with LiveRamp to Enable First-Party Data Matching for In-Game Ads Full Analysis → Read the source story →

No advertiser in the 20 largest US advertisers by ad spending will publicly cut or freeze its spending with Amazon's advertising business and give Amazon's auction pricing as the reason, as reported by Ad Age on or before September 30, 2027.

Why The Federal Trade Commission alleges Amazon shifted to first-price auction economics, where the winner pays exactly what it bid, while continuing to describe its auction as second-price, where the winner pays a penny over the runner-up; the big open-web exchanges announced that same switch in 2019 so buyers could recalibrate their bidding software. The test of whether auction transparency is a real purchasing lever is not the lawsuit but whether a single large buyer moves money over it. Amazon's shopper-purchase data has no substitute at scale, so a buyer walking away pays an immediate performance cost to make a point about six-year-old clearing prices. Agencies and advertisers also have their own problem: if the pricing shift was visible in bid logs and nobody caught it for a year, saying so publicly invites clients to ask what else went unwatched. The opposite outcome requires a buyer willing to absorb both worse performance and that question in public.

Right if: By September 30, 2027, no advertiser on Ad Age's list of the 20 largest US advertisers has publicly announced a cut or freeze in Amazon ad spending citing Amazon's auction pricing or disclosure. Wrong if: By September 30, 2027, at least one advertiser on that list has publicly announced a cut or freeze in Amazon ad spending citing Amazon's auction pricing or disclosure.

Open Web Exchanges Disclosed First-Price Switch; Amazon Allegedly Did Not Full Analysis → Read the source story →

Through December 31, 2027, none of the five biggest demand-side platforms (DSPs) — Google's Display & Video 360, Amazon DSP, The Trade Desk, Yahoo DSP and Criteo — will offer all of its customers a feature that lets AI software shift an advertiser's budget onto an ad platform it does not own without a person approving each move.

Why Google, Meta, Amazon and most large DSPs and SSPs have already shipped servers for Model Context Protocol, the Anthropic-created standard now housed at the Linux Foundation that lets AI software operate an ad platform in plain language, so the technical path to software that spends across platforms is built. What is not built is the approval layer: buyers have no written rule for who signs off when software moves budget at 2am, and a single misfire spreading spend across four platforms overnight is a real dollar loss with no clear owner, which turns this into a legal and headcount project rather than a feature toggle. The platforms also have no commercial reason to lead here, because the feature's whole purpose is to route money to competitors. Reporting, audience-building and draft-campaign agents ship instead, since they cut labor with no spending risk. For the opposite to happen, one of these companies would accept fresh liability for money moved to rivals before its own customers have asked for the capability.

Right if: On December 31, 2027, none of those five companies' published product documentation describes a generally available feature that lets AI software move budget to platforms the vendor does not own without per-change human approval. Wrong if: On December 31, 2027, at least one of those five companies' published product documentation describes such a feature as generally available to its customers.

MCP Servers Emerge as Unifying Interface Layer Across Fragmented Ad Stack Full Analysis → Read the source story →

The Trade Desk will report full-year 2026 revenue below its full-year 2025 revenue when it announces fourth-quarter results in February 2027.

Why Google, Meta, and Amazon took 56 cents of every US ad dollar in 2025, up from 53 the year before, a three-point move against a prior pace of one to one-and-a-half points a year, with Amazon alone at 9%. The Trade Desk sells access to the budget those three cannot capture, so the pool it fishes in is draining at roughly triple its old rate while Amazon's buying platform resells much of the same connected-TV inventory at a lower take. Advertisers who let Google's and Meta's automated buying tools set their budgets also build up data inside those platforms that makes moving money back out more expensive each year, which compounds the drain rather than reversing it. The bull case is that connected TV and retail media grow fast enough to cover the loss, but that is the exact inventory Amazon is pricing against, and The Trade Desk has already given up share there. A flat or growing year requires the independent side to hold ground it has been losing for two straight years.

Right if: The Trade Desk's reported full-year 2026 revenue, in its February 2027 fourth-quarter release, is below its reported full-year 2025 revenue. Wrong if: That same release shows full-year 2026 revenue equal to or above full-year 2025 revenue.

Google, Meta, Amazon Now Control 56% of US Ad Revenue Read the source story →

Quantum Path, the sell-side software from JWX pitched as replacing the human traders who set up ad campaigns, will be shut down, absorbed into an acquirer, or no longer sold as a replacement for human traders by June 30, 2027.

Why Quantum Path launched with one client onboarded and a feature-complete date of January 2027, and JWX's own framing keeps the agents out of the live auction where ad money actually changes hands, which means the product automates campaign setup rather than the work buyers pay traders for. Setup is the easy last mile; the expensive part is custom deal terms, brand-safety overrides, pacing failures on thin inventory and knowing when to escalate, so buyers will keep humans watching the agent and pay for both. That double cost pushes the payback decision out to the first renewal, and a single pilot client renewing is the only thing standing between this product and the scrapheap. Meanwhile the large sell-side platforms, Magnite and PubMatic among them, already sell buyers direct-access tools and can bolt setup automation onto them, which compresses the price a standalone vendor can charge. For the product to still be on sale in mid-2027 under the same pitch, the pilot has to renew, buyers have to accept the agent's output without redoing it, and a small private vendor has to hold a position two much larger platforms can copy.

Right if: Right if, on June 30, 2027, JWX's public product pages and announcements show Quantum Path discontinued, JWX acquired, or the product described as assisting traders rather than replacing them. Wrong if: Wrong if, on June 30, 2027, JWX is still independent and still publicly selling Quantum Path as a replacement for human ad traders.

JWX Launches Quantum Path: AI Agents to Replace Human Ad Traders Read the source story →

By May 31, 2027, at least one of The Trade Desk, Amazon DSP, Yahoo DSP, Criteo, or Viant will publicly state that Chrome's ad-density measurements feed into how it bids on or scores web inventory.

Why Google has added four columns to its free public Chrome user-experience dataset measuring how many ads a page carries, how much of the screen they occupy, how many bytes they load, and how much device processing they consume, and it has said its own buying tools get no privileged access to them. That hands every independent buying platform a zero-cost ruler for the cluttered, made-for-advertising long tail they already claim to filter out, and the first one to price on it gets a clean sales story: pay up for light pages, mark down heavy ones. Publishers are waiting for buyers to move and buyers are waiting for the metric to mean something, so the standing bet is that nobody acts, but the cost of acting is close to zero and the competitive pressure to show a quality differentiator against Google's own stack is constant. For the opposite to hold, every large independent buyer would have to leave a free, public, on-brand supply-quality signal untouched for a full year while selling clean supply as a product.

Right if: On or before May 31, 2027, one of those five platforms has said publicly, in a product announcement, documentation, or on-record executive statement, that Chrome CrUX ad-density data informs its bidding or inventory-quality scoring. Wrong if: As of May 31, 2027, none of those five platforms has made such a public statement.

Chrome Adds Four New Ad-Density Metrics to CrUX Dataset Full Analysis → Read the source story →

Warner Bros. Discovery will announce, on or before 30 June 2027, that it is retiring, selling, or handing to an outside partner the in-house technology it built to sell ads against its own streaming inventory.

Why Warner Bros. Discovery's linear television advertising revenue fell 27% year over year in its most recent quarter, while its streaming advertising grew only in the high single digits, so the inventory base funding its own ad-sales engineering is shrinking far faster than the replacement is growing. Running a proprietary system that sells your own ad space is a fixed cost that made sense when cable threw off cash; against a smaller base, renting demand from the buying platforms advertisers already use costs less even after giving up margin. Fixed measurement contracts with Nielsen and iSpot were priced in growth years and do not shrink with inventory, which turns a revenue decline into a margin problem and puts the ad-tech build first in line for cuts. The alternative, keeping the stack and trimming around it, requires streaming growth to close the gap with linear decline, and this year's numbers show the gap widening instead. The easy version of this call, simply adding another programmatic pipe while keeping the stack, is already standard practice; the real test is whether the company gives up owning the selling technology at all.

Right if: Warner Bros. Discovery states in a press release, regulatory filing, or quarterly earnings materials published on or before 30 June 2027 that it is shutting down, divesting, or outsourcing operation of its own streaming ad-sales technology. Wrong if: No such statement appears from Warner Bros. Discovery by 30 June 2027, including if it only adds or expands third-party buying-platform access while continuing to run its own ad-sales technology.

Linear TV Ad Revenue Craters at Paramount and WBD; Fox Surges 108% Full Analysis → Read the source story →

By August 15, 2027, Amazon's demand-side platform will publicly require a transaction ID — the unique tag that shows a buyer when the same ad impression is being offered to it over and over by different sellers — on the open-web inventory it bids on, per Amazon Ads' published bidding documentation or a public Amazon Ads announcement.

Why The Trade Desk pushed sellers to carry transaction IDs, and when the tags were broken inside Prebid, the open-source software publishers use to run their ad auctions, the IAB Tech Lab flagged the breach and then said The Trade Desk was within its rights to refuse inventory arriving without the ID. That ruling turns buyer-side enforcement from a standards violation into accepted market behavior, which lowers the cost for the next large buyer to copy it. Amazon sells its demand-side platform against The Trade Desk partly on transparency, and advertisers paying twice to reach the same person is the one complaint no buyer defends, so matching the requirement costs Amazon almost nothing technically because the ID already sits in the spec. The alternative, Amazon leaning only on its own internal deduplication and never stating a public requirement, leaves it buying inventory its biggest rival has already declared unverifiable. One buyer enforcing this is a preference; a second makes it the plumbing everyone runs.

Right if: On or before August 15, 2027, Amazon Ads has announced or documented that its demand-side platform will not bid on third-party inventory lacking a transaction ID. Wrong if: As of August 15, 2027, Amazon Ads has published no such requirement for its demand-side platform.

IAB Tech Lab Flags Trade Desk Transaction ID Move, Then Endorses It Read the source story →

Paramount Skydance's direct-to-consumer advertising revenue for full-year 2026, reported at its fourth-quarter earnings in February 2027, will grow by less than 18% over 2025.

Why Ad-supported streaming services raised commercial time per hour by about 18% between January and August 2026, according to Ampere Analysis, with Paramount+ now running roughly nine ad minutes an hour, close to broadcast levels. Advertiser demand for streaming inventory is not expanding at that rate in eight months, so the added slots either go unsold or clear at lower prices, and each extra minute earns less than the one before it. Netflix is the control case: it holds under 2.5 minutes an hour and prices the scarcity, while the services chasing load parity trade price per impression for volume. For revenue growth to beat the load increase, Paramount would have to fill every new slot and hold rates at the same time, which requires demand appearing faster than any current streaming forecast assumes.

Right if: Paramount Skydance reports full-year 2026 direct-to-consumer advertising revenue that is less than 18% above its 2025 figure. Wrong if: The reported full-year 2026 direct-to-consumer advertising revenue is 18% or more above 2025, or no comparable full-year advertising figure is published by 1 March 2027.

Ad-Supported Streaming Services Raised Commercial Load 18% in 2026 Full Analysis → Read the source story →

By May 31, 2027, one of the six largest agency holding companies (Publicis, Omnicom, WPP, Dentsu, Havas, Stagwell) will publicly announce that it uses Amazon's Bedrock Agent Core, the Amazon Web Services product Amazon pitches as the single control screen directing other advertising AI tools, to plan or buy client media on platforms Amazon does not own.

Why Amazon is putting its own engineers inside agencies to install Bedrock Agent Core, and it is moving before Google offers an equivalent. Ripping out an orchestration layer once it is wired into agency workflow is not a quick decision, so Amazon is subsidizing installation now to hold pricing power later, and a holding company gets an AI story it can sell to clients by saying yes in public. The case against is real: Google and Meta have no reason to let a rival's system run buys inside YouTube, Search, Facebook and Instagram, echoing the years Facebook kept its developer interface gated, and agencies must keep rival clients' data walled off, which is where mid-year installs usually break. But Amazon does not need either walled garden's blessing to land this announcement; it needs one holding company willing to be the reference account, and it is paying for exactly that.

Right if: Right if, on or before May 31, 2027, a press release or earnings presentation from Publicis, Omnicom, WPP, Dentsu, Havas, or Stagwell states that the company is using Amazon's Bedrock Agent Core to plan or buy client media on media platforms Amazon does not own. Wrong if: Wrong if no such statement from any of those six companies exists as of May 31, 2027.

Consolidation & Aggression Full Analysis → Listen to the episode →

Profound, the venture-backed startup that tracks how brands appear in AI chatbot answers, will be acquired by a larger advertising, measurement, or marketing-software company on or before 30 September 2027.

Why Matteo Franceschetti, co-founder and CEO of mattress maker Eight Sleep, says he now tracks how his brand surfaces in ChatGPT and Perplexity answers as a separate metric from Google and that the shift is already denting his Google search traffic, and he is steering hundreds of millions of dollars of media spend. That is the demand signal big measurement vendors respond to, and their habit is to buy the category rather than build it: DoubleVerify bought Scibids for AI-driven bidding, Integral Ad Science bought Publica for connected-TV measurement. Profound, which raised venture funding led by Kleiner Perkins in 2025, is the best-known independent tool in a category whose whole value is being first to name and price a new discovery surface, and that window closes fast once incumbents with global advertiser distribution decide to enter. Staying independent through 2027 would mean out-raising and out-selling those incumbents on their own turf at a moment when their sales forces are already fielding the questions Profound answers.

Right if: Right if a binding agreement for another company to acquire Profound, or a majority stake in it, is publicly announced on or before 30 September 2027. Wrong if: Wrong if no such acquisition is publicly announced by 30 September 2027, including if Profound remains independent, shuts down, or only raises further funding.

20VC: 7 Predictions for How AI Changes the World: Labour, Engineering, Social Media, GrokBots Buying Cybercabs and more with Matteo Franceschetti, Co-Founder @ Eight Sleep Listen to the episode →

By June 30, 2027, no advertiser or publisher will be publicly named, by LiveRamp or by itself, as using a LiveRamp age-verification signal to keep ads away from teenagers.

Why Meta's $17-18 billion, ten-year settlement over harm to children turns teen ad targeting into a legal liability, and the roughly $5.3 billion of it that Meta only pays if TikTok and YouTube adopt matching protections is designed to push the same rules across the industry. Publishers and app developers that rely on Meta login or audience extension inherit that obligation and need proof of a user's age from somewhere, which is the opening independent identity vendors like LiveRamp would need to sell into. The problem is that the trustworthy age signal sits on the device: Apple and Google know who set up the phone, and buyers will take an operating-system age signal over a probabilistic match from a third party, especially when the exposure being managed is a courtroom one. Vendors can announce an age product cheaply, but a customer putting its name to it is a legal statement, and few advertisers will publicly vouch for a signal they cannot defend in discovery. The opposite outcome requires an independent vendor to beat the phone makers on accuracy and on liability cover within nine months, which is a long way from where the market is now.

Right if: As of June 30, 2027, neither LiveRamp's public announcements and earnings materials nor any advertiser or publisher statement identifies a named company using a LiveRamp age-verification signal in its advertising. Wrong if: On or before June 30, 2027, LiveRamp or the customer itself publicly names an advertiser or publisher using a LiveRamp age-verification signal in its advertising.

Meta's $17–18B Teen Settlement Sets New Age-Assurance Standard Full Analysis → Read the source story →

The Trade Desk will announce the acquisition of a company whose main business is measuring whether ads actually caused sales, on or before June 30, 2027.

Why Hershey, which spends roughly $2 billion a year across media and retailer trade money, has run a pilot with The Trade Desk and the Australian measurement firm Mutinex that feeds sales-causation data straight into programmatic bidding, so bids chase proven business impact instead of cheap impressions. Whoever controls the connection between measurement and bidding controls a chokepoint every other buying platform has to route through, and The Trade Desk needs exactly that story while AppLovin takes the performance narrative away from it. Mutinex is small, already integrated, and cheap next to the value of owning that wire, which makes buying it the fastest route to a feature rivals cannot copy by signing the same vendor. Leaving the connection open is the comfortable path, but an open version is available to every competing buying platform within months and delivers The Trade Desk no advantage at all. The main risk is timing: the pilot is one advertiser, and The Trade Desk may wait for more proof before spending the money.

Right if: The Trade Desk publicly announces a signed agreement to acquire a company whose main business is measuring whether advertising caused sales, such as an incrementality or marketing-mix measurement firm, on or before June 30, 2027. Wrong if: No such acquisition is announced by The Trade Desk on or before June 30, 2027.

Mutinex and Trade Desk Pilot Incrementality-Driven Programmatic Bidding with Hershey Full Analysis → Read the source story →

By September 15, 2027, either ID5 or Optable will announce that it has been acquired.

Why Publicis paid about $2.2 billion for LiveRamp, roughly five to six times LiveRamp's expected next-year revenue, and that price is now the public benchmark for a company that matches customer records across different marketing systems. ID5 and Optable are the last sizeable independents selling that same neutral function, and the LiveRamp deal both sets a floor under what they are worth and gives cloud, data and measurement buyers a reason to move before another agency group does. Their standalone path is getting harder at the same time, because Snowflake and Databricks sell secure data-sharing as a feature bundled into products brands already pay for, which caps what an independent can charge. Staying independent through the next year would require them to out-wait a far better capitalised owner in a market whose pricing power is shrinking, which is the less likely outcome.

Right if: ID5 or Optable publicly announces a signed agreement to be acquired on or before September 15, 2027. Wrong if: Both ID5 and Optable remain independent with no announced acquisition agreement as of September 15, 2027.

Update: Publicis to Acquire LiveRamp for $2.2B to Power AI Data Agents Full Analysis → Read the source story →

By September 15, 2027, at least one of the 20 largest US advertisers by ad spending will publicly say it is moving its customer-data matching work away from LiveRamp because Publicis now owns it.

Why More than 500 brands ran shared-data projects through LiveRamp specifically because it sat outside every agency group and took no side between them, and Publicis's roughly $2.2 billion purchase ends that. A brand that buys media through more than one agency group now has its data connections owned by one of its agencies' competitors, which is the kind of issue procurement escalates at the next contract renewal rather than absorbs quietly. Rival groups already own substitutes in-house, Omnicom through Acxiom and WPP through InfoSum, so an advertiser leaving has somewhere to go without building anything. Pure inertia is the alternative, and ripping out data plumbing is genuinely painful, but switching costs buy delay rather than permanent silence, and large advertisers routinely make a public example of a vendor whose ownership changed underneath them.

Right if: A company ranked in the top 20 of US advertisers by ad spending states publicly, on or before September 15, 2027, that it is moving or has moved data-matching work off LiveRamp and cites Publicis ownership as a reason. Wrong if: No top-20 US advertiser makes such a public statement on or before September 15, 2027.

Update: Publicis to Acquire LiveRamp for $2.2B to Power AI Data Agents

The California Privacy Protection Agency will announce a privacy penalty of $1 million or more against an ad-tech or data-broker company on or before March 1, 2027.

Why The agency's largest privacy penalty against a data broker to date is about $130,000, a rounding error for the industry, but it has been hiring technologists who can run the same tracking and consent-inspection tools ad-tech firms use. California's privacy law sets fixed per-violation amounts that get multiplied by the number of consumer records involved, so once a technical audit produces a violation count against a company with any scale, the arithmetic passes seven figures without the agency needing to negotiate it there. Having sunk money into technical staff, the agency has every incentive to convert that capacity into a headline number that justifies its budget in the next California fiscal fight. The opposite outcome requires it to keep paying technologists to find violations and then keep pricing them at five figures, which stacks precedent but gives the agency nothing to show for the investment.

Right if: The California Privacy Protection Agency publicly announces a fine, settlement, or order requiring payment of $1 million or more from an ad-tech, data-broker, or identity-data company on or before March 1, 2027. Wrong if: No such penalty of $1 million or more against an ad-tech, data-broker, or identity-data company is announced by the agency on or before March 1, 2027.

California Enforcement Ramping Up With Statutory Fines That Cannot Be Negotiated Read the source story →

By June 30, 2027, Google will open its AI content payment program — the money that shows up in Search Console when a publisher's pages feed Google's AI answers — to any publisher who wants to enroll, ending the invitation-only pilot.

Why Google is currently paying only a few dozen invited publishers, the amounts are tiny, and nothing explains how the figure is calculated. The prize is not the money, it is the pricing shape: whichever model the market gets used to first becomes the starting point for every negotiation after it, and Google's per-use trickle is competing against the lump-sum cheques OpenAI has written to News Corp and other large publishers. That logic only pays off at scale — the more publishers who take a small cheque under Google's terms, the harder it is for anyone to argue the going rate should be a lump sum — and at these rates the expansion costs Google almost nothing. Leaving the program locked to a few dozen invitees would do the opposite: it keeps the price contestable and leaves the lump-sum deals as the only benchmark with real numbers attached. The case against is that the program is purely a legal hedge Google wants to cite in antitrust and copyright fights without ever scaling, but a hedge is worth more, not less, when thousands of publishers have opted in.

Right if: Google's Search Central blog or Search Console product documentation shows the AI content payment program open to publishers without an invitation as of June 30, 2027. Wrong if: As of June 30, 2027 the program is still invitation-only, or Google has discontinued it.

Google quietly scales pay-per-value AI content licensing pilot for publishers Full Analysis → Read the source story →

Omnicom will report negative organic revenue growth for full-year 2026 when it releases its Q4 2026 results in February 2027.

Why Omnicom is eliminating 15,000 roles, roughly 6 to 7% of the combined Omnicom-IPG workforce, to reach a $1.5 billion savings target, and the deepest cuts fall on the creative and brand side rather than media buying, which merged with little friction. Creative staff carry the personal client relationships, so when they leave the accounts they serviced go into review, and rivals recruit hardest from a holding company in the middle of eliminating the agency brands clients originally hired. Cost savings get booked the quarter they happen; the revenue that walks out with departed accounts shows up over the following year, which is exactly the 2026 reporting period. The alternative, clients sitting still through global brand eliminations and mass departures, asks advertisers to ignore the one change that reliably unsettles them, the loss of the named people they work with day to day.

Right if: Omnicom's reported full-year 2026 organic revenue growth, as stated in its Q4 2026 earnings release in February 2027, is below zero. Wrong if: Omnicom's reported full-year 2026 organic revenue growth, as stated in that release, is zero or above.

Omnicom Plans 15,000 Job Cuts Amid IPG Integration Struggles Read the source story →

By 31 October 2027, advertisers will be able to buy ads inside ChatGPT through a buying platform other than Amazon's, announced publicly by OpenAI or by that platform.

Why Amazon's deal to sell ads inside ChatGPT solved OpenAI's immediate money problem, which means OpenAI was the side that needed the deal and Amazon was the side buying a position early. A single preferred buyer caps how much of that inventory actually sells and lets that buyer set the price, and Amazon Ads, already running at roughly $56 billion a year, is also OpenAI's most dangerous rival in shopping search. Every ad marketplace that launched with one anchor buyer, from the early programmatic exchanges to the retail media networks, opened up to more buyers once the format worked, because exclusivity leaves unsold inventory and low prices behind. For the opposite to happen, OpenAI would have to keep handing its rate card to one competitor for more than a year while its own ad revenue is the thing it needs most.

Right if: On or before 31 October 2027, OpenAI or a named buying platform other than Amazon publicly states that advertisers can buy ChatGPT ad inventory through that platform. Wrong if: On 31 October 2027, Amazon's platform is still the only way to buy ads inside ChatGPT, including the case where OpenAI has ended the ad program entirely.

Amazon Ads Launches ChatGPT Pilot With OpenAI Read the source story →

WPP will lose a media account billing at least $500 million a year to a rival holding company with no competitive pitch, as reported by Ad Age, before 31 December 2027.

Why Publicis took PepsiCo's global media, about $1.7 billion in annual billings, away from Omnicom's OMD after 25 years without a pitch, its fourth such win in a year after Microsoft, Paramount and LVMH in Asia-Pacific. The mechanism is switching cost: Publicis runs Epsilon, which holds the client's own customer data, so pulling the media also means pulling the data plumbing, and procurement never opens a price-driven review in the first place. WPP is mid-restructure with no comparable customer-data asset to lock a client in, which leaves its largest relationships defensible only by relationship, and a no-pitch loss gives the incumbent no chance to defend at all. An account of $500 million is under a third the size of the PepsiCo book, so several WPP relationships clear that bar. The alternative, WPP holding every account of that scale or losing one only through an open pitch, requires clients to voluntarily run a process the winning side has every reason to keep closed.

Right if: Ad Age reports before 31 December 2027 that an account billing at least $500 million a year has moved from WPP to another holding company without a formal pitch or review. Wrong if: No such report appears by 31 December 2027, including any case where every WPP loss of that size followed a reported competitive pitch.

Publicis wins PepsiCo global media without a pitch Full Analysis → Read the source story →

Fox Corporation's agreement to buy Roku will be terminated by December 31, 2027.

Why Fox has reportedly bid about $22 billion for Roku, and the Department of Justice has issued a second request for documents, the tool it uses when it wants leverage over a deal rather than a quick sign-off. The same department is mid-fight with Google on the theory that a company owning both a marketplace and a big seller inside it rigs the auction, and a Fox-owned Roku is that arrangement in living-room television: Fox News, Fox Sports and the free streaming service Tubi sitting on top of one of the largest TV operating systems. Any remedy that stops Roku favoring Tubi or Fox-sold ad inventory removes the integration that justifies paying a premium for an operating system, and Fox executive chairman Lachlan Murdoch has already said the two would run separately, which leaves little reason to pay up at all. The Murdoch side has walked from a large deal before rather than overpay, as it did with Time Warner in 2014. A quiet clearance with light conditions is the comfortable outcome for both sides, but it hands Google a live example of the government blessing exactly what it is suing over, which is why the cheaper path for Fox is to stop writing checks to lawyers and drop it.

Right if: Fox Corporation or Roku announces in a securities filing or press release on or before December 31, 2027 that the acquisition agreement has been terminated, or a court blocks it and the parties abandon it by that date. Wrong if: The acquisition has closed, or remains pending under review or litigation, as of December 31, 2027.

MadTech Daily: DOJ Scrutinises Fox’s USD$22bn Roku Deal; US Warns UK Over Plan to Boost Public Service News on Social Media Listen to the episode →

PubMatic's first-quarter 2027 revenue, reported in May 2027, will be lower than its first-quarter 2026 revenue.

Why Judge Leonie Brinkema declined to break up Google's ad-tech business and instead ordered behavioral remedies, the central one being that Google's AdX exchange must bid through Prebid, the open-source setup that lets several exchanges bid on the same impression at once. Prebid president Garrett McGrath has said the technical work is an adapter built in days, so the change is cheap for everyone and confers no lasting advantage on any one independent exchange. What it does not touch is the asset that actually prints money: Google still controls the advertiser demand, and publishers route inventory to wherever the demand is, so more even auctions on the sell side do not create new buyers. Any incremental fill the independents pick up early gets competed away within a quarter or two, which leaves PubMatic fighting the same pricing pressure it faced before the ruling. The opposite outcome would require the remedy to shift buyer relationships rather than auction plumbing, and nothing in the order does that.

Right if: PubMatic's reported revenue for the quarter ending March 2027 is below its reported revenue for the quarter ending March 2026. Wrong if: PubMatic's reported revenue for the quarter ending March 2027 is equal to or above its reported revenue for the quarter ending March 2026.

Auctions In The Crosshairs Listen to the episode →

By 31 December 2027, the developer of an already-announced data-center campus in Virginia will publicly abandon or relocate the project after a county or town board votes it down.

Why Virginia's data-center corridor has dozens of pending proposals queued at a moment when grid operators there have flagged that the new load is pushing up what households pay for electricity, and local boards are the officials who hear that complaint first and face voters soonest. Approving power-hungry campuses has been the default because the tax revenue is large, and the usual developer response to a rejection is to redraw the site plan and come back, which is why a clean walk-away is the harder outcome to hit. What makes it plausible now is that the same companies hold options on pre-permitted sites in Ohio, Georgia and Texas, and transmission capacity in northern Virginia is already tight, so fighting a hostile board costs more time than moving does. If local anger stays at the level of loud meetings while every board eventually approves, the compute buildout absorbs the noise and the cost floor for AI capacity keeps falling on schedule.

Right if: A named developer publicly cancels or moves a previously announced Virginia data-center campus following a rejection vote by a Virginia county or town board, evidenced by that board's meeting record and the developer's own public statement, on or before 31 December 2027. Wrong if: No such public cancellation or relocation following a Virginia local rejection vote is on record by 31 December 2027.

The Bulwark’s Sarah Longwell on How Democrats Can Defeat MAGA for Good Listen to the episode →

Madison and Wall's next quarterly global ad growth update, covering the third quarter of 2026, will report growth below 11%.

Why Global ad growth ran 13.4% in the first quarter of 2026 and 12.7% in the second, against a full-year forecast of 11% and a total above $1.3 trillion, so the line is already bending downward before any macro shock. Much of the lift comes from AI companies buying users hard and from price inflation on scarce premium inventory, both of which front-load spending into a single year and then settle once the land-grab phase passes. Year-ago comparisons also get harder in the back half, because the AI advertising surge was already underway by late 2025. For the quarter to hold at 11% or better, that acquisition spending and premium pricing would need to keep climbing rather than level off, which is the least durable part of the model.

Right if: Madison and Wall's published global ad growth figure for the third quarter of 2026 is below 11%. Wrong if: Madison and Wall's published global ad growth figure for the third quarter of 2026 is 11% or higher.

Global Ad Spend to Top $1.3 Trillion in 2026, Growing 11% Full Analysis → Read the source story →

PubMatic's full-year 2026 revenue, reported in its fourth-quarter results in February 2027, will be lower than its full-year 2025 revenue.

Why The global ad market is growing about 11% in 2026, but the growth is pooling in Amazon, Meta, and Google, where an advertiser can see the sale happen inside the platform's own system, and in retail media taking budget from linear TV. PubMatic sells open-web inventory that sits outside those walls, so it captures a shrinking slice of each new dollar even while the headline market expands. Independent sell-side platforms also face pricing pressure from buyers who route more spend through direct platform deals at better terms, which compresses both volume and take rate at once. A rising year would require open-web programmatic to grow at least as fast as the market it is losing share to, which the concentration trend argues against.

Right if: PubMatic's reported full-year 2026 revenue comes in below its reported full-year 2025 revenue. Wrong if: PubMatic's reported full-year 2026 revenue is equal to or above its reported full-year 2025 revenue.

Global Ad Spend to Top $1.3 Trillion in 2026, Growing 11%

The UK government will lay a bill before Parliament on or before 30 June 2027 that requires developers of frontier AI models to give the UK AI Security Institute access to those models before public release.

Why Anthropic shipped its newest Claude model without giving the UK AI Security Institute a pre-release look, while several US organisations still received early access, which ended what had been standing practice. Voluntary review survives only while every lab participates, and a lab that skips a national regulator with no penalty hands its competitors the same permission on their own release schedules. The UK has already renamed its Safety Institute the Security Institute and has been building toward enforcement powers, and regulators that lose access typically legislate to get it back rather than ask twice. The alternative path, ministers accepting that pre-release testing is a favour labs grant when convenient, leaves the institute with no leverage over the exact companies it was created to inspect, which is a hard position for a government to defend publicly after a skipped evaluation.

Right if: A bill laid before the UK Parliament on or before 30 June 2027 contains a requirement that frontier AI developers provide the UK AI Security Institute with access to a model prior to its public release. Wrong if: No bill containing such a requirement is laid before the UK Parliament on or before 30 June 2027.

MadTech Daily: UK Safety Body Denied Access to Anthropic’s Mythos 5.1; M+C Saatchi MBO Falls Apart Full Analysis → Listen to the episode →

Before May 31, 2027, one of Netflix, Disney, Warner Bros. Discovery, Paramount, NBCUniversal, Fox or Roku will publicly announce that it is replacing the ad server that decides which commercials run on its own streaming inventory.

Why Publisher-side ad servers in streaming have been protected less by their technology than by the fear of migration: swapping the system that picks and places every ad is a months-long project that can end a career, and since Netflix moved off Microsoft's ad technology in 2024 no other large US streamer has changed its core ad server. Challengers are now selling side-by-side testing, running a second ad server against a slice of inventory so a publisher can compare the revenue before committing, which turns an irreversible decision into a reversible experiment and prices the incumbent's real advantage at close to zero. Streaming margins are thin enough that a verified single-digit revenue lift pays for the ad-operations pain, and a challenger with no reference accounts will discount aggressively to buy the first logo. Total stasis is the weaker case, because Google Ad Manager, FreeWheel and SpringServe contracts come up for renewal on ordinary cycles and Netflix already demonstrated publicly that a large streamer can change ad servers and survive it.

Right if: On or before May 31, 2027, Netflix, Disney, Warner Bros. Discovery, Paramount, NBCUniversal, Fox or Roku states publicly, or its incoming vendor states publicly, that it is moving its owned streaming inventory to a different primary ad server. Wrong if: As of May 31, 2027, none of those seven companies has made such an announcement.

Rebuilding the CTV Ad Server: Momentiv's Sandro Catanzaro Full Analysis → Listen to the episode →

A court will pause, suspend, or postpone Google's obligation to share expanded auction-level data with rival ad tech firms, ordered in the US ad tech antitrust case, on or before April 15, 2027.

Why Judge Leonie Brinkema found Google's ad tech business an illegal monopoly but declined to order a breakup, leaving behavioral fixes: publishers can set higher price floors against Google's tools, the last-look advantage that let Google see the final bid before an auction closed is gone, and Google must share more auction-level data with third parties. Google is appealing, and the data-sharing requirement is the one piece that forces it to build and operate new engineering for competitors while the appeal runs, which makes a request to pause it the cheapest and most obvious first move for its lawyers. A judge who has already chosen behavioral relief over a breakup has signalled tolerance for a slower path, and courts are more willing to hold back build-it-now obligations than conduct bans because an appeals court cannot un-ring the bell once rivals have the data. The alternative, Google standing up a production feed for Magnite, PubMatic, The Trade Desk and others while arguing on appeal that it should never have to, spends money it cannot recover and hands competitors signal it cannot claw back.

Right if: Right if an order on the docket in United States v. Google LLC in the US District Court for the Eastern District of Virginia, or from the Fourth Circuit, pauses, suspends, or postpones the auction-data-sharing requirement in whole or in part on or before April 15, 2027. Wrong if: Wrong if no such order appears on that docket by April 15, 2027, whether because Google never asked for one or because the request was denied.

Judge Rules Google Need Not Spin Off Ad Tech Business Full Analysis → Read the source story →

Criteo will announce an agreement to be acquired or taken private on or before December 31, 2027.

Why Criteo's market value has fallen below $1 billion and buyers hunting distressed assets have already approached it, with the board so far turning them away. Unlike The Trade Desk, where founder and chief executive Jeff Green holds super-majority voting control and can simply refuse any deal, Criteo has no controlling shareholder, so its board answers to ordinary investors who can be worn down by a cash offer priced above a depressed stock. The business still generates real cash from retail media and performance advertising, which is the exact profile private equity buys at this kind of price. Staying independent requires the board to keep refusing offers quarter after quarter while shareholders sit on losses, and that resistance usually breaks before the discount does.

Right if: Criteo publicly announces a signed agreement to be acquired or taken private on or before December 31, 2027. Wrong if: No such announcement from Criteo exists as of December 31, 2027.

Jeff Green's Super-Majority Control Blocks Acquisition Premium Read the source story →

Viant Technology's revenue for the first half of 2027, reported in its second-quarter 2027 earnings release in August 2027, will grow more slowly year over year than its first-half 2026 revenue grew.

Why Viant's shares rose about 30% over the past year on the argument that The Trade Desk is losing core open-web buying to cheaper, narrower rivals, yet that gain came off a depressed base during a broad ad-tech recovery and no challenger has shown a single large advertiser moving its general buying. The specialists winning today (healthcare and out-of-home buying tools) take vertical budget, which is additive and small, while the general-purpose seat stays put because agency planners consolidate platforms to cut logins, billing reconciliation and broken integrations, and each extra platform carries its own minimum spend and seat fee. For Viant's growth to accelerate instead, holding companies would have to start treating breadth of reach as waste they should pay less for and shift general budget wholesale, a trade buyers have refused repeatedly in ad tech. A narrative repricing lifts a stock quickly; it shows up in booked media dollars slowly, if at all.

Right if: We're right if Viant's reported first-half 2027 revenue growth rate is below its first-half 2026 growth rate. Wrong if: We're wrong if that growth rate is equal to or above the first-half 2026 rate.

Viant Gains 30% as Challenger DSPs Exploit TTD Vulnerabilities Read the source story →

The Trade Desk's full-year 2026 revenue, reported in its fourth-quarter earnings in February 2027, will be higher than its full-year 2025 revenue.

Why Media analyst Michael Nathanson argues that connected TV advertising has consolidated around five scaled sellers — YouTube, Disney, Amazon, Netflix, Fox paired with Roku, and Paramount paired with Warner Bros Discovery — and that agencies will increasingly buy from them directly, cutting out the independent software layer The Trade Desk sells. Buying direct from five sellers means five contracts, five incompatible reports, and no single view of how many times one household saw the same ad, which is the practical reason large agencies keep routing money through one buying platform. Amazon, Netflix, and Roku also still need programmatic pipes to reach advertisers they have no direct sales relationship with, so the inventory does not leave the platform simply because ownership of it has concentrated. Erosion of this kind shows up first as a thinner cut per dollar over two to three years, not as dollars disappearing inside twelve months. An outright annual revenue decline would require agencies to unwind unified buying faster than any previous direct-buy push has managed.

Right if: The Trade Desk's reported full-year 2026 revenue exceeds its reported full-year 2025 revenue. Wrong if: The Trade Desk's reported full-year 2026 revenue is equal to or below its reported full-year 2025 revenue.

CTV Consolidation Narrows to Five Scaled Players, Crowding Out TTD Read the source story →

By October 15, 2027, at least one of the three largest independent marketing mix modeling firms — Analytic Partners, Ekimetrics, or Nielsen's marketing mix business — will announce that it has been acquired, merged into a larger consultancy, or shut down.

Why Google now gives away Meridian, a free open-source tool that estimates how much each marketing channel drove sales, and has just added a build-and-debug chatbot, a graduated geo-testing feature, and the ability to use Google's own branded search volume as the measure of brand health. Meta has stopped pushing Robyn, its rival free tool, which removes the only other no-cost option and hands Google the adoption window. Independent modelers charge six figures for an engagement and are now squeezed from below by free software that mid-market advertisers will accept and from above by consultancies bundling modeling into transformation work. The opposite outcome requires enough advertisers to keep paying full freight for neutrality at a moment when the free alternative is good enough for a board slide, which is exactly the buying behavior that has collapsed in every other measurement category. Consolidation is how squeezed specialist vendors exit, and the private-equity backing behind these firms shortens the runway.

Right if: Analytic Partners, Ekimetrics, or Nielsen's marketing mix modeling business publicly announces a sale, a merger into a larger consultancy or holding company, or a wind-down on or before October 15, 2027. Wrong if: All three are still operating as going concerns under their existing ownership with no such announcement as of October 15, 2027.

Google Meridian MMM Adds Chatbot, GeoX, and Upper-Funnel Signals Full Analysis → Read the source story →

By June 30, 2027, a named brand advertiser will publicly say it increased its spending on Roku Channel, Tubi, or Pluto TV because an outside measurement firm such as Nielsen or Comscore verified that its ads ran.

Why US viewing of free ad-supported streaming channels rose 43% to 1.8 billion hours through August 2025, yet the price of a thousand ad impressions on that inventory has fallen from around $30 to single digits and revenue per viewing hour at some operators has dropped from 18 cents to 8 cents. The industry's flattering explanation is that buyers discount what they cannot verify, which implies that letting an outside referee in should bring premium brand budgets back and produce a case study the platform would shout about. The simpler explanation is oversupply: hundreds of near-identical channels of mid-tier content, which no verification tag repairs, and buyers who already pour money into YouTube and TikTok with weaker proof. If verification genuinely re-rated the inventory, the first seller to prove it would market that proof hard during upfront season, so the absence of any advertiser willing to attach its name to a spending increase is the strongest available evidence that the discount is about the content, not the data.

Right if: On or before June 30, 2027, a named advertiser states publicly, or is quoted by Roku, Tubi, or Pluto TV, that it raised its spending on that service because of outside verification that its ads ran. Wrong if: No such named-advertiser statement about Roku Channel, Tubi, or Pluto TV appears publicly by June 30, 2027.

FAST Platforms Hoarding Audience Data, Strangling Ad Economics Full Analysis → Read the source story →

Coca-Cola will publicly hand its media planning and buying business, either globally or for North America, to Omnicom by August 15, 2027.

Why Publicis won PepsiCo's roughly $100 million-a-year media account after 25 years with Omnicom, and no holding company runs the media for two direct beverage rivals, so Coca-Cola's business is now loose and in market. Omnicom is the obvious landing spot: it has the global footprint, it has publicly signaled new flexibility to pursue clients it was previously conflicted out of, and it needs a blue-chip win to change the story after losing PepsiCo. The alternatives are weaker: WPP's GroupM and Dentsu have no comparable first-party data and AI pitch to answer the one that just moved PepsiCo, and independents lack the worldwide buying scale a Coca-Cola assignment demands. The live risk is the calendar rather than the logic, since large packaged-goods reviews routinely run four to six quarters and Coca-Cola could split the work by region or stall past mid-2027.

Right if: Coca-Cola or Omnicom publicly announces on or before August 15, 2027 that an Omnicom agency has been appointed to Coca-Cola's media planning and buying, globally or for North America. Wrong if: No such announcement exists by August 15, 2027, whether the work goes to another agency group, to an independent, or stays where it is.

Omnicom CFO admits PepsiCo loss to Publicis blindsided leadership Full Analysis → Read the source story →

OpenAI will announce an agreement to acquire an advertising technology or ad sales company on or before 30 September 2027.

Why OpenAI is telling investors it will book $100 billion in advertising revenue by 2030, which would require growing ad sales more than 200% a year from a business that only began wiring in outside partners in February 2026: Criteo for retargeting, StackAdapt for programmatic display, and Amazon's demand-side platform for buyer demand. What is missing is not a feature but an operation, including an ad sales team, ad formats, pricing, and measurement, and each of those takes years to build while competing for engineering attention with the core model work. Buying a company that already has sellers, buyer relationships, and delivery technology is the only route that closes that gap on anything like the timeline being sold to investors. The alternative, staying a landlord who rents everyone else's plumbing, hands the margin and the advertiser relationship to Amazon and Criteo and leaves OpenAI as a supply source with no path to the number it has promised.

Right if: OpenAI publicly announces, on or before 30 September 2027, an agreement to acquire a company whose primary business is advertising technology or advertising sales. Wrong if: No such acquisition agreement is announced by OpenAI on or before 30 September 2027.

OpenAI Targets $100B Ad Revenue by 2030, Analysts Skeptical Full Analysis → Read the source story →

PubMatic's full-year 2026 revenue, reported in its fourth-quarter results in February 2027, will be lower than its full-year 2025 revenue.

Why The court let Google keep both its ad exchange and the ad server most large publishers run, killing the forced-sale scenario that was supposed to hand open-web supply back to independent exchanges. Google's sell-side sales force can now present those products as court-validated and structurally intact, which hardens publisher renewals at exactly the moment independent exchanges lose the regulatory argument they had been using to pry accounts loose. PubMatic's revenue is concentrated in open-web display, the part of the market where that pitch competes most directly with Google. Revenue growth in 2026 would require PubMatic to win publisher supply in the one arena where Google just received a clean bill from the court, or to add enough connected-TV business to cover the shortfall, and neither has shown up at that scale.

Right if: PubMatic's reported full-year 2026 revenue comes in below its reported full-year 2025 revenue. Wrong if: PubMatic's reported full-year 2026 revenue matches or exceeds its reported full-year 2025 revenue.

Google escapes ad-tech breakup; behavioral remedies now the focus Full Analysis → Read the source story →

Fox's acquisition of Roku will not have closed by December 31, 2027.

Why The Justice Department has issued a second request for information on the deal, and while that review runs the two companies cannot coordinate ad sales at all. The prize in the deal is Roku's account-level identity across more than 80 million active accounts attached to Fox's live sports and news, which would let one owner match premium inventory to known viewers and sell the whole package end to end; that same combination of viewer data and must-have content is what makes a regulator look twice, and data questions take longer to work through than old-fashioned distribution and carriage questions. A review that stretches past a year also invites a rival bid for Roku or a renegotiated price, either of which restarts the clock. Clearance inside fifteen months would require the government to drop a data question it opened on the most valuable piece of the transaction.

Right if: Neither Fox Corporation nor Roku has announced completion of the merger in a public filing or press release on or before December 31, 2027, including any case where the deal is abandoned or blocked. Wrong if: Fox and Roku complete the merger on or before December 31, 2027.

DOJ Requests More Information on Fox-Roku Merger Deal Read the source story →

Booking Holdings will spend the same share of its revenue on marketing in the first half of 2027 as it did in the first half of 2026, or a larger share, when it reports second-quarter results in August 2027.

Why The EU's Digital Markets Act, the law forcing large platforms to give rivals fair access, has made Google place travel sites above its own paid boxes on European searches, and Google paid a 460 million euro fine in July over the same conduct. If that placement genuinely hands Booking free, high-intent traffic, the cleanest evidence is Booking needing to buy less of it: marketing has run near 30% of its revenue in recent years, and a real windfall would push that ratio down. The likelier path is that European users scroll past mandated rival links to the brand and the format they already know, while Google keeps re-tuning a compliant layout that still favors its own units, leaving Booking bidding just as hard for the same clicks. Booking also has a standing habit of pushing every efficiency gain back into paid acquisition to defend share against Expedia and Airbnb, which works against the ratio falling even if free traffic rises. A drop in that share would be the strongest sign yet that a layout remedy can actually pry a click away from Google.

Right if: Marketing expenses as a percentage of revenue for the six months ended June 2027, in Booking Holdings' August 2027 results, are equal to or higher than the same measure for the six months ended June 2026. Wrong if: That percentage is lower than the same measure for the six months ended June 2026.

Google Forced to Restructure EU Search to Show Rival Results Full Analysis → Read the source story →

Momentiv, the startup selling software that sets ad prices and deal priority for publishers, will be acquired, shut down, or no longer marketing itself as an automated decision-maker for publisher pricing, according to its own public announcements and website on December 31, 2027.

Why Momentiv, backed by a seed check from Aperiam Ventures, is pitching publishers software that takes over floor pricing and deal prioritization inside the ad server. Publishers run Google Ad Manager because it arrives welded to Google's exchange, its demand and its reporting, and replacing it is a multi-quarter migration with live revenue at risk, so any challenger needs documented yield gains on real inventory that nobody has produced yet. Even publishers who stay put have to hand the software authority over prices rather than suggestions a human approves, and when a deal gets underpriced there is no clean owner of the miss, which is why most buyers stop at recommendations. For the company to still be independent and selling the same pitch at the end of 2027, publishers would have to both accept the migration and delegate pricing authority within roughly a year, and the court case that could force Google's bundle apart is unlikely to have changed any publisher's server by then.

Right if: On December 31, 2027, Momentiv has been acquired, has ceased operations, or its public materials no longer describe its product as making pricing or deal-priority decisions for publishers. Wrong if: On December 31, 2027, Momentiv is still an independent operating company whose public materials describe its product as making pricing or deal-priority decisions for publishers.

Aperiam Ventures invests in Momentiv; notes sell-side agentic gap Read the source story →

DoubleVerify's full-year 2026 revenue, reported in its February 2027 results, will be lower than its full-year 2025 revenue.

Why Google now answers many queries on its own results page, so the outbound click that used to send a reader to a publisher never happens, and the open-web page views those clicks created disappear with it. DoubleVerify gets paid per ad impression it checks, so its core business shrinks in direct proportion to open-web display volume, and publishers cannot manufacture replacement inventory from email lists, apps and podcasts fast enough to matter. The usual counterargument is that verification of ads on connected TV and social platforms grows fast enough to cover the open-web decline, but that work is priced lower, contested by rivals, and increasingly handled by the platforms' own tools, which caps how much of the gap it can close. A flat or growing year would mean open-web impression volume held up despite the search-referral collapse, which the traffic data does not support.

Right if: DoubleVerify's reported full-year 2026 revenue is below its reported full-year 2025 revenue. Wrong if: DoubleVerify's reported full-year 2026 revenue is equal to or above its reported full-year 2025 revenue.

Publishers Pivot to Direct Audience Relationships as Google Traffic Erodes Full Analysis → Read the source story →

By July 31, 2027, at least one of the 20 largest US advertisers will publicly move media spending away from Omnicom and say undisclosed agency markups are the reason.

Why Omnicom sold its barter unit ICON International in 2021 and then built four operations — Omnet, Apex, Midas and Orion — that buy ad space with Omnicom's own money and resell it to clients at a markup the client never sees. The Association of National Advertisers named this conflict publicly in 2015 and holding-company billings grew every year since, so nothing about the practice is secret; what changed is the size of the gap, because inventory prices fell after 2022 while client rate cards held, widening the spread the agency keeps. That makes this a finance problem rather than a marketing one: an audit now produces a number large enough that a CFO has to answer for it at the annual contract renewal, and independent auditors are already carrying the inbound. The case for nothing happening is that procurement teams chose these deals for a lower headline fee and would have to admit they were charged in a place they weren't looking — real, but that confession gets cheaper the moment one large advertiser goes first and the rest can point at it.

Right if: A company among the 20 largest US advertisers by measured ad spend publicly states before July 31, 2027, as reported by Ad Age, that it is cutting, moving or withholding media spending from Omnicom because of undisclosed markups on media Omnicom bought with its own money. Wrong if: No such public statement from a top-20 US advertiser is reported by Ad Age on or before July 31, 2027.

Omnicom Built Principal Media Empire After Selling Barter Unit Read the source story →

Through the close of the 2027 upfront negotiations on June 30, 2027, no named advertiser will publicly credit Gracenote's show-level TV content data with lifting sales or purchase intent.

Why Gracenote, the Nielsen-owned television content database whose show and episode data now feeds targeting inside The Trade Desk, Index Exchange, and PubMatic, has exactly one public result behind the pitch: a Dos Equis campaign bought through Dentsu, reported by Gracenote itself, claiming zero wasted impressions, 80% of ads running inside live play, and 7% better cost per thousand. Those are delivery numbers, and none of them says more beer was sold. Gracenote's head of global advertising sales calls the link between show-level targeting and purchase intent an early conversation, which is what a vendor says when the study does not exist. A proper holdout test, content-targeted against untargeted, costs money and carries the risk of showing the signal moves nothing, so neither Gracenote nor the brands paying a premium have a reason to fund and publish one inside a year. The buyers getting real value today are brand-suitability teams keeping airline ads off plane-crash dramas, and they have no sales result to announce.

Right if: As of June 30, 2027, no advertiser has been named in a Gracenote case study, press release, trade-press report, or conference presentation claiming a sales or purchase-intent lift from Gracenote show-level content targeting. Wrong if: On or before June 30, 2027, at least one named advertiser publicly claims a sales or purchase-intent lift from Gracenote show-level content targeting.

Beyond the Genre: Bill Condon on Gracenote’s Vision for CTV Transparency Full Analysis → Listen to the episode →

By May 15, 2027, a retailer, marketplace, or social platform that owns shopper purchase data will announce the acquisition of an independent company selling performance-based TV advertising.

Why Walmart bought Vibe and Pinterest bought tvScientific inside the past year, and both buyers had the same gap: they can see what people buy but they do not own the television screen, so the fastest way to tie TV exposure to a purchase is to buy the company that already does the wiring. The remaining independents in this niche are small, venture-funded, and priced within easy reach of any large retail media business chasing television budget. Every major retailer selling advertising faces the same build-or-buy choice, and building an attribution path from a smart TV impression to a checkout takes years. For this call to fail, the retail and commerce platforms currently expanding their advertising businesses would have to conclude that television conversions are not worth owning, which contradicts the two deals already closed and the money they are spending elsewhere to prove incremental sales.

Right if: On or before May 15, 2027, a retailer, marketplace, or social platform with its own shopper purchase data publicly announces it is acquiring an independent performance TV advertising company. Wrong if: No such acquisition is publicly announced by an acquirer of that type on or before May 15, 2027.

Ep 150: Apps, Ops and OEM with Mike Brooks Full Analysis → Listen to the episode →

Meta will announce an agreement to buy a company whose main business is making or optimizing advertising creative, on or before June 30, 2027.

Why Amazon switched its sponsored-products auction from second-price pricing, where the winner pays just above the runner-up's bid, to a hybrid with a hidden floor that charged winners more, and buyers shrugged because their return-on-ad-spend targets were still being met; the same month, the US court ordered remedies against Google that left its ad server and exchange intact. With regulators declining to break up the largest platforms and targeting and auction mechanics now automated, the ad creative itself is the last part of a campaign the platforms do not yet produce, and Meta is already collecting advertisers' creative variants through its optimization tools to train systems that will make that creative. Buying an existing creative production and testing platform gets Meta the workflows, the client relationships and the training data years faster than rebuilding them, and the Google ruling lowered the perceived antitrust cost of doing a deal. The alternative, Meta building the whole creative layer in-house, leaves the production workflow in agency and independent-vendor hands for the period when the platform is trying to close budget-in, results-out automation.

Right if: Meta publicly announces a signed agreement to acquire a company whose main business is producing, generating or optimizing advertising creative on or before June 30, 2027. Wrong if: No such acquisition agreement is announced by Meta on or before June 30, 2027.

The Amazon-FTC case and the myth of ad auction transparency Full Analysis → Listen to the episode →

By October 31, 2027, California will still have no signed law that forces the companies building the largest AI models, such as OpenAI, Anthropic, and Google, to meet enforceable safety-testing or guardrail requirements before releasing them.

Why Ripple co-founder Chris Larsen put $3.5 million behind California's most recent AI safety push against a target of $15 million, while OpenAI's political arm was reported to have committed $50 million to $100 million to defeating it, and the candidate carrying that effort lost. A roughly seven-to-one money gap is trivial for balance sheets already spending billions a year on computing power, so the opposition can keep funding it indefinitely in the state where most of the labs are headquartered. California has been here before: SB 1047, the 2024 attempt at frontier-model rules, passed the legislature and was vetoed. For the opposite to happen, the outspent side would have to win both a floor vote and a governor's signature in the one legislature most exposed to the industry's lobbying money, which is the harder path. The practical consequence is that anyone building AI into a product has no new California compliance burden landing in this cycle.

Right if: As of October 31, 2027, the California Legislative Information database shows no enacted state law imposing enforceable pre-release safety-testing or guardrail duties on developers of the largest AI models. Wrong if: At least one such law has been signed and appears as enacted in the California Legislative Information database on or before October 31, 2027.

Why This Crypto Billionaire Wants to Slow Down the AI Arms Race Listen to the episode →

Free ChatGPT users in the United States will be seeing paid advertising or sponsored placements inside the product on or before 31 December 2027, confirmed by OpenAI's own product and public announcements.

Why The people building AI assistants argue that ads are impossible in this format: the cost of answering each question is too high for a free ad-supported tier to make money, and a user who suspects the assistant is steering them toward whoever paid stops trusting the answer. Both arguments rest on a cost curve that is falling fast, with per-query compute costs roughly halving every nine to twelve months, and on a subscription base that excludes the large majority of users who will never pay $20 a month. OpenAI carries the compute bill for those free users today and earns nothing from them, which is exactly the position that historically produces an ad product. The alternative, holding a subscription-only model while the free base keeps growing, means paying indefinitely for users it has chosen not to monetize, and the competitive cover for refusing ads disappears the moment Meta puts commercial placements into its own assistant inside WhatsApp.

Right if: Paid advertising or sponsored placement is visible to free-tier ChatGPT users in the United States at any point on or before 31 December 2027. Wrong if: No paid advertising or sponsored placement is visible to free-tier ChatGPT users in the United States as of 31 December 2027.

20VC: The $100 Billion AI Assistant Race: Town vs Instinct vs GrokBot | We Spend $75K Per Engineer on AI Tools | Why the AI Assistant Market Is Not a Bubble & AI Assistants Will Replace Every App on Your Phone with JD, Founder of Town Listen to the episode →

Samba TV, the company that sells data on what shows are playing on TV screens, will be acquired, merged into another company, or shut down by December 31, 2027.

Why Gracenote has now pushed show-level programming data into The Trade Desk, Index Exchange and PubMatic, the three pipes that carry most open-market connected-TV buying outside Amazon, Google and Meta, so identifying what is on screen becomes a default field in the buying tool instead of a product a buyer shops for. Samba TV sells exactly that knowledge as its reason to exist, and it shelved a planned public listing in 2022 without ever reaching the scale that listing assumed. Once the signal is free inside the biggest buying platforms, the standalone vendor loses pricing power on its core product and its remaining value sits in its data and TV-maker relationships, which is what buyers of distressed data companies pay for. Staying independent through 2027 would require Samba TV to win back a premium for information that agency traders can now tick on by default, and the TV manufacturers it competes with (LG, Vizio under Walmart, Roku) already own that screen data without paying for it.

Right if: By December 31, 2027, Samba TV has announced, or an acquirer has confirmed in a public statement or securities filing, that Samba TV has been acquired, merged into another company, or wound down. Wrong if: On December 31, 2027, Samba TV is still operating as an independent company with no announced acquisition, merger, or shutdown.

Gracenote Partners With The Trade Desk to Distribute CTV Content Signals Full Analysis → Read the source story →

Comscore's full-year 2026 revenue, reported in its year-end results in early 2027, will be lower than its full-year 2025 revenue.

Why Comscore announced a restructuring targeting $20 to 25 million in annual savings against a revenue base of roughly $400 million, days after Nielsen agreed to buy ad-verification firm DoubleVerify for about $2 billion. Cost cuts of that depth extend runway; they do not arrive when customers are pulling harder on the product. Every renewal conversation now opens with the customer's doubt about whether the vendor will be around to service the contract, which shows up first as longer sales cycles and deeper discounts. For revenue to grow instead, Comscore would have to take share in television and local-market measurement precisely while nervous agency planners consolidate spending back toward the incumbent.

Right if: Comscore's reported full-year 2026 revenue is below its reported full-year 2025 revenue. Wrong if: Comscore's reported full-year 2026 revenue is equal to or above its reported full-year 2025 revenue.

VideoAmp and Comscore Both Cut Staff Near Nielsen Acquisition Announcement Full Analysis → Read the source story →

On or before June 30, 2027, one of the three largest agency groups (WPP Media, Publicis Media, or Omnicom Media Group) will publicly announce that it is buying local broadcast TV against audience numbers from VideoAmp, Comscore, or iSpot, not Nielsen's alone.

Why On August 31, 2026 Nielsen cut how long a household must watch before it counts as a local TV viewer, from five minutes within a quarter hour down to one minute, saying the old rule left up to 24% of tuning events unreported. Reported audiences grew in every US market without one extra person watching, and the change landed in the week stations set fall rates, so buyers who guaranteed delivery against the smaller pool now pay the same cost per thousand for shorter, thinner exposure. Nielsen made the switch without the Media Rating Council's audit stamp, which hands VideoAmp, Comscore, and iSpot their first dated, documented argument after years of arguing in the abstract and weakens the enforceability of guarantees written on Nielsen numbers. The big agency groups already run second measurement suppliers in national TV; local broadcast was the last market Nielsen owned outright. Quiet acceptance is the less likely path because the inflation shows up directly in the agencies' own year-end reconciliations with clients, where someone has to explain why the client paid the same money for more one-minute tune-ins.

Right if: By June 30, 2027, WPP Media, Publicis Media, or Omnicom Media Group has stated publicly, in its own announcement or in reported comments from a named executive, that its local broadcast television buying uses VideoAmp, Comscore, or iSpot audience numbers. Wrong if: By June 30, 2027, none of those three groups has made such a public statement about local broadcast television.

Nielsen Quietly Shortens Local TV Viewing Qualifier, Inflating Audiences Full Analysis → Read the source story →

Google will raise its published Buyer Direct fee above 10% for at least some buyers before October 31, 2027.

Why Google charged roughly 33% combined across its buying and selling tools for a decade, then priced Buyer Direct at 10% while a federal judge weighs remedies in the US ad-tech antitrust case. A cut that deep buys volume: it keeps advertisers transacting inside Google's system at the exact moment they have the most reason to build an alternative on open-source header-bidding software. Google sets this rate unilaterally and can move it by buyer tier or bolt on service and data charges, so restoring margin costs it nothing except a news cycle. Holding at 10% through 2027 means permanently giving up billions in margin on its own inventory, which it would do only under a court order capping the rate or a judgment that any increase invites fresh regulatory trouble.

Right if: Google's published Ad Manager pricing for Buyer Direct shows a fee above 10% for any group of buyers on or before October 31, 2027. Wrong if: Google's published Ad Manager pricing for Buyer Direct stays at 10% or lower for all buyers through October 31, 2027.

They Didn't Win Sh_t Listen to the episode →

By December 31, 2027, at least one of Index Exchange, OpenX, Magnite, or PubMatic will publicly announce that it is being acquired by or merged into another company.

Why The US court that found Google liable in the ad-tech case let it keep Google Ad Manager, the ad server publishers use to sell inventory, and AdX, its exchange, so no Google assets will be broken off and no Google demand will scatter into the open exchange market. Index Exchange and OpenX had positioned partly around picking up that fragmentation, and Magnite and PubMatic each carried a quieter version of the same upside in their optimistic case. With that possibility closed for years and open-web display losing budget to retail media and connected TV, the remaining route to scale for these four is combining customer bases and cost structures with each other or selling to a larger buyer. The alternative, all four funding their own connected-TV and retail-media buildouts while competing with Google, Amazon, and Meta at once, is the more expensive path and the one their investors are least likely to keep financing through 2027.

Right if: On or before December 31, 2027, Index Exchange, OpenX, Magnite, or PubMatic announces in a company press release or regulatory filing that it has agreed to be acquired by or merged into another company. Wrong if: None of Index Exchange, OpenX, Magnite, or PubMatic has announced such an agreement by December 31, 2027.

Google Antitrust Remedy: No Divestiture of Ad Server or Exchange Read the source story →

Google will start counting display ad impressions only once the ad begins painting on screen on February 17, 2027, the date it announced, with no delay to that date.

Why In September 2026 Google set February 17, 2027 as the day an impression stops counting at ad request or delivery and starts counting at first paint, which mechanically shrinks every publisher's reported inventory on the same day without changing a dollar of revenue. The five-month runway is short by Google's standards, and unlike the repeatedly delayed removal of third-party cookies, this change needs no replacement technology, no regulator sign-off, and no agreement from buyers or rival vendors before it can take effect; it is a counting rule inside Google's own serving and reporting stack. Google also has an incentive to move quickly, because its own inventory sits inside Chrome's rendering path and can observe a paint event more precisely than a publisher measuring through a third-party ad server. The case for a slip rests on messy render signals in apps and connected TV, but those are separate inventory types that can be handled with their own timelines while standard display switches on schedule. A delay would require Google to abandon a definitional advantage it controls unilaterally.

Right if: Google's ad products are counting standard display impressions on the begin-to-render basis as of February 17, 2027. Wrong if: That date passes without the begin-to-render basis in force for standard display, whether because Google moved the date or shipped it as an optional reporting metric alongside the old count.

Google to Shift Ad Impression Counting to Begin-to-Render in 2027 Full Analysis → Read the source story →

Roku will not name any additional retail partner for Curate — the product that matches Roku's logged-in TV households to a retailer's purchase records — beyond Best Buy and Instacart before the 2027 upfront season closes on May 31, 2027.

Why Curate's public proof so far is one campaign: Reese's during March Madness with Instacart, reported at 30% more new buyers and more than $4 back per dollar spent, which is the easiest possible test case — a famous snack brand in the biggest sports window of the year. Retailers built their own media businesses precisely to keep purchase data and the margin that comes with it, so the only ones likely to hand that data to Roku are those with no ad sales machinery of their own, a shrinking pool once Best Buy and Instacart are taken. Signing one costs a retailer months of legal and procurement work and an unproven match rate between Roku's roughly 100 million logged-in households and its own loyalty file, against demand Roku has not yet shown it can deliver away from a tentpole event. The opposite outcome — a steady drip of new retail logos — is what a product launch normally produces, which is why calling zero additions is the real test of whether Curate is a platform or a pair of partnerships.

Right if: Right if, as of May 31, 2027, Roku's public announcements and product materials list no retail data partner for Curate other than Best Buy and Instacart. Wrong if: Wrong if Roku publicly names any additional retail data partner for Curate on or before May 31, 2027.

Roku Curate and Retail Partnerships Drive Shoppable CTV Results Full Analysis → Read the source story →

Before the end of 2027, OpenAI will announce that an independent ad-measurement company, either DoubleVerify or Integral Ad Science, verifies ads shown inside ChatGPT.

Why OpenAI is testing advertising in ChatGPT, and the two companies that dominate ad verification built their businesses checking that a published web page was safe and that a real person saw the ad, which gives them nothing to inspect when the page is a paragraph generated fresh for one user. Large brand budgets have never moved at scale into a surface that counts its own impressions and grades its own brand safety; Meta, Amazon and Netflix all ran closed and all ended up onboarding outside measurement once they wanted brand money rather than performance money. OpenAI can keep the auction, the placement and the margin entirely to itself and still hand verification to an outside vendor, which is the cheapest way to make its inventory buyable by agencies and compliance teams. The opposite outcome, OpenAI holding out past 2027 with no independent check on what its ads run beside, would mean leaving brand budgets on the table for years while it writes a placement standard from scratch. The bet is that OpenAI takes the shortcut every large closed platform eventually took.

Right if: OpenAI or the vendor publicly announces, on or before 31 December 2027, that DoubleVerify or Integral Ad Science measures or verifies advertising served inside ChatGPT. Wrong if: No such announcement from OpenAI, DoubleVerify or Integral Ad Science exists by 31 December 2027.

OpenAI Testing Ads; LLM Ad Models Risk Biasing Served Content Read the source story →

Advertisers will not be able to buy DAZN's live-sports inventory from start to finish through DAZN Media+'s AI sales agent, with no DAZN salesperson involved, in the product as publicly available on December 31, 2027.

Why DAZN Media+ has said it will ship an artificial-intelligence sales agent in 2027 that lets advertisers browse its sports inventory, build a proposal, buy, and track the campaign with no seller in the loop. Self-serve buying takes hold where inventory is deep and interchangeable, which is why it works at Amazon and Roku; live-sports rights are scarce and exclusive, and the deals carry exclusivity clauses, audience guarantees, and make-good terms that automated systems handle badly. Friction also protects the price, because a salesperson defends a premium while a menu invites comparison shopping that drags rates toward the floor. The version that actually ships is far more likely to cover always-on and leftover inventory while the marquee matches stay with human sellers. The opposite outcome requires DAZN to let its scarcest, highest-priced inventory be bought like a commodity within roughly fifteen months of announcing the idea.

Right if: On December 31, 2027, DAZN Media+'s publicly available agent does not let an advertiser select, price, and complete a purchase of DAZN live-sports inventory without a DAZN salesperson involved at some step. Wrong if: On December 31, 2027, DAZN Media+'s publicly available agent lets an advertiser select, price, and complete a purchase of DAZN live-sports inventory with no DAZN salesperson involved.

DAZN Media+ Unveils AI-Powered Ad Sales Agent, Launching 2027 Read the source story →

Publicis Groupe will win another media account billing at least $1 billion a year away from Omnicom or WPP, announced and reported by Ad Age on or before June 30, 2027.

Why PepsiCo has moved roughly $3.4 billion a year in media buying from Omnicom's OMD to Publicis, which is standing up a dedicated unit called One PepsiCo and selling a single profit-and-loss line that fuses Epsilon's first-party customer data, creative work, and measurement. Publicis paid $4.4 billion for Epsilon in 2019 and has spent six years pitching that package to large packaged-goods advertisers; closing the biggest available test case hands every rival marketing chief a reason to put its own incumbent agency through the same review. Publicis also walked away from the competing Coca-Cola review, a sign it is choosing which giants to chase from a full pipeline. A threshold of $1 billion is under a third of the PepsiCo move, so it does not require a repeat of the largest account shift in years. The case that PepsiCo was a one-off won on price is weaker because procurement teams at other mega-advertisers now have a public template to make incumbents match, and Omnicom is defending those relationships while absorbing the IPG merger.

Right if: Ad Age reports on or before June 30, 2027 that Publicis Groupe has won a media account with at least $1 billion in annual billings that was previously held by Omnicom or WPP. Wrong if: No such Publicis win of a $1 billion-plus former Omnicom or WPP media account is reported by Ad Age on or before June 30, 2027.

PepsiCo Moves $3.4B Ad Account from OMD to Publicis Full Analysis → Read the source story →

Tinuiti's Q4 2026 Digital Ads Benchmark Report, published by February 2027, will show US Google Search cost per click up less than 10% year over year.

Why Google's AI Overviews place an AI-written answer at the top of the results page, taking space that used to hold paid ads, and Go Fish Digital measured roughly a 20% jump in cost per click during August 2026. August is the most expensive back-to-school bidding month of the year, and AI Overviews mostly land on informational questions, while the bulk of advertiser money sits on high-purchase-intent queries where Google has kept its ad slots intact. Google has managed a decade of page redesigns without letting auction prices run away, because a sustained double-digit price increase on its largest performance advertisers is precisely what pushes them to fund tests in Meta and retail media. For the full-quarter, all-advertiser number to come in at 10% or higher, the squeeze would have to spread from the long tail into commercial queries and survive the post-holiday quarter, which is a much stronger claim than one agency's peak-season snapshot supports.

Right if: Tinuiti's Q4 2026 Digital Ads Benchmark Report shows US Google Search cost per click up less than 10% year over year, including any decline. Wrong if: That same report shows US Google Search cost per click up 10% or more year over year.

Google AI Overviews Shrink Search Ad Inventory, Raising CPCs Full Analysis → Read the source story →

The European Commission will publish a decision requiring Google to sell its ad exchange (AdX) or its publisher ad server (DFP) on or before 31 July 2027.

Why Brussels has had Google's ad-tech business under investigation since 2021, has already taken enforcement action over Google sitting on both the buying and selling side of the same auction, and has kept the option of a forced sale explicitly on the table. Behavioral promises are weak here because the self-preferencing the Commission objects to lives inside auction mechanics that regulators cannot audit in real time, which is exactly the trap the Commission fell into with its earlier shopping-comparison remedy that took years of follow-up litigation to settle. European publishers, who are the complainants driving the file, gain nothing from another pledge of fair behavior and are pushing for a structural cut, and a US court has already found Google illegally monopolized parts of this same stack, giving Brussels political cover to order the harder remedy. The likelier-sounding alternative, a negotiated behavioral settlement, requires the Commission to accept commitments from a company it has already fined for the same conduct, which is a harder sell inside DG Competition than ordering a sale and letting Google appeal. An order is not an executed divestiture: appeals will run for years, and this call turns on the Commission writing the order, not on Google completing a sale.

Right if: Right if the European Commission publishes a decision on or before 31 July 2027 requiring Google to divest its ad exchange or its publisher ad server, whether or not Google appeals or has begun any sale. Wrong if: Wrong if no such decision is published by 31 July 2027, including if the Commission accepts behavioral commitments, issues only a fine, or takes no further action.

European Commission ad-tech case seen as bigger threat to Google than U.S. ruling Full Analysis → Read the source story →

The Trade Desk's fourth-quarter 2026 revenue, reported in February 2027, will be lower than its fourth-quarter 2025 revenue.

Why The Trade Desk guided third-quarter 2026 revenue down about 12% to roughly $650 million, its first decline since going public in 2016, and its cash margin has fallen from 39% two years ago to 34% last year to about 25% now. The 575 jobs it cut to buy back five or six points of that margin sat mostly in client success and solutions engineering, the people who walk buyers through migrating to Kokai, the company's newer automated buying interface, and Kokai adoption is the thing management is counting on to reaccelerate the top line. Meanwhile holdco procurement teams are using the down quarter as leverage to reopen volume commitments and run live tests on Amazon's DSP and Google's DV360, so any dollars that move out are slow to come back. A fourth-quarter recovery requires both an ad-market rebound and smooth migrations with fewer hands to run them, and the second cuts against the first. Q4 is the largest quarter of the year, which makes it the hardest one to backfill if commitments were renegotiated down in the autumn.

Right if: The Trade Desk's reported fourth-quarter 2026 revenue is below its reported fourth-quarter 2025 revenue. Wrong if: The Trade Desk's reported fourth-quarter 2026 revenue is equal to or above its reported fourth-quarter 2025 revenue.

Trade Desk Q3 Revenue Guidance Marks First Decline Since 2016 IPO Read the source story →

The final remedy order Judge Leonie Brinkema enters against Google in the US ad-tech antitrust case will not give an independent, court-appointed monitor direct access to Google's ad auction data, per the order as entered on the docket in United States v. Google in the Eastern District of Virginia by June 30, 2027.

Why Brinkema found Google monopolized the ad-serving pipes and then refused to break anything up, partly to avoid a 14-to-15-year appeals fight, and the fixes she signalled (sharing bid data with rival ad servers, ending the rules that stopped publishers setting different price floors for different exchanges) sit closer to Google's own proposed remedy than to the Justice Department's. Those fixes only bite if someone outside Google can check the auction, and an outside monitor with real log access is exactly the kind of intrusive machinery that takes years to stand up and invites the appeal the judge was trying to dodge. Google already tells the court it retired practices like first look and last look years ago, the Justice Department disputes that account, and nobody outside Google can settle it without the logs. The contrary precedent exists, since the search-case remedy installed a technical committee, which is why this is a live question rather than a formality, but adopting that here would reverse the same caution that produced the no-breakup ruling.

Right if: The remedy order entered on the docket by June 30, 2027 leaves verification of Google's compliance to Google's own reporting to the court or the Justice Department, with no independent monitor or technical committee granted direct access to Google's auction data. Wrong if: The entered order grants an independent monitor, auditor, or technical committee direct access to Google's auction data, or no final remedy order has been entered on the docket by June 30, 2027.

Google Had Its Day In Court. Now, It's Amazon's Turn Full Analysis → Listen to the episode →

By December 31, 2027, one of Amazon, Google, Meta, or Microsoft will publicly cancel or move a previously announced US data center campus costing $1 billion or more after the host county, parish, or town government blocked it.

Why Denver, New Orleans, and Kansas City have all imposed pauses on new data centers, and Quinton Lucas, president of the Democratic Mayors Association, expects the fight to intensify through the 2026 election cycle. The pressure point is physical: Denver Mayor Mike Johnston puts each local facility at roughly a quarter-million gallons of water a day plus heavy draw on the local grid, and the same load lands harder on rural systems hosting the giant campuses, such as the roughly $10 billion project in Richland Parish, Louisiana, whose incentive terms sit behind nondisclosure agreements. Local boards have already shown they will vote no, and the four largest builders have so many announced sites in the queue that only one holdout board needs to survive the lobbying for a campus to be abandoned or shifted to another state. The quiet outcome requires every host community facing a billion-dollar site to keep approving deals it cannot read, at a moment when utility bills are rising and the issue cuts across party lines. Working against the call: rural counties want the tax base and construction jobs, and the money usually buys agreement before organized opposition forms.

Right if: On or before December 31, 2027, Amazon, Google, Meta, or Microsoft announces it is cancelling or relocating a previously announced US data center campus of $1 billion or more, following a denial, revocation, or restriction vote recorded in the host local government's public minutes. Wrong if: Through December 31, 2027, every announced US campus of $1 billion or more from those four companies either proceeds, is merely delayed, or is dropped without any host local government having voted to block it.

Three Mayors of Blue Cities in a Polarized America Listen to the episode →

Alphabet's search advertising revenue will grow at least 15% year-over-year in the third quarter of 2026, reported in late October 2026.

Why Alphabet's search advertising line grew roughly 17% year-over-year in the second quarter of 2026, in the same period that clicks sent from search results to outside websites fell by about 30% on average across measured sites. Those clicks were always a cost Google absorbed, never a line it earned from, so answering the query on the results page keeps the user, keeps the auction, and removes the hand-off. Google has also raised minimum spend thresholds on advertisers, which lifts revenue per query independent of volume. The case for a sharp slowdown rests on ChatGPT or Perplexity pulling away enough query volume to dent the auction within a single quarter, and with AI Overviews reaching roughly two billion users and Google's default placement in Chrome and Android intact, that shift is too slow to land by September 2026.

Right if: Alphabet's reported Q3 2026 search advertising revenue is 15% or more above the same quarter of 2025. Wrong if: Alphabet's reported Q3 2026 search advertising revenue is less than 15% above the same quarter of 2025.

The Homepage Is No Longer the Front Door: Leah Nurik on AI Visibility, GEO, and the Future of Brand Discovery Full Analysis → Listen to the episode →

Before April 30, 2027, one of the 20 largest US advertisers will publicly say it has cut or frozen spending with Google or Meta because the platforms' estimated iPhone conversion figures overstate results.

Why Apple's tracking rules stop Google and Meta from seeing most purchases made on iPhones, so both estimate those sales statistically and report the estimates alongside real ones with no margin of error attached. Automated budget tools such as Meta's Advantage+ and Google's DV360 bidding steer money toward whichever channel reports the cheapest cost per sale, which can move up to a quarter of a large advertiser's iPhone budget toward the most flattering scoreboard within about 90 days, with no buyer left in the loop to discount it. A single geographic holdout test exposes the gap, and once finance sees budget compounding into sales that may not exist, the cheapest source of leverage a large advertiser has is saying so in public. The alternative is that big brands keep absorbing the gap quietly, as they have since 2021, but the automation is new and it removes the manual haircut that used to keep the problem inside one buyer's spreadsheet.

Right if: A company on Ad Age's most recent list of the 20 largest US advertisers states on the record, in a public filing, press interview, or conference appearance before April 30, 2027, that it has cut or frozen Google or Meta spending over unverifiable modeled conversion reporting. Wrong if: No company on that list makes such a statement on the record before April 30, 2027.

Google and Meta Use Fake "Modeled Conversions" to Steal Ad Attribution on iOS Read the source story →

On December 31, 2027, Google will not be operating under any in-force court order in the US ad-tech antitrust case requiring it to separate its publisher ad server from its ad exchange or to open its auction data to rival exchanges.

Why Judge Leonie Brinkema of the US District Court for the Eastern District of Virginia has refused the Justice Department's bid to force a sale of Google's ad exchange and publisher ad server, leaving only conduct rules on the table in a case about a business worth roughly 12% of Alphabet's revenue. Conduct rules still have to be written, and any version that bites at the auction layer will be appealed by Google and is routinely stayed while that appeal runs, which pushes real compliance years past the order date. Publishers and independent exchanges only gain leverage on the day a rule is actually binding, not the day it is drafted. The opposite outcome requires the same judge to write unbundling or data-access terms quickly and an appeals court to let them take effect immediately, a sequence that has not happened to Google in any prior ad-tech or search matter.

Right if: On December 31, 2027, the docket in the US ad-tech antitrust case against Google shows no order in effect that obliges Google to separate its publisher ad server from its ad exchange or to give rival exchanges access to its auction data, whether because no such order was issued or because it is stayed or under appeal. Wrong if: On December 31, 2027, such an order is in effect and Google is required to comply with it on that date.

MadTech Daily: DOJ Loses Bid to Break Up Google’s Ad-Tech Business; Havas Names Patrick Affleck to Lead UK & Ireland Village Listen to the episode →

Mediaocean will publicly announce an agreement to acquire a measurement, attribution, or ad-verification company on or before December 31, 2027.

Why Iván Markman, Mediaocean's president and chief operating officer, has staked the company's future on owning the reconciled record of how ad money and impressions actually moved — the same spend data that independent measurement firms currently rely on advertisers to hand over. Mediaocean has bought its way into every adjacent layer it wanted before, taking in Flashtalking, Innovid and 4C, and those ad servers are still being combined with customers mid-migration. Owning the billing record only turns into a real advantage if the company can also grade outcomes, and buying that capability is far quicker than building it while two serving stacks are still being merged. Sitting still until the migration finishes would leave the outcome-verification layer to DoubleVerify and Integral Ad Science at exactly the point when automated buying tools need verified results wired into the money flow, which is the position Mediaocean says it intends to own.

Right if: Mediaocean publicly announces, on or before December 31, 2027, an agreement to buy a company whose primary business is measurement, attribution, or ad verification. Wrong if: Mediaocean announces no such acquisition on or before December 31, 2027.

A Trillion Impressions and Two Hundred Billion Dollars Full Analysis → Listen to the episode →

To resolve the state attorneys general lawsuit over its purchase of Warner Bros. Discovery, Paramount Skydance will publicly agree by December 31, 2027 to a condition that limits how the combined company sells advertising, such as selling off ad sales assets or committing to keep its inventory open to competing ad exchanges.

Why California Attorney General Rob Bonta canceled settlement talks and accused Paramount Skydance of bad faith, and the 12-state coalition suing to block the deal has already won early rulings; the harm they have pleaded is that one combined seller of broadcast and streaming inventory gains pricing power over advertisers. Federal antitrust review under the current administration is permissive, so the states are the only real gate, and a gate that cannot deliver an outright block usually extracts conditions instead. Because the states' own case is built on advertising, advertising is the natural currency of any deal that lets the merger proceed. Equity markets price the acquisition above a 60% chance of closing, which means almost nobody is paid for the outcome where it closes carrying advertising strings. For the deal to close clean, the states would have to walk away from the exact harm they went to court over, after publicly burning a settlement meeting to signal they will not.

Right if: On or before December 31, 2027, a settlement, consent decree, court order or Paramount Skydance announcement filed in or arising from the state attorneys general case imposes a condition on how the combined company sells advertising. Wrong if: No such advertising-related condition is announced or filed by December 31, 2027, including cases where the litigation is still pending, the suit is dropped, the deal closes unconditioned, or the deal is abandoned.

California Cancels Paramount Settlement Talks, Calls Out Bad Faith Read the source story →

Google will agree to a child-safety settlement with a coalition of US state attorneys general covering YouTube's teen accounts, announced publicly or filed in court on or before December 31, 2027.

Why Meta has agreed to pay about $18 billion over ten years to 48 states, roughly $1.2 billion a year against annual revenue well north of $160 billion, and to turn on default daily time limits and nighttime blocks for teen accounts. That deal hands every attorney general a finished template with priced remedies, and Meta is openly pushing to have YouTube and TikTok held to the same floor, which means a well-funded party is feeding the effort. Google's exposure here is lopsided: teen viewing is a small slice of YouTube's business, while a multistate discovery process reaches internal research the company would rather keep sealed, and it has settled youth-privacy claims quickly before rather than litigate them. TikTok, by contrast, is tangled in ownership fights that make a clean multistate settlement harder to sign, which is why Google is the likelier first non-signatory to fold. The case for the opposite outcome rests on Google's willingness to spend years in court over an inventory pool it can afford to lose, which is the worse trade.

Right if: On or before December 31, 2027, Google announces, or files in court, a child-safety settlement with a group of at least ten US state attorneys general covering YouTube's handling of accounts held by minors. Wrong if: No such settlement is announced or filed by December 31, 2027, including cases where states only file or advance lawsuits against Google without a resolution.

Meta Agrees to $18bn Child Safety Settlement Across 48 States Read the source story →

The Trade Desk's Q3 2026 revenue growth, reported in early November 2026, will come in below the 3% year-over-year growth it posted in Q2 2026.

Why The Trade Desk grew just 3% year-over-year in Q2 2026, with chief executive Jeff Green attributing the weakness to automotive and consumer-goods advertisers that together make up roughly a quarter of the business, categories that rarely turn within a single quarter. The company then cut 575 roles, about 15% of staff, with the reductions landing largely inside Q3 itself, which strips client-facing coverage at exactly the point in the calendar where advertisers decide where incremental budget goes. Customer retention has held above 95% for more than a decade, so the problem is not clients leaving but clients spending less on the platform while Google and Amazon absorb the difference. For growth to hold at 3% or better, the soft categories would need to stabilize in the same quarter the account teams serving them were being dismantled. The cost side is trivial at $39 million to $51 million in charges against $1.5 billion in cash; the revenue side is where the damage shows up first.

Right if: The Trade Desk's reported Q3 2026 revenue growth is below 3.0% year-over-year. Wrong if: The Trade Desk's reported Q3 2026 revenue growth is 3.0% year-over-year or higher.

The Trade Desk cuts 15% of staff after growth collapses to 3%

Snap will agree to settle the US children's social-media addiction claims against it before August 1, 2027.

Why Meta has reportedly agreed to pay roughly $18 billion to close the children's social-media addiction claims against it, about 11% of Meta's $164 billion in 2024 revenue and almost certainly spread over a multi-year payment schedule, which Meta can absorb. Snap is a co-defendant in the same consolidated litigation and earns a small fraction of Meta's revenue, so a number scaled anywhere near that benchmark would threaten the company rather than dent a quarter. A headline settlement gives plaintiffs' lawyers a public price to anchor to, and the usual response from thinner-balance-sheet defendants is to buy certainty early rather than let a jury set the figure. Fighting to verdict is the alternative, and it means years of legal spend with a teen-heavy audience business hostage to one trial outcome, which is why the deep-pocket defendant settling first tends to pull the smaller ones in behind it.

Right if: We're right if filings in the federal Social Media Adolescent Addiction Product Liability multidistrict litigation (MDL No. 3047, Northern District of California) show Snap has entered a settlement of children's addiction claims before August 1, 2027. Wrong if: We're wrong if no Snap settlement of those claims appears on that docket by August 1, 2027.

Meta settles children's social media addiction lawsuits for $18 billion Read the source story →

Anthropic will raise the published list price of an already-released Claude model on its public pricing page at some point before 30 June 2027.

Why Anthropic has reportedly committed around $35 billion to compute capacity in a multi-year deal with Lambda, and that bill has to be earned back through what developers pay per call. Today's per-token rates and free developer credits are funded by investor capital rather than operating margin, and Andrew Bailey, Governor of the Bank of England, has publicly warned that a stumble in AI returns would hit the small group of names holding up the market and the borrowed money behind them. Every pricing move in this market so far has gone the other way, because each model generation was cheaper to serve and each lab cut list prices to hold developer share, so a rise means one lab accepting some volume loss to protect margin. Anthropic is the most likely to do it: its revenue skews to enterprise customers who have already wired Claude into production workflows and cannot swap it out in a quarter, and quietly repricing an older model costs less share than launching a new one at a higher rate. The alternative, holding or cutting list prices for another three quarters while servicing a $35 billion commitment, requires the capital window to stay open the whole way.

Right if: Anthropic's public API pricing page shows a higher price per million input or output tokens, at any point before 30 June 2027, for any Claude model that was already generally available on 20 September 2026. Wrong if: No model generally available on 20 September 2026 carries a higher published per-token price on Anthropic's pricing page at any point before 30 June 2027.

MadTech Daily: Bailey Flags AI Slowdown Risk for Global Markets; OpenAI Rejects Apple’s Trade Secret Theft Claims Listen to the episode →

Before June 30, 2027, a US federal agency or a state attorney general will sue a retail media network other than Amazon over how it sets or describes the minimum prices in its ad auctions.

Why The Federal Trade Commission and more than 20 state attorneys general have accused Amazon of quietly raising the minimum prices it charges advertisers on its retail ad auctions while telling them something different, which is a misrepresentation case rather than a pricing case. Every large retail media network runs the same setup: it owns the inventory, runs the auction, sets an invisible minimum, and publishes marketing claims about how fair that auction is. Once one enforcement theory is drafted and survives a first round of motions, re-filing it against Walmart Connect, Instacart, or a grocery network costs a state attorney general very little, and these offices reliably copy each other on consumer-protection theories that produce headlines. The case for nothing happening is that regulators wait for the lead case to develop before opening a second front, but the state coalition here is already large, elected AGs are rewarded for moving first, and retail media is now big enough that advertiser complaints are landing in those offices already.

Right if: A US federal agency or a state attorney general files a lawsuit or formal enforcement action before June 30, 2027 against a retail media network other than Amazon alleging it misstated or concealed how it sets minimum ad prices or auction pricing. Wrong if: No such lawsuit or formal enforcement action against a retail media network other than Amazon is filed by June 30, 2027.

MadTech Daily: FTC Sues Amazon Over Alleged Ad-Price Manipulation; Brussels Puts ChatGPT in the Same Bracket as Google Listen to the episode →

Hummingbirds, the startup that pays shoppers cash back for photographing store receipts to prove small creators drove in-store sales, will either be acquired or stop selling receipt-scan in-store attribution as its main product by June 30, 2027.

Why Hummingbirds runs about 70,000 small creators for roughly 350 paying brands, funding the loop with a software subscription plus consumer cash-back rebates, which leaves thin margin at challenger-brand scale. CEO Emily Steele has said the company will not work with publicly traded consumer-goods brands because their legal teams demand creative control, which kills the unscripted content the model depends on, so the largest packaged-goods budgets are structurally off the table. The measurement itself leaks: a shopper has to save an offer, buy the item in a physical store, then remember to photograph the receipt for a few dollars back. Retail media networks and measurement firms want offline signal and buy the plumbing rather than build it, which makes a tuck-in acquisition the likeliest exit for a company with this ceiling. Staying independent and still leading with receipt-scan attribution in mid-2027 would mean the model found budget inside exactly the risk-averse public advertisers its own founder rules out.

Right if: By June 30, 2027, Hummingbirds has announced an acquisition or a sale of its business, has shut down, or no longer presents receipt-scan in-store purchase attribution as its primary product in its own public marketing. Wrong if: On June 30, 2027, Hummingbirds is still an independent company presenting receipt-scan in-store purchase attribution as its primary product in its own public marketing.

Cracking The In-Store Attribution Code In Influencer Marketing Full Analysis → Listen to the episode →

Magnite or PubMatic will announce a deal to buy a company that identifies or measures which television shows streaming ads run against, on or before June 30, 2027.

Why FreeWheel, Comcast's platform for selling ads in professional streaming and broadcast content, began publishing show-by-show reports in July 2026 telling advertisers exactly which series their ads ran against, and brand-safety teams at the big agencies have every reason to write that question into 2027 buying requirements because it lets them build exclusion lists and defend the prices they already pay. Magnite and PubMatic route other people's inventory and have no content relationships of their own, so they cannot vouch for what a show is; building that capability in-house is a twelve-to-eighteen-month job, and the buying season that decides 2027 budgets closes well before then. Buying a content-classification or video-measurement business is the only route that fits the calendar, and such assets are small compared with the premium streaming revenue at stake. The alternative is that both firms stitch together publisher data partnerships instead, but partnerships leave them dependent on the same content owners they compete with for the ad dollar, which is a weak answer to give a holding company asking where its money went.

Right if: Either Magnite or PubMatic publicly announces, by June 30, 2027, an agreement to acquire a business whose product classifies or measures the television and streaming content that ads run against. Wrong if: Neither Magnite nor PubMatic announces such an acquisition by June 30, 2027.

FreeWheel Launches Series-Level CTV Transparency Report July 2026 Full Analysis → Read the source story →

By June 30, 2027, at least one of the six largest agency holding companies will publicly launch an AI agent that shifts live client budgets between Google, Meta, and Amazon without a person approving each shift.

Why Anthropic open-sourced the Model Context Protocol, the wiring that lets AI assistants operate other software, and handed it to the Linux Foundation; Google, Meta, Amazon and most major buying platforms have since stood up servers for it, so the cost of connecting an agent to a buying account has collapsed to near zero. That plumbing is already in place for reading reports, and the only savings that matter, engineering time on custom connectors and mid-level execution headcount, arrive solely when the human confirmation click disappears. Holding companies are under continuous client pressure to cut fees and are each marketing an in-house AI platform, which gives the first mover a commercial reason to ship the unsupervised version on managed budgets and say so loudly. The opposite outcome requires all six to keep a person signing every budget change for another year while selling AI as the reason their fees are worth paying, which is a harder position to hold than the liability argument suggests.

Right if: By June 30, 2027, WPP, Publicis, Omnicom, Dentsu, Havas, or Stagwell announces in its own press materials an agent available to clients that reallocates live campaign budgets across Google, Meta, and Amazon with no per-change human approval. Wrong if: On June 30, 2027, every such offering from those six companies still requires a person to approve budget changes or operates inside a single platform only.

MCP servers enabling AI agents to autonomously manage ad campaigns Read the source story →

By June 30, 2027, The Trade Desk will publicly grade or rank the exchanges it buys from on whether they pass a consistent transaction ID — the shared code that lets a buyer see when the same ad slot is being offered to it by several exchanges at once.

Why In August 2025 a change pushed into Prebid, the open-source auction software most publishers run, broke that shared code across exchanges, and the IAB Tech Lab called it a violation of the industry's auction rulebook; months later nothing had been rolled back and the matter closed with a clarification. Standards bodies here have no penalty to impose, so the only party that can make the identifier matter is a buyer that pays less for supply where it is missing or duplicated. The Trade Desk has both the motive and the machinery: duplicate offers inflate what it pays for the same impression, and it already publishes supply-path quality guidance to steer spend toward direct routes. The alternative — The Trade Desk raising the issue loudly through its own executives and then declining to price it — would leave money on the table for the middle-tier resellers whose margins depend on the duplication going unseen.

Right if: On or before June 30, 2027, The Trade Desk publishes supply-path or marketplace-quality materials, on its own site or in documentation reported publicly, that score exchanges on transaction ID consistency or duplicate-bid detection. Wrong if: No such score from The Trade Desk is published by June 30, 2027.

IAB Tech Lab Declared Transaction ID Change an OpenRTB Spec Violation Read the source story →

By December 31, 2027, at least one of the independent creator-management platforms Grin, Aspire, or Creator.co will be acquired, merged into another company, or shut down.

Why Unilever is now working with roughly 300,000 creators and L'Oreal with roughly 500,000, a roster size you only reach by automating discovery, contracting and payment, which is precisely the job these mid-tier platforms sell. Buyers operating at that scale want direct technical access to the social platforms' own systems, and YouTube, TikTok and Instagram want that spending settled inside their own walls, so the company in the middle is compressed from both ends at once. The same compression thinned out the middle of the programmatic display market once the largest buyers and largest sellers both wanted fewer hops. These vendors are venture-funded, sell into a category where the two ends of the market are both building the same features, and have no proprietary audience data or brand-safety scoring to fall back on. For all three to remain independent going concerns through 2027, brands and platforms would both have to keep paying for a layer each of them can build cheaply in-house.

Right if: By December 31, 2027, Grin, Aspire, or Creator.co has announced, or had announced by an acquirer, that it has been bought, merged into another company, or closed its creator-management business. Wrong if: On December 31, 2027, all three of Grin, Aspire, and Creator.co are still independently owned and still selling creator management as their main product.

Influencer marketing scale raises programmatic-style dilution concerns Read the source story →

By 15 May 2027, at least one of The Trade Desk, PubMatic, Magnite, or Viant will announce a signed agreement to be acquired or taken private.

Why In one summer, Publicis bought LiveRamp, Nielsen bought DoubleVerify, and Criteo entered talks to go private with Vista Equity Partners, following Integral Ad Science off the public market the year before, all at prices far below these companies' 2021 peaks. Each low-premium deal sets a cheaper reference price for the ones still listed, which makes the next buyer's bid easier to justify and makes shareholders more willing to take the exit. The Trade Desk and AppLovin have kept growing while their shares fell, which shows buyers are repricing the whole category of independent, publicly traded ad tech rather than punishing weak businesses, so health is no defence against a bid. For none of these four to sign a deal in the next eight months, ad-tech share prices would have to recover enough that boards can argue value is better realised by staying public, and nothing in this year's deal flow points that way.

Right if: Any one of The Trade Desk, PubMatic, Magnite, or Viant publicly announces a signed agreement to be acquired or taken private on or before 15 May 2027. Wrong if: None of those four companies has announced such an agreement by 15 May 2027.

Independent ad tech consolidation wave accelerates in summer 2026 Full Analysis → Read the source story →

Wunderkind's multi-app pause ad buy, which today covers Plex, Philo and DISH, will not carry pause inventory from Roku, Amazon's Prime Video or Disney's streaming apps as of May 31, 2027.

Why A pause ad runs on a screen the streaming app itself froze, so it costs the platform almost nothing to serve and can be priced above a standard break; TVision's study of 40 advertisers found pause ads drew about twice the attention of a 60-second in-stream spot. Roku, Amazon and Disney all already sell pause ads on their own apps and each runs its own ad sales stack, which means routing that inventory into an outside seller's single cross-app buy hands away margin on inventory they fully control. Wunderkind's three current supply partners — Plex, Philo and DISH — are exactly the kind of mid-sized services that need someone else's demand to fill a new format, and that is the profile of the platforms likely to keep joining. The opposite outcome requires one of the three largest streaming ad sellers to decide it cannot sell its own highest-attention slot without help, at the moment the performance case for charging a premium has just been handed to it.

Right if: On May 31, 2027, the platforms Wunderkind publicly lists for its pause ad buy include none of Roku, Amazon's Prime Video or any Disney-owned streaming app. Wrong if: On May 31, 2027, Wunderkind publicly lists at least one of Roku, Amazon's Prime Video or a Disney-owned streaming app as available through that pause ad buy.

Pause Ads Deliver 2x Attention vs. Standard 60-Second Streaming Spots Full Analysis → Read the source story →

As of the close of the IAB's NewFronts week in May 2027, none of YouTube, Instagram, or TikTok will give advertisers creator campaign results in a shared format that lets the same creator metrics be compared side by side across all three platforms.

Why The IAB has assembled a board of creator-economy executives to write shared terminology, brand-safety rules, and measurement definitions so large brands can buy creator media the way they buy other media, and standards bodies like this reliably produce the document. The three platforms that actually own creator distribution are not bound by it, and their native buying tools are the product they sell: if a buyer could grade a YouTube creator against a TikTok creator on one ruler, the platforms' own numbers stop being the only truth and creator inventory starts being priced against itself. Agreeing on what a word means costs a platform nothing; handing over the underlying performance signal costs it pricing power, which is why endorsement of industry standards has historically arrived without data access. For the opposite to happen, one of the three would have to volunteer the comparison that makes its rivals' inventory substitutable for its own, and none has done that under any prior IAB standard. Verification vendors can sell brand-safety checks on creator content without any of this, so no outside party has the leverage to force it open.

Right if: At the end of IAB NewFronts week in May 2027, no one of YouTube, Instagram, or TikTok makes creator-level campaign results available to advertisers in a format comparable across all three platforms. Wrong if: By the end of IAB NewFronts week in May 2027, at least one of YouTube, Instagram, or TikTok has publicly launched or committed to creator-level reporting that advertisers can compare across all three platforms.

IAB Launches Creator Board to Scale Influencer Advertising Standards Read the source story →

By June 30, 2027, at least one of the 25 largest US advertisers in Ad Age's Leading National Advertisers ranking will publicly confirm it is buying ads inside ChatGPT.

Why OpenAI says its ChatGPT ad business is running at a $1 billion annual rate less than 200 days after launch, and it has opened self-serve buying across 31 European markets, but its stated $2.5 billion target for 2026 needs roughly five times the current monthly revenue and that gap cannot be closed on small advertisers alone. Every new ad platform closes that gap the same way Meta and TikTok did, by paying for and publicising flagship campaigns from household-name brands, because one named blue-chip buyer unlocks the procurement conversations behind hundreds of smaller ones. The counter-case is real: ChatGPT has no independent brand safety or viewability certification, and a conversational placement does not reconcile against the attribution models big marketers already run, so the largest brands can test quietly and say nothing. But quiet testing is worth little to OpenAI, and a platform growing this fast has both the cash and the motive to buy the endorsement it needs before the 2027 planning cycle closes.

Right if: On or before June 30, 2027, a company ranked in the top 25 of Ad Age's Leading National Advertisers list is publicly confirmed, by its own executives or its own communications, to be running paid advertising inside ChatGPT. Wrong if: June 30, 2027 passes with no top-25 Ad Age Leading National Advertiser having publicly confirmed it is running paid advertising inside ChatGPT.

OpenAI's ChatGPT Ads Business Hits $1 Billion Annualized Run Rate Read the source story →

ShopSense AI, the startup that matches products in retailer catalogs to shoppable moments in TV and video content, will publicly announce an acquisition or a shutdown on or before June 30, 2027.

Why ShopSense AI sells a narrow model that reads content and returns relevant products across a self-reported 1,500 retailers and 500,000 brands, and that matching job is the exact capability OpenAI and Google are folding into their consumer assistants alongside shopping and checkout. Once the general models do the matching acceptably, the part of the business that still has value is the unglamorous work of cleaning and connecting messy product feeds so a 'shop this' button resolves to something actually in stock, and that asset is cheaper to buy than to build. Retail media and measurement buyers are already operating on the assumption that the independents get bought, which sets a low bar for a deal to happen. The alternative path, staying independent, requires raising more money on a model advantage that shrinks with every frontier release, and the shoppable-TV revenue underneath it is still tiny.

Right if: ShopSense AI, or a buyer, publicly announces an acquisition, a sale of the business, or a wind-down of the company on or before June 30, 2027. Wrong if: ShopSense AI is still operating as an independent company with no announced acquisition or wind-down as of July 1, 2027.

Ep 149: From Amazon to AI-Powered Commerce with Marissa Ramirez of ShopSense AI Full Analysis → Listen to the episode →

PubMatic will announce an agreement to be acquired or taken private on or before August 15, 2027.

Why Four independent ad-tech companies were claimed in a single summer: Publicis agreed to buy LiveRamp, Nielsen agreed to buy DoubleVerify, Integral Ad Science had already gone private, and Criteo entered talks with Vista Equity Partners. Public investors have stopped paying a growth multiple for standalone advertising software even when revenue rises, so a private-equity buyer or a strategic owner now values these assets more highly than the stock market does. PubMatic is the most exposed of the remaining independents: it is the smallest of the listed supply-side platforms, its share price has been punished through the same cycle, and its function is exactly the kind of plumbing an agency group or a measurement firm would rather own than rent. For this to fail, public multiples for standalone ad-tech would have to recover enough to make staying listed attractive again, and nothing in the past year points that way.

Right if: PubMatic publicly announces a definitive agreement to be acquired or taken private, in its own press release or an SEC filing, dated on or before August 15, 2027. Wrong if: No such PubMatic announcement or filing exists as of August 15, 2027.

5 questions facing advertisers, publishers and creators heading into fall Full Analysis → Listen to the episode →

On June 30, 2027, Roku will no longer sell Curate as a separately named product bundling retail purchase data with its TV ad inventory, having shut it down, renamed it, or folded it into general Roku ad buying.

Why Roku launched Curate in April 2026, stitching its viewing data on Roku-powered televisions to shopper purchase signals licensed from Best Buy Ads, Criteo, Instacart and Kroger, and pitched it as one line item covering inventory, shopper data and measurement. Every one of those four partners keeps selling the same signals directly, so any buyer who suspects a convenience markup can assemble the identical basket and negotiate each piece down. Amazon runs the same ad-to-purchase loop on Fire TV using shopper data it owns outright, with no licence fees and no partner able to pull the plug. Bundles hold together when the seller controls something no one else can supply, and the only genuinely exclusive asset here is viewing behaviour on Roku devices, an audience that skews older and budget-conscious. The alternative, Curate hardening into a standard that buyers stop pricing component by component, needs Roku to out-credential a rival that owns the actual receipts.

Right if: Roku's advertising product listings on June 30, 2027 no longer offer Curate by name as a distinct retail-purchase-data bundle, whether because it was discontinued, renamed, or absorbed into Roku's general ad products. Wrong if: Roku is still marketing Curate under that name as a distinct, separately buyable retail-data bundle on June 30, 2027.

Roku Pitches Data Quality Over Quantity to Win CTV Performance Ad Dollars Full Analysis → Read the source story →

By June 30, 2027, Publicis will be publicly announced as the new lead media agency for at least one more advertiser on Ad Age's ranking of the world's 25 largest advertisers, taking that business away from Omnicom.

Why PepsiCo handed Publicis its global media without running a competitive pitch, ending a relationship Omnicom's OMD had held in the US and UK for more than twenty years, and a no-pitch handover only happens when the buyer has already settled on the criterion it cares about. Here that criterion was an agency stack the advertiser can plug its own customer data into once, rather than the usual round of buying-power and rate-card negotiation. Every other large consumer-goods procurement team has now seen that shortcut, and Omnicom is the most exposed incumbent because it is simultaneously absorbing Interpublic, which pulls attention and staff away from defending existing accounts. The alternative reading, that Publicis simply undercut on fees and this stays a one-off, is weaker because a client does not rebuild twenty years of audience data, platform seats, and taxonomy to save a few points of commission.

Right if: On or before June 30, 2027, Ad Age reports that an advertiser on its list of the world's 25 largest advertisers has appointed Publicis as lead media agency in place of Omnicom. Wrong if: No such appointment is reported by Ad Age on or before June 30, 2027.

Publicis Wins PepsiCo Global Media Without a Pitch Full Analysis → Read the source story →

Permutive, the privately held publisher data company, will announce it has been acquired by May 15, 2028.

Why Permutive co-founder Joe Root has put a number on the addressability squeeze: fewer than 30% of consumers now carry a persistent ID, the tracking token buyers use to aim an ad at a specific person, which means most ad space on the open web cannot be matched to a named individual. That math hurts the companies whose core business is matching names to impressions, and the cheapest way for an identity, cloud, measurement, or agency buyer to own the publisher-side data layer is to buy a company already embedded in publishers' pages rather than build one. Permutive sits in exactly that layer and is private, which makes it purchasable on a normal corporate timetable. Staying independent is the harder path: a publisher-focused data vendor has few routes to scale against buyers who can bundle the same capability with cloud storage, identity resolution, or verification they already sell, and the funding market for identity-adjacent tooling has thinned. The main way this call fails is if probabilistic and IP-based matching quietly restores enough reach that the scarcity thesis cools and nobody feels urgency to buy.

Right if: Permutive or an acquiring company publicly announces a change of control of Permutive on or before May 15, 2028. Wrong if: No such acquisition announcement exists as of May 15, 2028, including if Permutive is still operating independently, has raised new funding, or has wound down.

Only 30% of Consumers Are Addressable, Reshaping Ad Targeting Full Analysis → Read the source story →

The Trade Desk will close free public access to OpenSincera, the free version of Sincera's publisher quality scores, requiring a Trade Desk account or a paid licence to pull the data, by December 31, 2027.

Why The Trade Desk, the largest independent buyer of open-web advertising, bought Sincera, the company that grades how clean and transparent a publisher's bid requests are, and its founder Mike O'Sullivan now reports directly to chief executive Jeff Green. A slice of that grading is published free as OpenSincera, which is how a private rubric becomes the industry's default definition of clean supply: agencies wire the scores into buying rules and junk-site exclusion lists, and publishers fix their sites to chase a better grade. Once that dependence is set, the free tier stops being a land-grab and becomes a standing gift to anyone who wants to reverse-engineer the rubric and build a rival grader, while the buyer that owns the scale has every reason to keep the dial inside the building. Leaving the data permanently open to all comers is the less likely path, because it hands competitors the blueprint for the exact asset The Trade Desk paid for and does nothing to settle the conflict of a buyer grading its own sellers.

Right if: On December 31, 2027, pulling OpenSincera publisher scores requires a Trade Desk account, a paid licence, or an approval process, or the public feed has been discontinued. Wrong if: On December 31, 2027, anyone with no Trade Desk relationship can still access OpenSincera publisher scores free of charge.

The Trade Desk's Sincera Now Controls Open Supply Auditing Full Analysis → Read the source story →

Before June 30, 2027, at least one of the 20 largest US consumer-goods advertisers will publicly cut or freeze its advertising spending with Amazon and give the way Amazon sets its auction prices as a reason.

Why The Federal Trade Commission and more than 20 state attorneys general have sued Amazon, alleging it quietly raised the minimum prices in its own ad auctions and misdescribed how those auctions worked, collecting billions from advertisers in the process. Big packaged-goods buyers have complained about Amazon's pricing opacity since its search ads launched and have never pulled back, because being able to tie an ad to a purchase inside Amazon's own store still pays better than the alternatives. What the lawsuit changes is cover: a brand's performance lead who wants concessions can now point at a government filing rather than a suspicion, and the only way to make that threat cost Amazon anything is to make the pullback visible. Squeezing quietly for rebates buys far less leverage than one publicized freeze during the biggest spending year of the cycle, which is why at least one large advertiser taking the loud route beats the whole category staying polite.

Right if: By June 30, 2027, an advertiser among the 20 largest US consumer-goods ad spenders has stated publicly, in its own announcement, earnings call, or an on-record statement to the press, that it is cutting or freezing spend with Amazon over how Amazon prices its ad auctions. Wrong if: By June 30, 2027, no advertiser among the 20 largest US consumer-goods ad spenders has publicly tied a cut or freeze in Amazon spending to Amazon's auction pricing.

FTC and 20+ States Sue Amazon Over Ad-Price Manipulation Read the source story →

Integral Ad Science will announce an agreement to be acquired or taken private on or before 31 December 2027.

Why Ad verification is being pulled inside the companies that sell the media: DoubleVerify has traded down since its IPO and is now the subject of takeover talk from Nielsen, the audience-measurement seller, while retail and connected-TV sellers increasingly bundle their own campaign grading into the buy at no extra cost. That leaves Integral Ad Science, the last sizeable independent verifier, selling neutrality into a market where the largest sellers give a substitute away, with revenue concentrated in a handful of agency groups that are themselves squeezing vendor fees. A business in that position is worth more inside a bigger stack or under private ownership than as a small public company carrying public-company costs and quarterly growth expectations. The opposite path, IAS staying independent and public through 2027, requires buyers to start paying a premium for neutral grading at exactly the moment the sellers are making grading free.

Right if: Integral Ad Science issues a press release or SEC filing on or before 31 December 2027 announcing a definitive agreement to be acquired by another company or taken private. Wrong if: No such announcement exists by 31 December 2027 and Integral Ad Science remains an independent publicly traded company.

DoubleVerify Acquired by Nielsen; Vibe TV Acquired by Walmart Full Analysis → Read the source story →

Paramount Skydance will not own Warner Bros. Discovery on 31 December 2027.

Why California's Attorney General cancelled a scheduled meeting with the parties and cited bad-faith negotiating, which is unusual friction this early and signals that state regulators intend to extract concessions rather than wave the deal through. A purchase of this scale needs clearances from federal antitrust reviewers and several state attorneys general at once, and each holdout can add months by demanding conditions on pricing, carriage, or job commitments. Paramount Skydance also has to fund the whole of Warner Bros. Discovery, and the pool of lenders willing to underwrite premium media assets while cable declines is thin. Closing inside fifteen months would require the California friction to disappear, every other reviewer to stay quiet, and the financing to hold, and nothing in the record so far points that way. The likelier path is a deal that is still grinding through review, reshaped, or dead by the end of 2027.

Right if: Right if, as of 31 December 2027, Warner Bros. Discovery has not become a Paramount Skydance subsidiary according to the companies' filings with the US Securities and Exchange Commission, whether because the deal is still pending, was restructured, or was abandoned. Wrong if: Wrong if those filings show the acquisition completed on or before 31 December 2027.

Comcast Announces Full NBCUniversal Spin-Off; Media M&A Wave Uncertain Full Analysis → Read the source story →

One of the 25 largest US advertisers will pull, re-edit, or stop running an ad that uses AI-generated visuals after public criticism, on or before January 15, 2027.

Why Coca-Cola ran AI-generated holiday ads in both 2024 and 2025 and kept them on air through heavy public criticism, which is the pattern most large advertisers have followed: the complaints stay on social platforms while the cost savings stay in the budget. What has shifted is where the anger lands. Audiences react hardest when AI visibly stands in for the person they came to watch; AI used behind the scenes in production draws almost no heat. Holiday 2026 creative leans harder on synthetic on-screen talent and AI-made creator content, which puts more campaigns in the zone that provokes the strongest reaction. The case against is that big brands have so far absorbed the criticism without retreating, and pulling a campaign is a public admission most marketing teams will pay to avoid.

Right if: Ad Age reports on or before January 15, 2027 that an advertiser in its ranking of the 25 largest US advertisers pulled, re-edited, or stopped running an ad using AI-generated visuals following public criticism. Wrong if: No such report appears in Ad Age on or before January 15, 2027.

OpenAI creator brand trip sparks mainstream AI backlash signal Full Analysis → Read the source story →

By the close of the 2027 upfront selling season on June 30, 2027, Omnicom will publicly state that it is moving client identity work — matching customer records to advertising audiences — off LiveRamp, which is now owned by rival agency group Publicis.

Why LiveRamp built its business as neutral plumbing that every agency group could use to match customer records to ad audiences, and Publicis has now bought it outright. A rival holding company paying a competitor for the layer that decides which shopper sees which ad is funding the other side's pitch, and Omnicom already owns Acxiom, so it has an in-house substitute and little reason to keep quiet about switching once Publicis starts selling LiveRamp as a Publicis-only advantage in new-business pitches. The quiet alternative — Omnicom letting contracts run out without comment — is the comfortable path, but competitive pitches force public answers, and "our data spine is not owned by Publicis" is the cheapest one Omnicom has. If Omnicom instead stays on LiveRamp through the upfront, it signals that holding companies value the matching infrastructure more than they fear the conflict, and that the neutrality argument independents are counting on has no commercial bite.

Right if: Right if Omnicom, in a company announcement or in Ad Age, states on or before June 30, 2027 that it is shifting client identity matching away from LiveRamp. Wrong if: Wrong if no such Omnicom statement appears in a company announcement or in Ad Age on or before June 30, 2027.

Publicis acquires LiveRamp; Nielsen to acquire DoubleVerify Full Analysis → Read the source story →

YouTube will delay, remove, or loosen the rule that suspends a creator's ad earnings whenever their Shorts views fall below 10 million in any rolling 90-day window, before that rule takes effect in February 2027.

Why YouTube is raising its Partner Program entry bar in February 2027 from 4,000 to 8,000 long-form watch hours and from 10 million to 20 million Shorts views, and it is adding a recurring test that cuts off earnings in any 90-day stretch below 10 million Shorts views. The entry-bar increase is safe for YouTube because the channels it removes earn almost nothing, so creator anger there costs YouTube no revenue and buys it a cleaner brand-safety story with large advertisers. The rolling suspension is a different animal: it reaches creators who are already monetizing and already earning, turning one soft quarter into a pay cut, and every past YouTube monetization reversal has come when a rule bit people with real income. Recurring on-off eligibility also breaks the guaranteed minimums that multi-channel networks and creator platforms negotiate, which puts commercial partners, not just creators, in the complaint queue. The case for the rule surviving intact is that YouTube simply ships what it announced, but it has eighteen months of notice, no revenue riding on this clause, and a cheap way to defuse the loudest objection by softening only this piece.

Right if: As of March 31, 2027, YouTube's published Partner Program eligibility rules show the rolling 90-day Shorts suspension postponed, dropped, or triggered at a threshold below 10 million views. Wrong if: As of March 31, 2027, YouTube's published Partner Program eligibility rules have the rolling suspension in force at 10 million Shorts views per 90 days, as announced.

YouTube Doubling Watch-Hour Thresholds for Monetization in 2027 Full Analysis → Read the source story →

Sony will make its free, ad-supported PlayStation TV channels available on the PlayStation 4 by June 30, 2027.

Why Sony has put more than 100 free ad-supported channels on the PlayStation 5 and handed the ad serving to PubMatic and Publica, with chief executive Hiroki Totoki saying the priority is earning more from the roughly 125 million people already logged into PlayStation rather than selling more consoles. Only the PS5 slice of that 125 million can see the channels today, so the audience Sony can actually sell is well short of the number it quotes, and advertisers price reach they can verify. The buying platforms have no console category in their targeting yet, and they will not build one for a footprint this small, which keeps ad prices and fill soft until the addressable base grows. Extending to the PS4 costs Sony little incremental work because the ad stitching already sits with an outside vendor, and it is the only lever that changes the reach math quickly. The case against is that Sony wants PS4 owners to upgrade and treats the older box as end-of-life, but a wall of unsold ad slots on a sub-scale device is a worse outcome for the monetization strategy Totoki has staked out.

Right if: By June 30, 2027, PlayStation's free ad-supported channels are available to PlayStation 4 owners, per an announcement on Sony's official PlayStation Blog or a live listing in the PS4 store. Wrong if: On June 30, 2027, the free ad-supported channels remain PlayStation 5 only, or Sony has discontinued them.

Sony Launches FAST Channels on PS5, Taps PubMatic for Ad Serving Full Analysis → Read the source story →

By December 31, 2027, the US Federal Trade Commission will file a complaint against Google alleging that it placed undisclosed bids or hidden reserve prices in ad auctions it ran.

Why In September 2026 the FTC sued Amazon for allegedly inserting its own secret bid into its ad auctions for seven years, and at the same time opened an investigation into Google on the identical theory. Google's exchange has the structure the complaint attacks: the operator sees every bid before setting the clearing price, and its past floor-price programs are already documented in the record of the US ad tech antitrust case. Having written the first complaint, the agency has the template, the economists and the theory of harm, so a second filing costs it far less than the first. The obvious counterargument is that the Amazon case is weak on harm, since advertisers grew Amazon ad spend from about $4.6 billion in 2018 to roughly $68 billion last year while the alleged overcharging ran, but waiting for a ruling on that would take years and the FTC's interest is in outlawing the practice rather than punishing one company, which argues for moving while the theory is live.

Right if: The FTC files a complaint in court or an administrative action against Google on or before December 31, 2027 that alleges undisclosed bidding or undisclosed reserve pricing by Google in auctions Google operated. Wrong if: No such FTC complaint or administrative action against Google exists by December 31, 2027.

FTC Sues Amazon for Secretly Inflating Ad Auction Prices Seven Years Full Analysis → Read the source story →

OpenAI's ChatGPT advertising revenue booked for calendar 2026 will come in under $1 billion, against the $2.5 billion the company is targeting, as reported by The Information by March 31, 2027.

Why OpenAI's ad business is being described at a $1 billion annualized run rate, a figure that multiplies one strong month by twelve; the money actually booked in the first roughly 200 days of selling is about $330 million against a stated $2.5 billion target for 2026. Clearing that target needs roughly $540 million a month for the rest of the year, more than six times the current pace. Opening self-serve buying across 31 European markets is the move a seller makes when direct deals with large brands are not filling fast enough, and those brands stay capped at test budgets while impressions and brand safety on this inventory are still self-reported with no independent verification. For booked 2026 revenue to clear even $1 billion, monthly ad revenue needs to roughly double inside four months while advertiser procurement teams are still hand-writing waiver language to buy at all.

Right if: The Information reports by March 31, 2027 that OpenAI's ChatGPT advertising revenue for calendar 2026 was below $1 billion. Wrong if: The Information reports a 2026 ChatGPT advertising figure of $1 billion or more, or publishes no such figure, by March 31, 2027.

Update: OpenAI ChatGPT ads hit $1B run rate in under 200 days Full Analysis → Read the source story →

Sony will move its global media buying account out of WPP Media, with the change confirmed publicly by 15 May 2027.

Why Sony commissioned an audit of how WPP Media handled its media spending and came away disputing the rebates it was owed, the money agencies receive back from media owners that is supposed to flow to the client. WPP is fighting in court to keep those audit findings under seal, which is the behaviour of a group that expects other clients to copy the exercise if the numbers become public. The backdrop is worse than a billing dispute: executives at WPP's media arm in China were convicted of bribery and fraud, one of them sentenced to life, so a procurement team now has grounds to escalate from asking for an explanation to ending the relationship. The alternative, a quiet settlement that keeps the account, is how the 2016 transparency fight ended, but a client that has already paid for a forensic audit and taken the dispute to court has spent most of the goodwill that kind of settlement needs.

Right if: Sony, WPP, or reporting in Ad Age confirms by 15 May 2027 that Sony's global media buying has moved to another agency or in-house. Wrong if: Sony's global media buying still sits with WPP Media on 15 May 2027, including if neither party says anything.

ExchangeWire on WPP Vs. Foster, Apple App Data Consent, and YouTube Monetisation Full Analysis → Listen to the episode →

WPP will announce a signed agreement to sell its remaining stake in the research firm Kantar on or before it publishes its first-half 2027 results in August 2027.

Why WPP is carrying about $3.3 billion of debt with profit falling, and the cleanest cash available is the roughly 40% of Kantar it still holds alongside Bain Capital, which controls the business. That stake sits outside WPP's agency operations, so selling it raises money without disturbing a single client relationship, which is why it keeps surfacing whenever disposals are discussed. Bain has its own reason to consolidate ownership ahead of any eventual exit, so both sides have a motive to transact rather than wait. The alternative is another year of servicing the debt with no visible balance-sheet move from chief executive Cindy Rose, which is a harder position to defend to shareholders than selling a passive minority holding in a business WPP no longer runs.

Right if: WPP has publicly announced a completed sale or a signed agreement to sell its entire remaining Kantar stake by the publication of its first-half 2027 results. Wrong if: WPP still holds all or part of its Kantar stake at that publication with no completed or signed full sale announced.

WPP faces potential asset divestitures to cut $3.3 billion debt Full Analysis → Read the source story →

WPP will still wholly own both Ogilvy and VML when it publishes its first-half 2027 results in August 2027.

Why Bankers and press reports have floated Ogilvy and VML as disposal candidates to attack WPP's roughly $3.3 billion debt pile, but the value of those networks is welded to WPP's shared services and its ability to cross-sell creative alongside media buying. The moment either name appears in a formal sale process, the partners who generate the largest share of billings start taking recruiter calls and Fortune 500 procurement teams open contingency files, so the asset is marked down before a price is ever agreed. A buyer prices in both the standalone discount and that bleed, which means WPP would be selling its core at the worst possible multiple to fix a debt load its revenue base can still service. Carrying the debt through another year is the cheaper trade, and boards facing this choice have consistently sold the passive holdings first.

Right if: WPP wholly owns both Ogilvy and VML at the publication of its first-half 2027 results, with no announced agreement to sell, spin off, or take on an outside owner in either. Wrong if: By that publication WPP has sold, spun off, or announced an agreement to sell any ownership stake in Ogilvy or VML.

WPP faces potential asset divestitures to cut $3.3 billion debt

The Trade Desk will announce the acquisition of, or an equity stake in, a company that owns consumer purchase or behavioral data before it reports fourth-quarter 2026 results in February 2027.

Why Amazon's and Yahoo's ad-buying platforms keep winning head-to-head return-on-spend comparisons because they hold purchase and behavioral data on real shoppers, and The Trade Desk deliberately owns none of it, a position co-founder Jeff Green has defended for years while passing on chances to buy data owners. The pressure has moved from theory to money: Walmart dropped The Trade Desk as its exclusive partner and bought its own mobile ad-buying platform, Vibe, for a reported $1.4 billion, turning an anchor customer into a competitor. Once artificial intelligence makes campaign optimization a commodity, a platform that licenses everyone else's data has little left that a buyer cannot get from Amazon directly, so buying a data asset is the fastest available answer. Founder principles rarely survive a repricing stock plus defecting anchor clients, and the board is the party that has to keep funding the gap. Holding the line instead means absorbing another full year of losing the targeting comparison while rivals compound their advantage.

Right if: The Trade Desk publicly announces an acquisition of, or equity stake in, a company whose primary asset is consumer purchase or behavioral data, on or before its fourth-quarter 2026 earnings report in February 2027. Wrong if: No such acquisition or equity stake is announced by The Trade Desk's fourth-quarter 2026 earnings report in February 2027.

Episode 188: Rich Greenfield on the Meta Settlement and TTD's Stock Woes Full Analysis → Listen to the episode →

Snap will agree to a settlement with a coalition of US state attorneys general over harm to under-18 users, announced on or before December 31, 2027.

Why The multi-state action against Meta, reported at up to $18 billion plus mandated age-gating and cuts to data collection on minors, gives the attorneys general a reusable template built on the claim that algorithmic feeds are designed to hook teenagers, and that claim applies to every large social app rather than to Meta alone. Snap is the most exposed of the remaining targets: it already faces state litigation over youth safety, it lacks the legal budget and cash cushion Meta used to drag its cases out for years, and the design remedies now on the table are cheaper for Snap to accept than to fight. Once one platform signs a landmark order, the coalition's cost of bringing the same case again drops to near zero, and the holdout looks worse in front of a judge. The opposite outcome, Snap litigating past the end of 2027 with nothing signed, requires it to spend on a fight whose likely endpoint is the same set of design restrictions it could take now. The genuine risk is pace: multi-state negotiations routinely slip, and a court-approval timetable can push an agreed deal past the deadline.

Right if: On or before December 31, 2027, the office of at least one US state attorney general announces, or a court filing shows, a settlement with Snap resolving claims about harm to under-18 users. Wrong if: No such settlement announcement or court filing exists as of December 31, 2027.

MadTech Daily: Meta Faces Up to $18bn Teen Safety Settlement; X Revamps Creator Pay With New Rewards Program Listen to the episode →

By June 1, 2027, a top-20 US commercial bank or a major private-credit firm (Apollo, Ares, Blackstone, Blue Owl, KKR or a peer) will publicly announce a lending facility to a named ad-tech company secured against that company's unpaid customer invoices.

Why Ad-tech companies typically wait about 90 days to be paid by buyers while owing their own suppliers in about 30, and Silicon Valley Bank's 2023 failure removed the main lender that used to bridge that gap for venture-backed technology firms. Small specialty lenders such as OAREX have filled part of the hole by plugging directly into buying and selling platforms so they can watch a borrower's revenue in real time, proving the collateral is trackable and that the loans get repaid by large, creditworthy buyers. Private credit funds have raised record amounts and are actively buying or seeding specialty finance books rather than building underwriting teams from scratch, which makes an ad-tech invoice facility a cheap way in rather than an engineering project. The conventional view, that big lenders will keep avoiding a fragmented industry full of loss-making borrowers, describes banks better than it describes private credit, which is paid precisely to underwrite the assets banks decline. Invoices owed by large advertisers and platforms are self-liquidating collateral, and that is the easiest kind of book for a large fund to scale into quickly.

Right if: A top-20 US commercial bank or a major private-credit firm publicly announces, by June 1, 2027, a credit facility to a named ad-tech company secured against that company's unpaid customer invoices. Wrong if: No such announcement from a top-20 US commercial bank or major private-credit firm exists by June 1, 2027.

Adtech’s Financing Tax Full Analysis → Listen to the episode →

Through the close of the Cannes Lions festival in June 2027, no consumer-goods company outside Dollar General's own private-label brands will publicly say it shifted advertising money into Dollar General Media Network.

Why Dollar General is selling advertisers a shopper the big retail networks miss: 21,000 stores, about 2 billion transactions a year, and a claim that adding its audiences to a campaign delivers 50% reach no one else has. That 50% figure is Dollar General's own arithmetic with no published method, and unique is not the same as valuable, because a rural, low-income basket is small and skews to value products that national brands fund thinly. Buying it is cheap and easy, since it activates inside The Trade Desk and Google's DV360 with no new seat to learn, so brands that wanted to test it already can have; a year of quiet after easy tests means the results did not justify a public endorsement. Dollar General has every commercial reason to publish a win the moment it has one, having spent Cannes 2025 announcing itself as a media company. The opposite outcome requires the incremental-audience story to survive an advertiser's own conversion math on branded goods, not just private label, which is the part nobody has shown yet.

Right if: As of June 30, 2027, Dollar General Media Network's published announcements and case studies contain no named consumer-goods advertiser stating it moved advertising budget to Dollar General Media Network. Wrong if: By June 30, 2027, at least one named consumer-goods advertiser has publicly stated it moved advertising budget to Dollar General Media Network.

Austin Leonard: Awakening America’s Secret Retail Media Giant in an AI World Full Analysis → Listen to the episode →

By December 31, 2027, the US Federal Trade Commission will announce an enforcement action against an ad-tech intermediary — a demand-side platform, supply-side platform, or exchange — over consumer opt-out signals that failed to reach downstream partners.

Why When a user tells a website not to sell or share their data, that instruction routinely dies at the first hop: the consent management tool records it, and vendors further down the chain never receive it. Privacy enforcement so far has landed on consumer-facing brands and publishers, which are easy to test from a browser but fix only one leak at a time. Intermediaries are where the signal actually stops, and one case against a platform that resells data to hundreds of buyers forces compliance across the whole chain at once, which is why regulators eventually work upstream rather than staying at the retail end. The FTC has been examining how consent flows are implemented in practice, and state privacy laws now give it a well-documented trail of opt-outs that were recorded but never honored. The case for nothing happening rests on the agency staying focused on softer consumer-facing targets, which spreads enforcement effort thin for little structural effect.

Right if: The FTC publicly announces a complaint, consent order, or settlement by December 31, 2027 against a demand-side platform, supply-side platform, or ad exchange that cites failure to honor or pass along consumer opt-out signals. Wrong if: No such FTC action against a demand-side platform, supply-side platform, or ad exchange is announced by December 31, 2027.

S2E11: Just Be a Star. With Boltive CEO, Pamela Slea Full Analysis → Listen to the episode →

Informa TechTarget, the B2B technology publisher that sells advertising and buyer-intent data built on research-stage web traffic, will report full-year 2026 revenue lower than its 2025 revenue when it publishes results by March 2027.

Why Business software buyers increasingly do their early research inside AI chat assistants, arriving at vendors with a shortlist already formed and without visiting the trade sites that used to host that research. Informa TechTarget's core product depends on that visit happening: it sells ads against category research and sells vendors data on which companies are reading about their category on its properties. Fewer research visits shrinks both the ad inventory and the quality of the intent signal at the same time, so the damage compounds rather than offsetting. Growth in 2026 would require research-stage traffic to hold up even as the tools that answer those questions get better and more widely used inside enterprises. Relationship-sold ad-tech and enterprise deals still close in the room, but the publishers monetizing the research stage sit directly in the path of the change.

Right if: Informa TechTarget's reported full-year 2026 revenue is below its reported full-year 2025 revenue. Wrong if: Informa TechTarget's reported full-year 2026 revenue is equal to or above its reported full-year 2025 revenue.

Asana CMO Prachi Gore on how AI workflows are ‘completely disrupting’ the B2B playbook Listen to the episode →

By 31 May 2027, the Federal Trade Commission will announce a children's-privacy enforcement action carrying a penalty of $100 million or more against a company other than TikTok.

Why TikTok agreed to pay $400 million in 2026 to settle US claims it collected data from under-13 users without parental consent, a figure that dwarfs the $5.7 million it paid under its 2019 settlement over the same conduct. Part of that penalty was for breaching the 2019 order itself, which is extra proof work a regulator only takes on when it intends to treat old privacy settlements as live obligations across the board, and a long list of platforms and data businesses signed similar orders over the past decade. Nine-figure children's-privacy penalties already have precedent at this scale, with Epic Games at $275 million in 2022 and Google and YouTube at $170 million in 2019, so the ceiling is established rather than hypothetical. The case against this happening is docket speed: these actions take years to build and the agency may simply not have one ripe before mid-2027, which is why this is a call and not a certainty.

Right if: The FTC announces, on or before 31 May 2027, a children's-privacy enforcement action against a company other than TikTok with a monetary penalty or settlement of $100 million or more. Wrong if: No such FTC children's-privacy action at $100 million or more against a company other than TikTok is announced on or before 31 May 2027.

MadTech Daily: New Support for ASA Awareness Campaign; TikTok Settles Children’s Privacy Case for USD$400m Full Analysis → Listen to the episode →

No top-five US streaming service — Disney+, Hulu, Max, Paramount+, or Peacock — will announce that its subscriptions can be bought and watched inside the Netflix app before the 2027 upfront selling season closes on 15 May 2027.

Why Netflix is reported to be weighing whether to sell rival streaming services through its own app, which would turn the largest logged-in video audience outside Google and Meta into the front door for everyone else's subscriptions. The closest working precedent, Apple TV Channels, has operated for roughly seven years and has never pulled in a top-tier competitor, because joining means handing a rival the subscriber relationship, the sign-in, and a cut of the ad-supported tier that is now each major's growth story. Disney, Warner Bros Discovery, Paramount, and NBCUniversal are all assembling their own bundles and selling their own ad inventory, so becoming a tenant in Netflix's interface costs them the two assets they are spending most heavily to build. The services with a reason to sign are the smaller ones that cannot afford to buy subscribers on their own. A major signing would require one of these companies to conclude its standalone app has failed badly enough to rent shelf space from its biggest competitor, and none of them is at that point within this selling cycle.

Right if: By 15 May 2027, none of Disney+, Hulu, Max, Paramount+, or Peacock has publicly announced distribution of its subscription inside the Netflix app. Wrong if: By 15 May 2027, at least one of Disney+, Hulu, Max, Paramount+, or Peacock has publicly announced distribution of its subscription inside the Netflix app.

Netflix Considering Selling Access to Other Streaming Services Read the source story →

The Trade Desk will announce a deal to buy a supply-side ad technology company, meaning a business that sells advertising space on behalf of publishers, such as Magnite or PubMatic, before it reports full-year 2027 results in February 2028.

Why The Trade Desk announced Ventura, its own smart-TV operating system, in November 2024, promised it for the first half of 2025, shipped no hardware, and saw its Sonos tie-up go nowhere; Matthew Henick, the senior vice president who owned the project, left in April 2026 to run the immersive art company Meow Wolf, and no successor effort has been named. Chief executive Jeff Green spent two years telling investors the connected-TV supply chain is broken and that his company would fix it, and owning the software layer between viewer and ad was the fix. Rebuilding that effort needs hardware spending and a consumer brand The Trade Desk does not have, while buying an established publisher-side platform is an off-the-shelf purchase it can fund from cash and stock. Staying purely on the buying side leaves it renting access from Roku, Amazon, and Google on their terms with no answer to the problem it publicly diagnosed, which is why a supply-side acquisition is the likelier route back to controlling inventory.

Right if: The Trade Desk publicly announces a definitive agreement to acquire a company whose primary business is selling publisher advertising inventory at any point before it reports full-year 2027 results in February 2028. Wrong if: No such acquisition agreement is announced by The Trade Desk's full-year 2027 results in February 2028.

Trade Desk's Ventura TV OS Exec Departs for Immersive Art Company Read the source story →

The Trade Desk's total revenue growth for full-year 2026, reported at its fourth-quarter earnings in February 2027, will be lower than Magnite's total revenue growth for the same year.

Why After The Trade Desk's February 2025 revenue miss, media buyers at the large agency groups began qualifying Amazon's DSP and Google's DV360 as primary buying tools, and the company's shares are down roughly 90% from their December 2024 high. In August 2026 The Trade Desk made the Programmatic Table, the signature interface of its 2023 Kokai relaunch, optional, which addresses workflow complaints rather than the demand loss that preceded them. Magnite sells inventory to whichever buying platform wins, so budget shifting away from The Trade Desk still passes through Magnite's pipes and shows up in its revenue. The opposite result, The Trade Desk out-growing the sell-side, requires agencies to unwind migrations they have already paid to build and staff, and the money and internal politics run the other way.

Right if: The Trade Desk's reported year-over-year total revenue growth rate for full-year 2026 is below Magnite's reported year-over-year total revenue growth rate for full-year 2026. Wrong if: The Trade Desk's reported year-over-year total revenue growth rate for full-year 2026 equals or exceeds Magnite's.

Trade Desk Launches 'Zuma' Update, Makes Kokai's Signature UI Optional Read the source story →

As of June 30, 2027, the share of European iPhone users who agree to app tracking when Apple's permission prompt appears will still be 30% or below, according to AppsFlyer's published opt-in benchmark.

Why Since 2022 the opt-in rate has sat at roughly 24-25%, because most people reflex-tap the refusal button rather than read the screen, so changing the words on the prompt changes what is written, not what users do. Germany's competition regulator found Apple's consent design favored Apple's own advertising business, but what it extracted is corrected wording, not a requirement that Apple's prompt work exactly like the one it forces on other apps. Apple has a direct financial interest in keeping the friction that starves rival measurement firms of signal, so it will comply narrowly. For the rate to jump past 30% the current number would have to be mostly an artifact of Apple's language, and three flat years across several prompt revisions say it is behavioral instead. If copy were that powerful a lever, the number would already have moved.

Right if: AppsFlyer's published opt-in benchmark shows European iOS tracking opt-in at 30% or below on or before June 30, 2027. Wrong if: AppsFlyer's published benchmark shows European iOS tracking opt-in above 30% on or before June 30, 2027, or AppsFlyer publishes no such European figure by that date.

Germany Finds Apple's ATT Framework Violated Competition Law Read the source story →

By 31 March 2027, Ad Age will report that one of the world's 20 largest advertisers has moved media buying out of WPP, with the advertiser or its incoming agency publicly giving rebates or markups on resold media as a reason.

Why Richard Foster's amended $100 million wrongful-termination suit against WPP ties his claims to a Sony-prompted audit of WPP Media's rebate practices in China, the same market where WPP executives were convicted of bribery and fraud, and WPP is asking a New York court to seal parts of the filing. A company fighting that hard to keep pages out of the public record gives every client procurement team a written reason to trigger the audit clauses that sit in most holding-company contracts, and once an audit shows how much the agency earns on media it buys and resells, the cheapest fix for a chief financial officer is to move the account to a shop that does not do it. The 2016 industry transparency fight produced anger but no lawsuit forcing documents into daylight, so budgets stayed put; this time there is a live filing, a named client audit, and rivals such as Publicis selling clean hands as a product. The opposite case is that procurement asks questions, receives paperwork, and renews anyway, which is what happened for a decade, but that pattern held when no large advertiser had a headline reason to justify the switching cost to its own board.

Right if: Ad Age reports on or before 31 March 2027 that a company ranked in the world's 20 largest advertisers has shifted or ended media buying with WPP, and the advertiser or its new agency states publicly that rebates or markups on resold media were a reason. Wrong if: No such Ad Age report exists by 31 March 2027, including cases where large advertisers run transparency audits of WPP but no top-20 advertiser publicly moves business on those grounds.

WPP Seeks to Seal Parts of $100M Foster Whistleblower Lawsuit Read the source story →

By June 30, 2027, OpenAI will publicly offer merchants a way to pay for placement or ranking in ChatGPT's shopping results.

Why ChatGPT's shopping tab turns product hunting into a front-door feature, and OpenAI's public position is that what appears there is ranked on merit, with money changing hands only when a sale closes. Shopping intent is the most valuable inventory on the internet, and Google built its search business by auctioning exactly that moment; a commission on completed purchases pays a small fraction of what the same shopper is worth to an auction. OpenAI is carrying very large compute costs and put Fidji Simo, formerly chief executive of Instacart, in charge of its applications business, a background built on selling placement to consumer brands. Staying purely organic would mean giving away the highest-value queries it has while Amazon and Google keep charging for theirs, which is the harder choice for a company that needs revenue at scale quickly.

Right if: By June 30, 2027, OpenAI's public announcements or its published merchant documentation show merchants can pay OpenAI for placement, ranking, or visibility in ChatGPT shopping results. Wrong if: On June 30, 2027, OpenAI's public announcements and merchant documentation show shopping results remain unpaid, with OpenAI earning only sales commissions, referral fees, or nothing from merchants.

OpenAI Adds Dedicated Shopping Tab to ChatGPT Sidebar Read the source story →

By December 31, 2027, Amazon will pay Twitch streamers for the right to use their videos to train its AI models.

Why Amazon is being sued over training AI on Twitch streamers' videos even though Twitch gave streamers a way to opt out, and the case argues that a default-on setting is not the same as permission. Amazon cannot un-train a model, so the only practical way to de-risk the next generation of models is to buy the rights going forward rather than assume silence counts as a yes. Amazon has already run this play on the publishing side, paying the New York Times on the order of $20-25 million a year for AI training rights, which means the contracts, the price-setting and the internal approval path all exist. The alternative, holding the opt-out default until a court rules, leaves every model Amazon builds on Twitch video carrying an unpriced liability that grows with each new release, and buying peace from creators is cheaper than losing the argument once.

Right if: Amazon or Twitch publicly announces payments, revenue share or license fees to streamers in exchange for using their video to train AI models, on or before December 31, 2027. Wrong if: No such payment, revenue share or license program for AI training on streamer video is publicly announced by Amazon or Twitch on or before December 31, 2027.

MadTech Daily: Amazon Sued Over AI Training on Twitch Videos; UK Digital Infrastructure Investment Hits Dot-Com Era Levels Listen to the episode →

Alphabet's Google Network revenue — the ads Google sells on other companies' websites and apps — will come in below $29 billion for full-year 2026 when Alphabet reports fourth-quarter results in February 2027.

Why Google's AI Overviews and AI Mode answer questions on Google's own page, and search referral traffic to publishers is already down 30 to 40% at some sites; People Inc. has cut its reliance on Google search from roughly 60% of traffic to about 20%. The sessions that disappear are exactly the open-web page views that AdSense and Google Ad Manager monetize for third-party sites, and the audience that remains valuable is moving to email, direct visits and apps, surfaces Google's network business does not reach. Google Network revenue has already drifted down from about $31.7 billion in 2022 to about $30.4 billion in 2024, and the erosion mechanism is now accelerating rather than easing. The opposite case — that network revenue stabilises or grows — depends on Google replacing lost open-web impressions with new third-party supply, but its own AI answer formats keep the impression and the ad dollar on Google-owned pages, where it books as Search revenue instead.

Right if: Alphabet's reported full-year 2026 Google Network revenue is below $29 billion. Wrong if: Alphabet's reported full-year 2026 Google Network revenue is $29 billion or higher.

The case for and against publishers blocking Google Full Analysis → Listen to the episode →

Snap's advertising revenue for the fourth quarter of 2026, reported in its February 2027 earnings release, will be lower than its advertising revenue for the fourth quarter of 2025.

Why Meta is closing on a settlement with nearly every US state, worth up to $18 billion, over claims it designed Facebook and Instagram to hook teenagers, and the states also want limits on how under-18s use the apps. Meta can absorb both the cash and the compliance work against roughly $164 billion of annual revenue, but the settlement sets a public floor on what counts as acceptable teen targeting, and agency brand-safety teams scrub youth-adjacent audience segments on the headline rather than waiting for a court order. That repricing lands hardest on the platform with the youngest user base and the thinnest revenue cushion, dragging on declared 13-17 buckets and on the 18-24 lookalike audiences trained off the same restricted signals. The case for Snap growing anyway rests on the argument that age inference survives any ban on explicit targeting, which is true of the machinery but beside the point when the buyer, and not the regulator, is the one moving the money. Advertisers pull first and return quietly later, and the return will not happen inside this quarter.

Right if: Snap's reported advertising revenue for Q4 2026 is below its reported advertising revenue for Q4 2025. Wrong if: Snap's reported advertising revenue for Q4 2026 is equal to or above its reported advertising revenue for Q4 2025.

Meta Faces Up to $18bn Teen Safety Settlement Across US States Full Analysis → Read the source story →

MNTN, the streaming-TV ad platform that sells simplified campaign buying to mid-market advertisers, will report lower revenue for the first half of 2027 than for the first half of 2026 when it publishes second-quarter 2027 results in August 2027.

Why Roku has hired Patrick Harris, who spent a dozen years selling to small businesses at Meta, to build a self-serve streaming-TV product that lets a local advertiser buy airtime without a salesperson, and Amazon is handing its sellers free AI tools that turn a product page into a video spot. MNTN's whole business is being that easy front door onto other companies' streaming inventory, so when the inventory owners build the front door themselves and price it near cost, the reseller's margin goes first and its volume follows. MNTN also buys much of what it resells from the same platforms now competing with it, which caps how far it can undercut them on price. The case for MNTN continuing to grow rests on streaming ad budgets rising fast enough to lift every seller, but rising budgets are exactly what pull Roku and Amazon into the small-advertiser segment directly.

Right if: MNTN's reported revenue for the six months ending June 2027 is below its reported revenue for the six months ending June 2026. Wrong if: MNTN's reported revenue for the six months ending June 2027 is equal to or above its reported revenue for the six months ending June 2026.

Roku Hires Meta's Patrick Harris to Chase SMB CTV Advertisers Read the source story →

By June 30, 2027, at least one of the twenty largest US advertisers will publicly say it has moved performance budget - spend judged on sales, not awareness - into ads inside ChatGPT.

Why OpenAI is rebuilding Meta's direct-response ad product feature for feature: self-serve buying, product carousels, audience uploads, cost-per-click and cost-per-thousand bidding, conversion tracking and automated creative. Because the buying tools look familiar, agency performance teams can stand up a test in days, the same way they stood up TikTok in 2019, and big advertisers have a long habit of publicising early wins on a new surface as proof they are ahead of their category. The counter-case is that conversational sessions are a noisier path to purchase than a product-catalogue click and OpenAI has shown no audited proof of incremental sales, which would cap it at small brand experiments nobody brags about. That caps the size of the money, not the willingness to talk about it, and one public claim of real performance spend is a low bar for a surface this heavily covered. The quiet outcome requires every large advertiser to both test and stay silent through two full planning cycles.

Right if: On or before June 30, 2027, a named advertiser among the twenty largest US ad spenders states publicly that it has shifted performance budget into ads inside ChatGPT. Wrong if: No such public statement from a top-twenty US advertiser exists by June 30, 2027.

OpenAI Ad Platform Rapidly Replicating Meta's Feature Set Read the source story →

Peacock's ad-supported tier will not be available to YouTube Premium subscribers in the United States by December 31, 2027, going by YouTube's published list of Premium benefits.

Why NBCUniversal and YouTube announced a multiyear deal to fold Peacock's ad-supported tier into YouTube Premium starting in 2027, but said nothing about how ads will run inside a subscription whose entire selling point is no ads, or who gets the ad revenue. Resolving that contradiction means either cutting the ad load, which removes NBCU's reason to do the deal, or showing ads to people who paid to avoid them, which drives cancellations, and neither side has said which. Underneath the marketing sits roughly a year to eighteen months of grinding technical work: stitching ads into the stream, reconciling viewer identity across two systems, and agreeing on whose measurement grades the impressions. The announcement's timing is driven by Comcast spinning NBCUniversal into a standalone public company that needs a growth story for its first earnings decks, which is a reason to announce early rather than a reason to be ready. An on-time launch would require both parties to have already settled the ad economics they conspicuously declined to describe.

Right if: As of December 31, 2027, Peacock's ad-supported tier is not available to US YouTube Premium subscribers as part of their subscription. Wrong if: On or before December 31, 2027, US YouTube Premium subscribers can access Peacock's ad-supported tier as part of their subscription.

NBCU and YouTube Strike Multiyear Global Distribution Deal for Peacock Read the source story →

By 30 June 2027, at least one of the five largest agency holding companies (WPP, Publicis, Omnicom, Dentsu, Havas) will publicly announce client advertising spend committed through the Ad Context Protocol, the open standard backed by Scope3 that lets AI software buy ads on its own.

Why Live spend running through agent-based buying is tiny today: pubX chief executive Andrew Mole puts real volume at $2,000 to $3,000 a day, and both Scope3 and the IAB Tech Lab, which backs a rival standard called AAMP, have said in writing that the plumbing is not running at production scale. The forcing event is the first quarter of 2027, when annual programmatic spending commitments come up for renewal and budget can be re-pointed without breaking a contract. Holding companies are selling clients an artificial-intelligence story and a public commitment to a named protocol is cheap to announce and expensive to skip while rivals do it, which is why at least one moving first is more likely than all five staying quiet. The alternative, total silence from every large agency group through a full renewal cycle and another year of AI positioning, would require an industry that announces partnerships weekly to pass on the loudest available one.

Right if: One of WPP, Publicis, Omnicom, Dentsu or Havas publicly announces, on or before 30 June 2027, client media spend committed through the Ad Context Protocol. Wrong if: None of those five agency groups has made such a public announcement by 30 June 2027.

Agentic Ad Buying Not At Scale, Standards Bodies Admit Full Analysis → Read the source story →

The Trade Desk's year-over-year revenue growth in the first quarter of 2027 will be slower than its year-over-year growth in the first quarter of 2026.

Why Advertisers have been shifting programmatic budget to Amazon's DSP because Amazon ties ads to retail purchase data and moves measured return, not because The Trade Desk's screens were hard to use. Ask Koa, the conversational front end that moved from alpha to closed beta this summer, speeds up campaign setup and audience building but does nothing to connect open-web buying to purchase outcomes. The Kokai relaunch made the same promise that better tooling would bring migrated budget back, and it did not, which is why a new interface exists at all. For growth to accelerate instead, agencies would have to move spend back on the strength of a better planning interface, and buyers do not change demand-side platforms for that. A closed beta also cannot carry enough spend during the period to change the top line either way.

Right if: The Trade Desk's reported first-quarter 2027 revenue growth rate versus the prior year is below its first-quarter 2026 growth rate versus the prior year. Wrong if: The Trade Desk's reported first-quarter 2027 revenue growth rate versus the prior year equals or exceeds its first-quarter 2026 growth rate versus the prior year.

The Trade Desk Launches 'Ask Koa' Agentic AI Interface Read the source story →

By March 1, 2028, the newly independent NBCUniversal spun off from Comcast will have announced an agreement to be acquired outright or to sell a controlling stake in itself to another company.

Why Comcast's cable cash flow was what paid for Peacock's losses, and once the two companies trade separately that subsidy is gone for good, leaving a distant-fourth streaming service to fund sports rights and content against Netflix and Amazon on its own public balance sheet. A clean, separately listed entertainment company with a premium film and television library, live sports and live news is a far easier thing for a large technology or media buyer to purchase than the same assets buried inside a cable conglomerate, which is the practical effect of the spin. The opposite path, an independent NBCUniversal buying its way to scale, is the weaker case, because the split removes exactly the capital that would fund those purchases while public shareholders press for a near-term path to profit. Sellers under quarterly margin pressure in the middle of the most expensive content-and-sports arms race in the industry's history tend to look for a bigger balance sheet to stand behind rather than build one.

Right if: NBCUniversal or a named buyer publicly announces a signed agreement for an outright acquisition of NBCUniversal or for a stake giving the buyer control, on or before March 1, 2028. Wrong if: No such signed agreement has been publicly announced as of March 1, 2028.

NBCU Spinoff from Comcast to Create Independent Public Entertainment Company Read the source story →

By May 15, 2027, OpenAI will offer advertisers an ad product in ChatGPT that is priced on, or automatically optimized toward, purchases completed inside ChatGPT.

Why OpenAI is building its advertising business around leaders recruited from Meta's performance ad teams, and the products those people built depend on a direct line from an ad to a purchase. ChatGPT's audience is large but mostly logged-out and its sessions are short, so browsing behavior alone does not supply that line; its in-chat checkout with merchants does, because it records the actual sale. The people now running the business have spent their careers pricing ads on outcomes, and wiring checkout records into the ad manager is a faster route to revenue than selling reach numbers to performance buyers who will not pay much for them. Staying on impressions and engagement through spring 2027 would leave OpenAI chasing brand budgets with no proof of results, which is the weaker commercial position, and holdco buyers have shown they will not fund a new channel on unverified reach alone.

Right if: On or before May 15, 2027, OpenAI has publicly made available to advertisers an ad product in ChatGPT whose pricing or automated bidding uses purchases completed inside ChatGPT. Wrong if: As of May 15, 2027, every ChatGPT ad product available to advertisers is still priced and optimized solely on impressions, clicks, or engagement.

Meta Alumni Flood OpenAI, Amazon, TikTok, Roku Ad Teams Read the source story →

No advertiser will be publicly named as a paying buyer of Gannett's sponsored-content placements inside AI chatbot answers on or before Gannett's second-quarter 2027 earnings report, due in late July 2027.

Why Gannett has rebuilt USA Today's pages into machine-readable formats and blocks roughly 99% of crawlers that are not paying, and chief executive Mike Reed told investors on the August 6, 2026 earnings call to expect more AI licensing deals this year; alongside that, the company is pitching brands on making their sponsored content visible inside answers from ChatGPT and similar assistants. A brand buyer asked to fund that placement has no way to learn how many people saw the answer, what sat next to it, or what sentence the model wrapped around the message, because the companies running the models do not hand out logs of what their answers surfaced. Advertisers have been burned by 'new format, trust us' pitches before and typically park such ideas in small curiosity budgets that neither side wants its name attached to. For the opposite to happen, a brand would have to accept a placement it cannot verify and then agree to be the public face of it, which is the order in which almost no ad category has ever grown.

Right if: Right if, as of Gannett's second-quarter 2027 earnings report in late July 2027, neither Gannett nor any brand has publicly identified a named advertiser that paid for sponsored-content placement inside AI chatbot answers. Wrong if: Wrong if, on or before that report, Gannett or the brand itself publicly names at least one advertiser that has paid for such placement.

Gannett / USA Today Reformats Content to Win AI Licensing Deals Full Analysis → Read the source story →

Nielsen's Gauge report covering July 2027, published in August 2027, will show ad-supported streaming at 48% or more of ad-supported television viewing among adults 18 and up.

Why In 2026 Nielsen stopped counting viewers aged 2 to 17 in its headline Gauge figure and recentered on adults 18 and up, and ad-supported streaming's share of ad-supported TV viewing fell from 48.2% to 44.4% overnight with no change in what anyone actually watched, because children skew heavily to streaming. That handed broadcast and cable sales teams a 3.8-point reach argument to defend rates with for one upfront season, and they will use every day of it. But streaming keeps adding both viewing time and ad-supported inventory among adults, which is what has pushed its share up several points a year, so the definitional gift has a short shelf life. Clearing 48% among adults within a year would erase the entire swing Nielsen created. The alternative, adult streaming share stalling in the mid-40s, requires broadcast and cable to hold adult viewing time flat, which continued cord-cutting and the migration of sports to streaming services have not allowed.

Right if: Nielsen's Gauge for July 2027 puts ad-supported streaming at 48.0% or more of ad-supported TV viewing among adults 18 and up. Wrong if: Nielsen's Gauge for July 2027 puts that figure below 48.0%.

Nielsen Gauge: 71.5% of TV Viewing Is Ad-Supported; Methodology Shift Depresses Streaming Share Read the source story →

Through the close of the 2027 upfront on June 30, 2027, no national game inventory from the NFL, NBA, or MLB will be sold on a guaranteed-audience deal where iSpot or VideoAmp is the currency the guarantee is measured against.

Why Paul Ballew, the NFL's chief data and analytics officer, has publicly attacked Nielsen's revised method for counting how many people watch the same screen together, warning it will push reported sports audiences down, and he did it right after the league's best-rated season since 1989. A seller objecting to a measurement change immediately after posting record numbers is angling for better terms from its measurement supplier, because the complaint is that the next count will be smaller, not that the last one was wrong. For a rival measurement firm to actually take the business, an agency holding company has to put nine-figure delivery guarantees behind iSpot or VideoAmp and accept liability for free makegood inventory when those numbers come in under a record-season benchmark, and no buyer has taken that risk at football scale. Agencies also carry no budget or headcount to reconcile two scorecards that disagree about the same game every Monday morning. The outcome where a league actually moves requires a buyer willing to be first on an untested currency, and the leagues extract more by threatening to leave than by leaving.

Right if: By June 30, 2027, no public announcement from a league, its network partner, or a buying agency, as reported in Ad Age, describes guaranteed NFL, NBA, or MLB national inventory transacted against iSpot or VideoAmp numbers. Wrong if: By June 30, 2027, at least one such guaranteed deal on iSpot or VideoAmp numbers is publicly announced by a league, network, or agency and reported in Ad Age.

NFL Publicly Challenges Nielsen Over Measurement Methodology Changes Read the source story →

On or before May 15, 2027, Roku will announce that a defined block of its US advertising inventory can be bought only through Amazon's demand-side platform, closing that inventory to third-party exchanges.

Why Roku and Amazon published a joint study claiming their direct integration delivered 21% lower cost per unique viewer and 66% more unique viewers than buying the same Roku inventory through a third-party exchange, a comparison in which the exchange path had its audience matching switched off, which Amazon's own copy described as 'artificially disabled.' Companies do not publish a loaded comparison like that unless they plan to act on the conclusion: Amazon owns the buying platform, Roku owns the screens, and the two now jointly own the measurement that says the middle layer is unnecessary. The cheapest way to make that claim true is to fence off inventory only Amazon's platform can buy, which strips exchange fees out of the path and gives Amazon a reason to keep steering budget to Roku. The case against is that Roku leans on exchange competition to hold up prices, but that argument applies to leftover inventory, not to the logged-in premium audiences where the joint deal has the most to gain.

Right if: Roku's public announcements or its first-quarter 2027 earnings materials state, on or before May 15, 2027, that some defined portion of its US ad inventory is purchasable only through Amazon's demand-side platform. Wrong if: As of May 15, 2027, every portion of Roku's US ad inventory remains purchasable through at least one third-party exchange.

Roku/Amazon Benchmark Test Disabled Competitor's Targeting to Claim Superiority Read the source story →

Paramount will complete its $110 billion acquisition of Warner Bros Discovery on or before December 31, 2027.

Why Paramount opened settlement talks with California early, which is the behavior of a buyer paying to clear an obstacle instead of litigating for years against California and the eleven other states challenging the deal on film distribution and pay-TV competition. State attorneys general have no power to block a federally reviewed transaction on their own; they can only delay it, and the standard outcome is behavioral commitments banked quickly instead of a coin-flip trial. Both boards need the transaction to happen for balance-sheet reasons, with roughly $40 billion of debt sitting on Warner Bros Discovery, so neither side has an incentive to let the clock run. The realistic failure mode is not a court killing the deal but the calendar slipping past 2027 while conditions are negotiated, and a fifteen-month runway from here is tight enough that this is a live call rather than a formality.

Right if: The acquisition has legally closed, with Warner Bros Discovery under Paramount ownership, on or before December 31, 2027. Wrong if: The acquisition has not legally closed by December 31, 2027, whether because it is still pending, blocked, or abandoned.

Paramount Opens Talks with California Over $110B Warner Bros Deal Read the source story →

By February 28, 2027, Meta will again let advertisers exclude at least one specific named placement, such as Facebook search results or Reels in-stream video, from their campaigns.

Why Meta has removed the manual controls that let buyers block individual placements, so ads are now assigned by its automated system on the basis of audiences and outcomes. Banks, pharmaceutical companies, and airlines operate under audit and outside-counsel review that requires them to show where an ad ran, and "the system decided" is not an answer those reviews accept, so this slice escalates formally instead of complaining quietly. Meta has run this play before: strip a control for the sake of automation, absorb the complaints, then bring back a narrower version once large spenders threaten budget. The case for Meta holding the line is real, since performance advertisers are the bulk of billings and have no better destination, but one ugly, screenshottable adjacency at a household-name brand is cheap for Meta to prevent and expensive to survive. Restoring a single narrow exclusion costs Meta almost nothing in returns and buys off the loudest buyers.

Right if: On February 28, 2027, Meta's Business Help Center placement documentation shows advertisers can exclude at least one specific named placement from a campaign. Wrong if: On February 28, 2027, that documentation shows no way to exclude any specific named placement, offering only audience and outcome buying with broad brand-safety inventory tiers.

Meta Removes Manual Ad Placement Exclusion Controls for Advertisers Read the source story →

Meta's advertising revenue will grow faster in 2026 than it did in 2025, measured from the full-year figures in its fourth-quarter 2026 earnings report in late January 2027.

Why Jerry Murdock, a co-founder of the investment firm Insight Partners, has publicly argued that Meta is a short and is becoming a dividend-paying utility that stops investing in its product, which for a media buyer would show up as targeting and measurement on Instagram and WhatsApp quietly stalling while prices keep climbing. The counter-mechanism is that Meta's automated ad ranking and automated creative generation let it charge more for the same inventory every quarter, and advertisers with performance budgets have no substitute at comparable scale to move that money to. A company harvesting yield cuts product spending and lets pricing power decay slowly, which would show as a growth rate falling below the prior year's; a company still compounding pricing power on the same ad load shows the opposite. Deceleration off a large base is the easy assumption, and that is precisely why the call is worth making on the other side.

Right if: Meta's full-year 2026 advertising revenue growth rate, as reported in its fourth-quarter 2026 earnings, is higher than its full-year 2025 advertising revenue growth rate. Wrong if: That 2026 growth rate is equal to or lower than the 2025 growth rate.

20VC: The AI Bubble Will Burst: Half the Neoclouds Will Die | China: Should We Ban Chip Exports & Be Fearful of Chinese Open-Source | Mag7: Who Dies and Who Thrives: Why Meta is Meh and Microsoft is Mega Listen to the episode →

Google will publicly commit to selling or spinning off AdX, its ad exchange, by October 1, 2027.

Why Google is rewiring its ad server, Google Ad Manager, so buyers and sellers can negotiate and run campaigns through AI tools plugged directly into it, which means less of the market has to clear through the exchange at all. The US Justice Department's remedy push in the ad-tech monopoly case is aimed squarely at that exchange, and the ad server is the piece that actually locks publishers in, so the exchange is the cheaper thing for Google to surrender. Regulators are fighting over a toll booth on a road Google is already rerouting, and Google's lawyers can see that trade. Holding out to keep both assets means years more litigation and the live risk that a court orders the ad server broken off instead, which is a far worse outcome than divesting a business Google is designing around.

Right if: Right if, by October 1, 2027, Google has announced or agreed in a filing in the US Justice Department's ad-tech case in the Eastern District of Virginia that it will sell or spin off AdX. Wrong if: Wrong if no such announcement or filed agreement exists by October 1, 2027 and Google is still contesting divestiture of AdX.

Episode 187: Jack Raines Trolls LinkedIn and Ends Up with a Book Deal Full Analysis → Listen to the episode →

By 30 September 2027, DoubleVerify, Integral Ad Science, or Comscore will announce the acquisition of a company whose main business is measuring how often brands are mentioned in AI chatbot answers.

Why Agencies including Havas Media Network are now selling clients their own in-house scoring of how often a brand shows up in AI answers, which puts the seller in charge of grading its own work, an arrangement buyers have eventually rejected in every earlier phase of search measurement. That gap has already attracted a crop of venture-funded startups building crawlers that sample chatbot responses at scale, and sampling rights, model coverage and historical data are the hard part to replicate from a standing start. The verification vendors have consistently bought their way into new categories rather than growing them internally, as DoubleVerify did with Scibids and OpenSlate and Integral Ad Science did with Publica, because buying an existing customer list is faster than convincing brands a bolt-on feature counts as independent. The alternative path, quietly adding a chatbot-mention report to an existing dashboard, leaves them selling a free-feeling feature against dedicated challengers already signing brand contracts, which is the weaker commercial position of the two.

Right if: DoubleVerify, Integral Ad Science, or Comscore publicly announces an agreement to acquire a company whose primary product measures brand presence in AI chatbot or AI search answers, on or before 30 September 2027. Wrong if: None of the three has announced such an acquisition by 30 September 2027, including cases where they build the capability in-house, partner instead, or do nothing.

Havas Media Network's Amy Banks on SEO, GEO, and Zero Click Search Full Analysis → Listen to the episode →

By June 30, 2027, at least one of Peacock, HBO Max, or Paramount+ will make ads on its TV home screen — the menu screen a viewer sees before choosing what to watch — buyable through a demand-side platform, announced publicly by the streaming service or its platform partner.

Why Device makers have already moved: Samsung put its home screen inventory on The Trade Desk, Magnite standardized pause ads, and Nexxen built a tool that resizes one creative across TiVo, Vizio/VIDAA and other manufacturers so buyers can traffic home screen units without hand-cutting specs per device. The streaming services have held their home screens back as hand-sold deals to protect a premium, but that premium rests on scarcity rather than proof, since the only measurement available for these units today is media mix modeling and most buyers do not run one. As device-maker supply sets a cheap clearing price for the same screen and agencies consolidate buying into a handful of platforms, a hand-sold-only home screen unit gets skipped in planning rather than bid up. Continuing to hold the line is the comfortable assumption, but it requires the ad-supported tiers to keep turning down incremental demand on their single most-viewed surface through a full upfront cycle, which is a harder position to defend the longer the cheap substitute is live.

Right if: By June 30, 2027, Peacock, HBO Max, or Paramount+ has publicly made its home screen ad placements available to buy through at least one demand-side platform, including via a private marketplace deal open to multiple buyers. Wrong if: On June 30, 2027, all three of Peacock, HBO Max, and Paramount+ still sell home screen ad placements only through their own sales teams with no demand-side platform path.

Bridging The World Of TV And Ad Tech Full Analysis → Listen to the episode →

By December 31, 2027, one of OpenAI, Google, Microsoft, Meta, or Amazon will agree to pay at least $1 billion to settle a US copyright case over content used to train its AI models, confirmed by a settlement filed in US federal court.

Why Anthropic's author settlement put a public number on training data at roughly $3,000 a book, and the licensing deals signed by News Corp, Axel Springer, and the New York Times put a second set of prices on the record. Once a going rate exists, damages stop being guesswork a defendant can argue down and become arithmetic a court can multiply across a corpus, which is what turns a nuisance suit into an existential one. With 80 to 100 active cases and at least one major trial expected in 2027, the deep-pocketed defendants face a choice between writing a very large check and risking a statutory-damages verdict that prices every remaining plaintiff. The alternative path, where these companies grind every case to verdict or settle quietly for small sums, is less likely because a single adverse ruling would set the comparable for all the others, and because the same firms are already paying publishers nine figures voluntarily for licenses.

Right if: A settlement agreement filed in US federal court by December 31, 2027 shows OpenAI, Google, Microsoft, Meta, or Amazon paying $1 billion or more to resolve copyright claims over AI training content. Wrong if: No such filing exists by December 31, 2027, whether because the cases are still running, went to verdict, or settled below $1 billion.

Big Tech’s Litigation Era Listen to the episode →

By May 15, 2027, Taboola or Outbrain will announce a deal to sell all programmatic display advertising on a site ranked in Comscore's top 25 US news properties, matching the arrangement NBC News signed for NBCNews.com and Today.com.

Why NBC News handed exclusive global programmatic display on NBCNews.com and Today.com to Taboola, a company most buyers still think of as a supplier of recommended-links units at the bottom of articles, while keeping its higher-priced video sales direct. That gives every tier-one news site a public template: hand the low-margin auction work to one outside firm, cut the headcount that runs it, and keep the inventory that actually pays. After years of falling open-web display prices, most national news sites run display operations too small to beat a specialist on yield once salaries are counted, and publisher monetization teams copy each other quickly once a peer of NBC News's standing goes first. The opposite outcome requires these publishers to believe their own auction operations out-earn a specialist net of cost, which is exactly the belief the NBC News deal undercut; the real brake is that exclusivity hands one counterparty pricing power at renewal, and that objection tends to lose to an immediate cost saving.

Right if: Taboola or Outbrain publicly announces, on or before May 15, 2027, that it will run all programmatic display sales for a US news site in Comscore's top 25 news properties. Wrong if: No such announcement from Taboola or Outbrain exists on or before May 15, 2027.

MadTech Daily: OpenAI Expands ChatGPT Ads Pilot to 31 European Markets; Apple's Latest Spyware Alert Wave Appears Largest Yet Listen to the episode →

WPP's full-year 2026 revenue excluding pass-through costs, reported in its annual results in February 2027, will be lower than its 2025 figure.

Why Retail media has split consumer-goods budgets between sales and marketing teams, and the internal confusion that creates is now being sold as a consulting product: map the marketing department's workflow, then use the map to show which delays are the agency's fault. A client walking into a renewal with that evidence can cut the agency fee and point to process data rather than to bad work, which gives a chief marketing officer a cost reduction their finance chief can bank this year. WPP is the most exposed of the large holding companies, with a client base weighted toward packaged goods, several quarters of the weakest organic growth in its peer group, and account losses already in the base. The opposite outcome, WPP growing revenue in 2026, would require it to win enough new business to outrun both procurement pressure on fees and the budget that keeps migrating directly to Amazon and Walmart's own ad platforms. Fee cuts are fast and visible; new business wins take a year to show up in revenue.

Right if: WPP's reported full-year 2026 revenue less pass-through costs is below its reported 2025 figure. Wrong if: WPP's reported full-year 2026 revenue less pass-through costs is equal to or above its reported 2025 figure.

The CMO's AI Change Management Problem Full Analysis → Listen to the episode →

Through June 30, 2027, The Trade Desk will not offer advertisers a way to set and pace a budget against a specific window inside a live game, such as the final two minutes of a basketball game or an overtime period.

Why Live sports is the last reliable mass audience in US television, and rights holders are now routing guaranteed upfront commitments into auction-based private deals: Warner Bros. Discovery says these biddable deals are its fastest-growing ad segment and that it has been running the play for three years. The buying software on the other side of that pipe was built to spread a budget smoothly across a day or a week and across millions of small impressions, so when a buzzer-beater sends bid requests spiking into a two-minute window, a normal campaign either spends out early and goes dark by the fourth quarter or misses the moment entirely. Rebuilding pacing logic for a few dozen tentpole events a year is expensive engineering aimed at a small slice of total spend, while the demand-side platforms earn their margin on the high-volume long tail. The practical workaround today costs the platforms nothing to support, because it is a buyer phoning the publisher's yield team during the game. That is why the absence of a real in-game budgeting control is the more likely outcome than a purpose-built one arriving inside the next nine months.

Right if: As of June 30, 2027, The Trade Desk's public product documentation and release notes show no control that lets a buyer set or pace a budget against a defined window inside a live event. Wrong if: The Trade Desk's public product documentation or release notes describe such an in-game budgeting or pacing control on or before June 30, 2027.

Warner Bros. Discovery’s Marisa Crocker: 'It's an attention economy' and live sports is the last great gathering place Full Analysis → Listen to the episode →

PubX, the startup whose AI agents negotiate ad buys directly between advertisers and publishers, will be acquired, shut down, or stop selling itself as a replacement for demand-side and supply-side platforms by June 30, 2027.

Why PubX is live and has proved the technical part works, running agent negotiations over standard bidding protocols inside Prebid and the Google ad server that premium publishers already operate, but the volume is still a few thousand dollars a day across more than 3,000 sites and the payment float is run by a second company, Scope3, because money movement is the unsolved piece. The pitch to publishers is arithmetic: they keep about 41 cents of the advertiser dollar today, and PubX models a 10 to 15% take rate instead of stacked platform fees. The problem is that every piece of the plumbing is open source, so there is no defensible layer to own, and the parts that are not solved, who guarantees the impression was real and who eats an underdelivered campaign, are exactly the parts publishers' finance teams refuse to route real revenue through. Experian shipped agent-connection software for this kind of trading and reports zero usage, which is what it looks like when the rails exist and nobody trusts them with budget. A company with no technical moat, single-digit-thousands of daily volume, and an eighteen-month runway typically gets bought for its team or repositions toward selling tooling to the incumbents it promised to replace, rather than sustaining an independent challenge to them.

Right if: By June 30, 2027, PubX has been acquired, has ceased operations, or no longer publicly markets itself as a replacement for demand-side and supply-side platforms. Wrong if: On June 30, 2027, PubX is still independent, still operating, and still publicly marketing itself as a replacement for demand-side and supply-side platforms.

When Publishers Get AI Agents: Andrew Mole on Agentic Trading and the Future of Media Full Analysis → Listen to the episode →

Paramount will complete its purchase of Warner Bros. Discovery on or before May 31, 2027.

Why California's attorney general walked out of settlement talks with Paramount and publicly called the company a bad-faith actor, and the fight has pushed the deal's timetable into 2027 while 4,500 announced job cuts sit on the ad and programmatic teams. State attorneys general almost never stop a media merger outright; the usual ending is a community-benefit fund, a revised jobs commitment and a podium victory lap, and both companies need the combination badly enough to pay for one. The prize is the only premium video seller with enough scale to package against Disney and Comcast, so the incentive to settle and close runs in one direction. The case for missing the date rests on litigation dragging past the spring selling season or a court imposing conditions Paramount refuses, which would mean a state regulator succeeding where federal challenges to media deals have mostly failed. The likelier cost of this fight is a gutted ad organization arriving with the closed deal, not the deal dying.

Right if: Right if Paramount announces the acquisition of Warner Bros. Discovery has been completed on or before May 31, 2027. Wrong if: Wrong if the acquisition has not been completed as of May 31, 2027, including if it is abandoned or blocked.

Update: California AG accuses Paramount of bad faith in WBD acquisition lawsuit Full Analysis → Read the source story →

AgenticAdvertising.org, the coalition formed to write rules for AI software that buys ads on its own, will have published no technical specification on its public website by May 31, 2027, or will have shut down or merged into another standards body by that date.

Why The coalition entered this fight with a domain name and a grievance against the IAB Tech Lab, which has roughly 650 members and 25 years of running the technical plumbing most programmatic buying sits on. Nothing is forcing a choice: no advertiser is routing meaningful live budget through AI buying agents, so engineering teams at demand-side and supply-side platforms are hedging at the connection layer rather than committing, and a young coalition with no volume behind it has little to keep funders and member engineers in the room. Meanwhile The Trade Desk and Google can define the interface inside their own systems and let everyone else conform, the same way identity resolution settled after the cookie fight, which strips a neutral body of the one thing that would give it leverage. The opposite outcome, a fast public specification from an insurgent group that survives on its own, would require agentic buying to arrive at scale faster than header bidding or OpenRTB ever did.

Right if: Right if, on May 31, 2027, AgenticAdvertising.org's public website carries no technical specification, draft or final, for AI agents buying ads, or the organization has ceased operating independently. Wrong if: Wrong if, on May 31, 2027, an independently operating AgenticAdvertising.org has a technical specification, draft or final, publicly posted on its website.

AgenticAdvertising.org and IAB Tech Lab clash over agentic protocols Full Analysis → Read the source story →

At least one of the six largest agency holding companies — WPP Media, Publicis, Omnicom, Dentsu, Havas or Stagwell — will publicly announce a connected TV campaign bought directly through a supply-side platform's AI briefing tool, with no demand-side platform in the chain, on or before 31 March 2027.

Why PubMatic and the independent agency Butler/Till ran a CTV campaign for Geloso Beverage Group where the brief was typed into an AI assistant and the seller's own platform built and ran the buy, and the two parties reported supply-chain and technology costs down about 80% with 40% more impressions on the same budget. Numbers that size travel fast through client budget reviews, and the holding companies are already defending their fees, so a public pilot is the cheapest way for one of them to look like it is passing savings on. The force pushing the other way is that a holdco's trading desk earns its margin on the demand-side platform layer this setup removes, and handing strategy to a platform that also sells the inventory means the seller picks the media and grades the results with no independent referee. That conflict is why most holdcos will keep the human trader and the DSP in the loop and say nothing publicly. It only takes one to break ranks for competitive optics, and the fee pressure on all six is heavier than the reputational risk of a single named pilot.

Right if: On or before 31 March 2027, one of WPP Media, Publicis, Omnicom, Dentsu, Havas or Stagwell is publicly named — by itself, by its client, or by the vendor — in a connected TV campaign bought through a supply-side platform's AI briefing interface with no demand-side platform involved. Wrong if: No such campaign involving any of those six holding companies is publicly announced by 31 March 2027.

Butler/Till and PubMatic run fully autonomous agentic CTV campaign Full Analysis → Read the source story →

Alphonso, the ad-technology company that sells the advertising on LG's smart TVs, will announce an agreement to be acquired at more than $118 a share on or before February 27, 2027.

Why A court-ordered tender offer priced Alphonso at $118 a share, a number co-founder Ashish Chordia called "incredibly shitty," while a Koch private-equity letter of intent put roughly $200 a share, about $1 billion, on the same business in the same month. Comcast is separately circling a joint venture through its investment arm Atairos, which would put LG's screen-viewing data, Comcast's free streaming service Xumo, and its FreeWheel ad server under one commercial roof. Two credible bidders sitting about 70% above the tender price is a gap that normally gets closed with a signed deal, because a seller who has already called the floor price insulting has no reason to accept it while a control premium is on the table. The confidential IPO filing works mainly as leverage: public investors would have to underwrite $850 million to $1 billion of revenue produced by roughly 100 people without knowing how much of it is outside advertisers versus LG routing its own spend through the pipe, and a private buyer can settle that in diligence while the market discounts it. The $118 price only wins if the Koch bid and the Comcast talks both collapse in the same window.

Right if: Alphonso publicly announces a definitive agreement to be acquired, or a control transaction valuing the company above $118 a share, on or before February 27, 2027. Wrong if: No such agreement is announced by February 27, 2027, including any outcome where Alphonso completes an initial public offering or the $118 tender price stands.

Alphonso Eyes IPO, Comcast JV, or $1B Koch PE Deal Full Analysis → Read the source story →

By June 30, 2027, Walmart will announce that an outside measurement company reports the sales impact of brand advertising campaigns running on its VIZIO smart-TV inventory, rather than Walmart scoring those campaigns itself.

Why Walmart paid about $2.3 billion for VIZIO and is now pitching its retail media arm as a destination for brand and television budgets, not just ads that close a sale, and the VIZIO deal already pulled reported advertising growth down to about 38% from roughly 43% for the standalone business. Ryan Mayward, Walmart's ad sales chief, has said the company is controlling the attribution model in all cases, which is exactly the reason brand planners at the big agencies keep awareness money at YouTube, Roku and The Trade Desk: a seller that grades its own results cannot settle an argument about whether a campaign worked. Retail media budgets were approved in the first place because the payback was checkable against a neutral yardstick, so reclassifying them as brand spend without an independent scorekeeper asks buyers to give up the one thing that got the money released. That leaves Walmart with a choice between keeping measurement control and unlocking the budgets that justify the VIZIO price, and the cheaper concession is the measurement. The alternative, holding the scoring rights and waiting for brand money to arrive anyway, caps the business at the bottom of the funnel and leaves a $2.3 billion television acquisition unpaid for.

Right if: On or before June 30, 2027, Walmart or a named outside measurement firm publicly announces an arrangement under which that firm reports the sales impact of brand campaigns on Walmart's VIZIO television inventory. Wrong if: No such announcement exists by June 30, 2027, or the only outside involvement disclosed is ad delivery, viewability or brand-safety verification rather than reporting sales results.

Walmart Repositions Connect as Full-Funnel 'Growth Platform,' Eyes Brand Budgets Full Analysis → Read the source story →

By 31 December 2027, either Target's Roundel or Kroger Precision Marketing will have announced the purchase of a company whose product lets small businesses buy streaming TV ads themselves.

Why Walmart paid between $1.2 and $1.4 billion for Vibe.co, a platform with roughly 10,000 advertisers, about $130,000 per advertiser, a price that only makes sense as a way to turn its hundreds of thousands of Marketplace sellers into TV advertisers on the same screen where they already buy search ads. Target and Kroger pitch the same small business the same 'one place to buy your ads' story, and neither owns a streaming buying tool; building one from scratch takes longer than buying one, and the Vibe price has already set the going rate for the handful of independent platforms left. Renting the capability from an outside ad-buying firm is cheaper, but it leaves billing, seller identity and purchase-based results reporting outside the retailer's own systems, which is the part that makes the sales pitch work at all. The less likely outcome is that both sit out a full year while Walmart converts its seller base, because the small-seller relationship is where these networks compete hardest and the entry point into television is the most expensive piece to lose.

Right if: Target or Kroger publicly announces, on or before 31 December 2027, an agreement to acquire a company selling self-serve streaming TV ad buying. Wrong if: Neither Target nor Kroger has announced such an acquisition by 31 December 2027.

Walmart Closes $1.2–1.4B Acquisition of Self-Serve CTV Platform Vibe.co Full Analysis → Read the source story →

By June 30, 2027, Walmart will publicly name an outside measurement company allowed to independently check Walmart Connect's attribution results — the figures that tell an advertiser which ads drove store and site sales.

Why Ryan Mayward, Walmart Connect's head of sales, has said Walmart controls the measurement of what worked no matter which pipe the money flows through, including buys placed via Yahoo, Magnite or an outside demand-side platform. Whoever measures the outcome decides what the advertiser's money bought, which is why consumer-goods finance teams treat a seller grading its own sale as a conflict and keep their own read of incremental sales running alongside. Amazon spent years and real money building outside credibility before buyers stopped arguing with its numbers, and Walmart is claiming that endpoint while its data plumbing is still being built. The pull toward opening the books is money: last-click budget will keep flowing on Walmart's own scorecard, but the larger brand and incrementality budgets move only once someone unaffiliated has verified the number, and Walmart wants those budgets. The opposite path — holding the scorecard closed through the 2027 planning cycle — keeps Walmart Connect capped at the dollars it already wins.

Right if: By June 30, 2027, Walmart or Walmart Connect has publicly named a specific independent measurement or accreditation body verifying its attribution or incrementality results for advertisers. Wrong if: No such named independent verifier has been publicly announced by Walmart or Walmart Connect as of June 30, 2027.

Walmart Connect Claims Attribution Control Across All Buying Paths Full Analysis → Read the source story →

By Taboola's fourth-quarter 2027 earnings report in February 2028, none of Disney, Warner Bros. Discovery, Paramount Skydance, or Fox will have announced that Taboola is selling its programmatic display advertising.

Why Taboola now sells all of NBCUniversal's programmatic display inventory on a 65/35 revenue split, which on paper beats the open-market chain where stacked exchange and buying-platform fees take 40 to 50 cents of every dollar before the publisher is paid. The split only matters if Taboola's small-business and native advertiser demand pays premium prices for premium inventory, and the clearing prices under the new arrangement will not be visible for two to three quarters. Large media companies treat control of pricing as the thing that protects their rate card, and letting one outside vendor bid first on everything means trusting that vendor's advertisers to value the inventory correctly. For a rival to copy this before NBCUniversal's numbers are known would mean moving on faith about its own pricing, which is not how these companies behave. The likelier read is that NBCUniversal offloaded leftover display inventory it no longer wanted to staff, and one contract stays one contract.

Right if: As of Taboola's fourth-quarter 2027 earnings report in February 2028, none of Disney, Warner Bros. Discovery, Paramount Skydance, or Fox has publicly announced an arrangement under which Taboola sells its programmatic display advertising. Wrong if: At least one of Disney, Warner Bros. Discovery, Paramount Skydance, or Fox has publicly announced such an arrangement by that report.

Taboola Takes Over NBCUniversal Programmatic Sales Read the source story →

By 15 May 2027, at least one of OpenAI, Anthropic, or Perplexity will publicly announce that it has bought a company whose business is tracking which ads lead to sales.

Why The IAB's framework for measuring ads shown to AI assistants is due 12 November 2026, and it is a voluntary recommendation with no power to make any AI platform open its records to outside auditors. That leaves open the path Facebook took, where the company serving the ad also grades whether it worked and outside vendors are left checking the seller's own numbers. The scarce piece in that setup is the layer that links publisher and advertiser records to an AI platform's logs, and buying an existing tracking firm delivers it in months where building takes years, at prices that are trivial for companies raising tens of billions. Waiting until purchases made through AI assistants are a large share of sales would mean launching an advertising business on numbers advertisers have no reason to believe, which is why the purchase is worth making before the ad revenue arrives rather than after.

Right if: On or before 15 May 2027, OpenAI, Anthropic, or Perplexity publicly announces it has acquired a company whose main business is ad measurement, attribution, or identity matching. Wrong if: No such acquisition has been publicly announced by any of those three companies on or before 15 May 2027.

IAB Developing Framework to Measure Ads Served to AI Agents Full Analysis → Read the source story →

OpenAI will acquire an advertising identity, measurement, or data clean-room company outright by June 30, 2027.

Why OpenAI is selling enterprise advertising features that only large advertisers use, including custom audiences with a 25,000-person minimum and account-health dashboards, while owning none of the underlying plumbing: it has no reliable way to recognize the same person twice across sites and no verified path to prove an ad drove a purchase. Every new ad platform faces that gap, and the usual fix is renting it through a partner or a data clean room, the neutral system where an advertiser and a platform compare customer lists without either handing over raw data. Renting leaves the pricing power and the data rules with the partner, which cuts against OpenAI's core pitch that what people type into a chatbot is a purchase signal nobody else holds. With the cash it has raised and an enterprise ads team it is staffing at roughly $374,000 to $415,000 for a single marketing lead, buying the layer is affordable and faster than building it. The comfortable alternative, a string of partnership press releases, is what advertisers already expect, and it leaves OpenAI renting the one thing that would make its data defensible.

Right if: OpenAI announces or completes the acquisition of a company whose main business is advertising identity resolution, ad measurement, or data clean rooms on or before June 30, 2027. Wrong if: No such acquisition is announced or completed by June 30, 2027, with partnerships, reseller deals, and integrations not counting.

OpenAI Targets Enterprise Ad Teams With New Hires and Platform Features Full Analysis → Read the source story →

At least one US state will sign into law a tax charged on companies' collection or use of consumer data by July 15, 2027.

Why Maryland's digital advertising tax, which charged up to 10% of gross digital ad revenue, was struck down as unconstitutional after Google, Comcast's NBCUniversal and Apple challenged it under the federal ban on taxes that single out online commerce. The budget need that produced that tax has not gone away, and copycat ad-revenue statutes in Illinois, Utah and Washington now sit under the same legal cloud, so the ad-dollar route is closed. A levy on data collection or use is far harder to knock down because it charges for conduct anyone can engage in, and Illinois already runs a biometric privacy law that gives drafters a working template. The cost would land on identity, clean-room and data-broker businesses such as LiveRamp rather than on the sell-side platforms that just escaped. The alternative, that legislatures write nothing new after one courtroom loss, would be out of character for states hunting recurring revenue from large technology firms.

Right if: A governor signs a bill into law that imposes a tax or fee on the collection or use of consumer data by commercial entities, on or before July 15, 2027, as recorded in the Tax Foundation's state tax legislation coverage. Wrong if: No such law is signed by July 15, 2027, including cases where data-tax bills are introduced or pass one chamber but are not enacted.

Google, Peacock, Apple Kill Maryland Digital Ad Tax; Other States at Risk Full Analysis → Read the source story →

S4 Capital, the digital advertising group founded by Martin Sorrell, will cease to be an independent listed company by 31 December 2027, through a takeover, a take-private, or a merger.

Why Three founders who sold their agencies to S4 Capital are suing over acquisition payments the group is withholding, and S4 asked the court to seal the dollar figures rather than publish numbers that would clear it. Those deferred payments come from deals struck at 2021 prices that the acquired businesses have since undershot, so the group is choosing between paying sums the units no longer justify and fighting the people whose client relationships it bought. Fighting them is the cheaper choice on cash and the more expensive one on revenue, because agency leaders mid-payout leave and take accounts with them, which shrinks the business further against the same obligations. The shares have already repriced hard from their 2021 level, which makes the units cheap for a cash-rich buyer or a private equity bidder who can settle the founders in cash rather than promises. The alternative, that S4 trades its way out alone, requires the acquired units to stabilise while their leadership is actively litigating with the parent.

Right if: A regulatory announcement to the London Stock Exchange on or before 31 December 2027 discloses an agreed offer, scheme of arrangement, or merger under which S4 Capital ceases to be an independent listed company. Wrong if: S4 Capital is still trading as an independent listed company with no such agreed transaction announced as of 31 December 2027.

S4 Capital Sued by Executives Over Withheld Acquisition Payouts Full Analysis → Read the source story →

By April 30, 2027, one of WPP, Omnicom, or Dentsu will pull or suspend its recommendation of The Trade Desk for client spending.

Why Publicis commissioned a forensic audit of The Trade Desk, told clients to hold off on the platform, and on June 12, 2026 the two sides issued a joint statement saying the issues had been "addressed"; The Trade Desk's shares rose about 5% on the news and not one finding, remedy or pricing term was made public. The lever worked: the largest agency holding company on earth forced its biggest independent buying platform back to the table and took private concessions without losing the relationship or the client spend. The Trade Desk is a shared vendor for WPP, Omnicom and Dentsu, all of them cutting costs and reopening vendor terms, so the same move is available to each at almost no cost beyond the audit fee. The opposite outcome asks three margin-pressured procurement desks to watch a rival extract concessions in public and decline to copy it for a full year, which is not how holding-company procurement has ever behaved with a shared supplier.

Right if: By April 30, 2027, a statement from WPP, Omnicom, or Dentsu, or a trade-press report of that company's client guidance, shows it has pulled or suspended its recommendation of The Trade Desk. Wrong if: None of WPP, Omnicom, or Dentsu has pulled or suspended its recommendation of The Trade Desk as of April 30, 2027.

Publicis and The Trade Desk Quietly Settled Audit Dispute in June Full Analysis → Read the source story →

By the end of July 2027, Omnicom will have announced a further transfer of its own technology staff to Endava or another outside contractor, on top of the 468 Omni platform engineers moved in 2026.

Why Omnicom justified its $13.5 billion purchase of IPG partly on Omni, its in-house data and audience platform, then seven months after closing moved at least 468 of the engineers who build and run Omni onto the payroll of Endava, a publicly listed IT contractor with roughly 11,000 engineers. Once the build-and-run team sits with a vendor, the cost and political difficulty of moving adjacent teams drops, and the $750 million-plus cost-saving target Omnicom promised investors keeps the pressure on fixed engineering headcount all through 2027. Endava's contract is non-exclusive and its scale makes absorbing another few hundred people routine, so the path of least resistance runs in one direction. The opposite outcome, Omnicom halting or reversing the shift, would mean rehiring at higher cost the people who hold the undocumented knowledge it just handed away, and abandoning the savings story the market has already priced.

Right if: Omnicom or Endava publicly announces a further move of Omnicom technology or platform staff to an outside contractor on or before 31 July 2027. Wrong if: No such further transfer has been publicly announced by either company as of 31 July 2027.

Omnicom Transfers 468+ Omni Platform Engineers to Contractor Endava Full Analysis → Read the source story →

By December 31, 2027, a court will let the Texas attorney general's lawsuit accusing Netflix of designing its service to hook children proceed past Netflix's bid to have it dismissed under Section 230, the 1996 law that shields platforms from liability for what their users post.

Why State attorneys general are seeking roughly $1.4 trillion from Meta in the Oakland addiction trial on a theory that sidesteps Section 230 entirely: they are attacking engineering choices such as autoplay, infinite scroll and notification cadence, which are nobody's speech, instead of anything users posted. Texas Attorney General Ken Paxton has copied that theory onto Netflix, and a second court accepting it turns one lawsuit into a doctrine that reaches every business selling attention, from streamers to the buying platforms that optimize for time on surface. Courts have already shown willingness to treat a recommendation surface as a product that can be defectively designed, which gives the Texas court a well-marked path to follow. A clean Section 230 dismissal would require treating autoplay defaults and feed engineering as publishing decisions, a reading that has been losing ground for several years. The main way this fails is timing rather than merits: dismissal briefing can slip past the deadline, and a settlement or withdrawal would end the case before any ruling.

Right if: By December 31, 2027, the court denies Netflix's Section 230 dismissal at least in part and allows the design claims to go forward. Wrong if: The court dismisses those claims on Section 230 grounds, the case ends by settlement or withdrawal, or no such ruling has issued by December 31, 2027.

Social Media Addiction Trial Against Meta Begins; $1.4 Trillion at Stake Read the source story →

Paramount Skydance will complete its purchase of Warner Bros. Discovery on or before March 31, 2027.

Why Paramount Skydance owes its lenders roughly $650 million for every quarter the Warner Bros. Discovery purchase stays open, a penalty that started on September 30, 2026 and is now running past the deal's original outside date. That meter makes Paramount a buyer willing to trade promises about its future conduct for speed, and the 12-state antitrust case led by California Attorney General Rob Bonta rests on theatrical release windows and cable carriage, two sources of leverage that shrink every year while Netflix and Amazon reshape the market without asking permission. The states can bank announced concessions now or spend years litigating over declining assets, which is why a negotiated exit is more likely than a verdict. Outright collapse would push Warner's assets into a distressed break-up that neither the buyer, the lenders, nor the studios' employees want. The genuine uncertainty is timing: whether the parties close before Paramount eats two more quarters of fees.

Right if: Paramount Skydance publicly announces or files with the SEC that the acquisition of Warner Bros. Discovery has closed, dated on or before March 31, 2027. Wrong if: As of April 1, 2027 the acquisition has not closed, for any reason including a court block, an abandoned deal, or continued litigation.

Paramount-Warner Bros. Merger Faces March Trial, Settlement Likely Read the source story →

OpenAI will open ChatGPT advertising to self-serve buying, letting an advertiser create and run a campaign without speaking to a salesperson, by 30 June 2027.

Why OpenAI has pushed ads inside ChatGPT into 31 European markets and says ad revenue grew more than 25% since early August, but every one of those dollars is hand-sold as a direct deal negotiated by a person. Hand-selling caps revenue at the size of the sales team and the CPMs brand buyers will pay on faith, which is the wrong shape for a company that needs advertising to help cover enormous compute costs. Google, Meta, and Amazon each moved to self-serve buying within a couple of years of selling their first ad, because an auction open to small and mid-sized advertisers is where the volume and the margin live. The case against is that OpenAI cannot yet offer verified reach or a clean consent trail for targeting in Europe, which stalls large-brand procurement, but self-serve is aimed at the long tail of advertisers who buy on their own dashboards without demanding outside verification, so the measurement gap slows big budgets more than it slows the product.

Right if: By 30 June 2027 an advertiser can register with OpenAI and launch a paid ChatGPT ad campaign through an OpenAI interface without contacting a salesperson. Wrong if: On 30 June 2027 buying ads in ChatGPT still requires going through an OpenAI salesperson or a managed-service arrangement.

OpenAI Expands ChatGPT Ad Pilot to 31 European Markets Full Analysis → Read the source story →

Before the 2027 upfront selling season closes on June 30, 2027, a top-20 US advertiser will publicly say it pulled sports money out of a locked-in upfront commitment and spent it instead on slots bought inside live games through automated auctions.

Why Warner Bros. Discovery's automated ad revenue on live sports grew 174% between the 2025 and 2026 March Madness tournaments after it began swapping targeted ads into the live stream in real time, and the tournament drew 50 advertisers that had never bought its sports inventory before. Those buyers had been priced out by the high floors of directly sold sports packages, so the auction route is the only way performance-driven budgets reach a live mass audience. Every other major sports seller already streams live games through the same ad plumbing, so the capability itself is spreading regardless of who announces what; the real contest is whether guaranteed upfront dollars start leaving the handshake deals. The comfortable outcome is that everyone praises automated sports buying while nobody admits to moving a committed dollar, because agencies protect negotiated guarantees and sellers punish defectors with worse inventory. That holds only until one large advertiser finds auction-bought sports cheaper than its upfront rate and decides saying so publicly is worth more leverage than silence, which is the likelier break given proven delivery in the highest-pressure inventory in television.

Right if: By June 30, 2027, a top-20 US advertiser by measured ad spend, or its media agency speaking for it, states on the record in Ad Age that it shifted sports budget out of an upfront commitment into automated live-game buying. Wrong if: No such on-the-record statement from a top-20 US advertiser or its media agency appears in Ad Age by June 30, 2027.

WBD Programmatic Live Sports Revenue Grows 174% Year-Over-Year Read the source story →

VideoAmp will be acquired, merged into another company, or shut down by June 30, 2027, ending its run as an independent measurement company.

Why VideoAmp says it ran $6 billion in "currency and measurement" deals last year, a phrase that can cover pilots and influenced budgets as easily as money that actually changed hands; an industry expert who says he watched the wording get drafted puts the cleared figure closer to $2 billion, and neither VideoAmp nor the buyers and sellers it names answered 13 written questions about it. The company's entire pitch was to replace Nielsen as the number networks and agencies transact on, and that job runs on trust in the transacted total, so a threefold gap between the promoted number and the defensible one raises the price of the next funding round at exactly the moment procurement and legal teams are asking vendors to show their methodology. Accredited rivals such as Nielsen, Comscore and iSpot win this argument by standing still, which slows VideoAmp's commercial momentum while its cash need continues. Staying independent through the 2027 upfronts would require either a clean audited number that puts the $2 billion estimate to bed or a fresh round raised on a wounded story, and the company has declined the cheaper of those two options.

Right if: VideoAmp announces a sale, a merger, a controlling investment that hands ownership to another company, or a wind-down of operations on or before June 30, 2027. Wrong if: VideoAmp is still operating as an independent, unacquired company on July 1, 2027.

VideoAmp's $6B 'Currency' Claim Cannot Be Verified by Reporters Full Analysis → Read the source story →

By June 30, 2027, at least one of the New York Times, News Corp, or the Washington Post will publicly confirm a signed agreement with Apple to be paid for news content used in Siri's AI-generated answers.

Why Apple is reported to be negotiating multiyear deals that pay publishers based on how often Siri draws on their articles, and the New York Times is named among the targets. Apple's motive is not the journalism itself but a legally clean body of licensed text that Google and OpenAI cannot copy without inviting lawsuits, and Apple has both the cash and the shipping deadline of a rebuilt Siri pushing it to close. For a publisher, a per-use royalty costs nothing up front and converts traffic many of them already expect to lose into a payment, which makes signing easier than holding out. The main thing that could block a deal is the measurement problem, since only Apple can count Siri queries, but that argument stalls terms rather than killing a marquee agreement, and Apple can buy past it by offering a guaranteed minimum.

Right if: On or before June 30, 2027, the New York Times, News Corp, or the Washington Post states publicly, or Apple states publicly, that a content licensing agreement covering Siri answers has been signed with that publisher. Wrong if: As of June 30, 2027, none of those three publishers has a publicly confirmed signed Apple agreement covering Siri answers.

Apple in Talks With Publishers on AI Siri Content Licensing Deals Full Analysis → Read the source story →