The week of 20 September 2026

00The week in one paragraph

This was the week Google’s ad tech ruling stopped being a number and became a calendar. The opinion unsealed September 16 added a six-year compliance monitor with source-code access, a “functionally equivalent” bar for Prebid integration, and global scope — details that outlast the 12-and-15-month build clock the market already knew, while a separate breakdown showed those remedies cover only a shrinking slice of the auction, leaving DV360, CTV, in-app and retail media untouched. The FTC, watching the same ruling as the precedent for its own case against Amazon’s hidden reserve price, got a pointed warning from France’s 2021 fine on Google: a behavioral remedy can look like progress and leave market structure unchanged five years later. Google spent the same week quietly testing a much smaller mechanism on publishers directly — an opaque, usage-based AI licensing payout two of its own program participants call a black box — while Chrome shipped four new ad-density metrics with no benchmark attached yet, and one InMarket engineer argued most of the industry still isn’t validating the very supply-chain object that proves an impression’s real route. CTV spent the week arguing with itself: nine executives split on whether buyers or sellers hold pricing power, one interoperability pitch against forcing linear and CTV onto a single system, and a new IAB figure showing 43% of buyers still don’t trust the inventory they’re buying — blocked less by bad data than by a privacy law and a deal structure never built for single-show targeting. Retail media, meanwhile, is being asked to prove something harder than attribution to keep growing into brand budgets, and gaming’s own trade body spent its biggest showcase of the year arguing it has to look exactly like every other channel before it’s allowed to look different.

31The Google Ad Tech Opinion Is Public. Four Details Sit Beyond The 12-And-15-Month Timeline

Judge Leonie Brinkema’s ad tech remedies opinion, filed under seal September 2, went public unredacted on September 16 after both sides let their 14-day window to flag confidential material pass without objection. The unsealed text adds detail the 12-and-15-month build clock didn’t carry: AdX and DFP must connect to Prebid through integrations the court calls “functionally equivalent,” not merely present; AdX owes rival ad servers the same real-time bid terms DFP gets, and Google must share win/loss bid data with publishers and document how DFP picks a winner. A court-appointed monitor gets six years — down from the 15 the DOJ requested — with full access to Google’s staff, systems and source code, and the order applies worldwide, though DV360 stays outside its scope entirely. Both sides owe proposed final judgments by October 2, and the remedies themselves take effect 60 days after Brinkema signs the order — both dates ahead of the multi-year build work.

29Google’s Ad Tech Remedies, Decoded: What Changes In 12 Months, What Takes 15, And What The Court Left Alone

Google has 12 months to open AdX to competing ad servers on non-preferential terms and 12 to 15 months to wire AdX and DFP into Prebid — the first remedies in the case with an actual build-and-test period attached rather than an open-ended mandate. What the clock leaves alone matters just as much: the court found insufficient evidence to regulate DV360, and the remedies apply only to the open-web display auction, not CTV, in-app or retail media — the three categories where ad tech volume has actually been moving, in a segment where AdWords’ own share of traditional open-web display fell from more than 40% in 2019 to 11% by 2025. Enforcement runs on a quarterly monitor-report cadence with a response window of more than a month before a judge can act on a flagged violation — what privacy expert Alan Chapell called Google being “brilliant at playing the behavioral remedies game,” warning publishers to expect “breadcrumbs,” not structural change. The court also flagged Google Partner Bidding (gBid Direct), which lets Google bid directly into publisher in-app auctions bypassing an exchange entirely, as a template Google could extend to open-web display once the new requirements bite.

22Google’s Ad Tech Antitrust Cases Are Winding Down. CTV and AI Are Next.

Google has now been ruled a monopolist in two major U.S. antitrust cases inside twelve months — general search and ad tech — and in both, judges chose behavioral remedies over divesting Chrome, Google Ad Manager or AdX. The ad tech trial record surfaced internal programs, including “Red State” (aimed at suppressing rival exchanges’ bid prices) and “Jedi Blue” (a quid pro quo giving Meta preferential auction treatment), that shaped the monopoly finding even without a breakup. With the structural fight largely settled, regulators are turning toward where Google is expanding fastest: AI Max has moved from search into shopping and travel, and Performance Max already spans search, YouTube, display and shopping in one campaign, both riding the same auction infrastructure the courts just examined. The European Commission is pursuing a separate case targeting Google’s sell-side ad tech stack, and the U.K.’s competition authority has proposed conduct requirements that include an opt-out for AI Overviews — a direct answer to publisher complaints about lost traffic.

23What the FTC Can Learn From Google’s Ad Tech Case as It Pursues Amazon

Two days after the FTC filed its soft-reserve-price complaint against Amazon on September 1 — alleging a hidden floor that ran for seven-plus years — Judge Brinkema handed down her Google remedies and declined to break the company up, opting instead for limits on self-preferencing, real-time auction-data sharing with publishers, and Prebid integration. That ruling, not the Amazon complaint itself, is now the FTC’s operating precedent, and the industry reaction is a warning worth reading: France’s competition authority fined Google €220 million in 2021 over the same kind of self-preferencing conduct, and five years later market dynamics were largely unchanged. Equativ CEO Arnaud Créput: “the market structure has fundamentally not changed. Google remains dominant, and independent ad tech companies still compete within an ecosystem where Google controls critical infrastructure.” Kevel CEO James Avery’s framing: “Proving the harm is the easy part; fixing it without breaking something else is the hard part.”

21Google’s New AI Licensing Pilot Pays Publishers by the Use. Most Say It’s a Black Box.

Google is quietly scaling an “AI contribution pilot” that pays publishers a usage-based fee — not a flat license — through a Search Console widget, for content pulled into Gemini, AI Overviews and AI mode; there’s no minimum spend and publishers can opt out at any time. The pitch targets small and mid-sized publishers rather than the largest newsrooms, who typically negotiate direct licensing deals instead, and Google has approached at least dozens of publishers so far. The widget shows monthly earnings but not the methodology behind them. “It’s quite black box,” one publisher executive in the program told Digiday; another said flatly, “Do I wish they were more transparent? Definitely.” For scale, Google’s separate News AI licensing scheme already covers more than 200 titles globally, and Google News Showcase covers more than 2,800 publications across 33 countries — the AI contribution pilot is smaller and newer than both.

25Chrome Has A New Way To Measure Ad Overload On The Web

Chrome now reports four new ad-experience metrics inside the Chrome User Experience Report: ad count, ad density, ad weight by network usage, and ad weight by CPU usage — all running on the same real-user measurement pipeline that already underlies Core Web Vitals, so no new data collection was required and publishers can pull the numbers today rather than waiting on a rollout. Chrome group product manager Alex Cone said there are “no plans” for benchmarks at launch: “We want to better understand if [the metrics] are valuable for making decisions.” WPP, The Guardian, Mediavine, Index Exchange and Adelaide have publicly backed the initiative, and Google’s own DV360 already supports it, though Cone declined to say whether other DSPs or SSPs have committed to using the signals.

26Why SChain Validation Deserves More Attention Than It Gets

SChain, the OpenRTB supply chain object attached to every bid request, records every entity that touched an impression on its way to auction — closing a gap ads.txt never did: authorizing who’s allowed to sell inventory says nothing about which path an impression actually took through resold and rebrokered supply. Per InMarket engineering manager Piyush Deolikar, most of the industry still isn’t validating it, leaving room for fabricated hops whose asi/sid pairs don’t exist in any seller’s sellers.json file, payment-handling nodes missing from a publisher’s own ads.txt, and spoofed domains carrying bundle IDs that can’t be resolved. Deolikar’s guidance: exchanges should validate SChain before sending outbound bid requests, SSPs should insert their own node and keep sellers.json synchronized, and DSPs should discount or drop unverifiable chains — what he calls “the cheapest, highest-leverage fraud filter available.”

28Linear and CTV Don’t Need One System. They Need One Layer Between Them.

Converging linear TV and CTV doesn’t require replacing either system, argues Virtual Minds’ Max Deyerl heading into DMEXCO — it requires an interoperability layer that connects the two while keeping each side’s domain-specific tooling intact. “Many broadcasters, content providers, and inventory holders are still operating in different silos,” he said, despite presenting a unified viewing experience to audiences; the handover cost between systems doesn’t just slow campaigns, it excludes digital-first buyers from linear inventory they’d otherwise use. Austria already ran a version of the fix: its broadcasters collectively moved linear TV to digital buying using real-time HbbTV signals and a recalculated CPM-based currency, Teletest 2.0, that eliminated the “zero ratings” problem of measuring linear the old way in a converged market. Deyerl’s warning on AI: “AI only creates operational excellence if you have the right data, and if you already have the silos combined.”

24Is CTV a Buyer’s Market or a Seller’s Market?

Nine industry leaders surveyed on where CTV pricing power sits today mostly land on the same split verdict. dentsu UK&I’s Alexandra Sturgess: “CTV is increasingly shifting from a seller’s market to a buyer’s market,” driven by ecosystem diversification and programmatic platforms that let buyers optimize in real time. Mindshare’s Dave Sargent calls it “Both”; Titan OS’s Joe Evea says a buyer’s market “but only in aggregate,” since TV-OS gatekeeping still creates distinct, defensible opportunities. BBC Studios’ Lindsay Turner: overall inventory is abundant, but “premium attention is still a seller’s market” — distinctive, fan-driven content remains genuinely scarce. Bloomberg Media’s David Bradford calls the 2026 landscape “quite clear” as a buyer’s market overall, though premium sellers keep the advantage when they can deliver genuinely engaged audiences.

30Retail Media’s Measurement Bar Just Moved From Attribution To Incrementality

Retail media’s measurement bar is moving from attribution — tracing an ad exposure to a transaction — to incrementality: whether that transaction would have happened without the ad. DoorDash Ads GM Katie Daleo made the case at Ascendant Network’s inaugural Retail + Commerce Media Upfront, citing DoorDash’s own numbers: two-thirds of its users are undecided about a purchase when they open the app, and in one beverage-brand campaign, 56% of buyers reached through DoorDash advertising were net-new customers the brand hadn’t already acquired in-store — figures DoorDash and research partner Circana are using to argue the channel earns credit beyond last-touch attribution (vendor-supplied; DoorDash is the platform making its own case). The Alliance for Audited Media was named as a body that could independently certify retail media measurement methodologies — the kind of third-party validation attribution-only claims never needed because the claim itself was narrower.

32CTV Buyers Have Show-Level Data Now. VPPA And Deal Structure Still Block Show-Level Buying.

A new IAB report found 43% of CTV buyers have little or no confidence in the inventory they’re actually buying — but Tatari’s Tom Discepola argues the fix isn’t more reporting, it’s the plumbing underneath it. Two structural walls block show-level buying regardless of data quality: the Video Privacy Protection Act, since combining show-level viewing with household IDs creates real legal exposure that linear TV’s anonymous program logs never carried; and deal structure, since PMP-based optimization — the workhorse of performance CTV buying — has no show-level lever to pull without a programmatic-guaranteed deal or sponsorship. One top-five streamer told Tatari its SSP integration can’t pull show-level reporting at all; a second doesn’t broadly share it; a third offers only unranked top-25/50/100 lists its own team calls “directional at best.”

33Gaming’s Advertisers Pitch: Prove Parity First, Make The Differentiation Case Later

Gaming draws 2.4% of media spend even though 67% of Americans play video games every week — “the single biggest mismatch between attention and investment in media today,” per Overwolf CEO Shahar Sorek, speaking at the fifth annual IAB PlayFronts. IAB CEO David Cohen told the event, “We need to resist the temptation to take the models of yesterday and simply put them into the environments of tomorrow” — then PlayFronts spent most of its floor time showing tools that do exactly that: Overwolf’s Gamer Grid identity platform and its new In-Game Audience Map, spanning more than 5,000 titles, are pitched at agency budgets rather than a separate gaming line item. The proof case buyers cite is rewarded video: Sam’s Club’s Roblox campaign drew 5 million engaged impressions and 1 million virtual-card claims, with an 18% sales lift and a 6% conversion lift per cardholder — a format WPP Media’s Frank Puma said went from “a joke” on the buy side to credible once premium platforms delivered results like that.