00The week in one paragraph
This was the week the leverage turned. Last week the story was that the numbers had stopped adding up — supply contracting, measurement failing, publishers building their own meters. This week publishers, regulators and courts all reached for the switch. USA Today Inc. said it is prepared to delist from Google Search within six to twelve months, People Inc. said blocking Google is “100% on the table,” and Reddit is reportedly weighing cutting Google off from a ~$60M-a-year data deal — a threat that only exists because USA Today has already signed licensing agreements with Meta, Microsoft and Amazon. The regulators moved in the same direction: the European Commission fined Google €890M under the DMA and ordered structural changes to how Search presents its own services, eight days after the July 16 specification decisions; a bipartisan bill would force AI crawlers to identify themselves, with FTC enforcement, against a backdrop where 57.4% of all web requests are now automated; and children’s privacy quietly became the strictest regime in digital advertising, applying by audience rather than by intent. A federal judge halted the $110B Paramount–Warner Bros. Discovery merger two days before closing — brought by 12 state attorneys general, not by any federal agency. Meanwhile Alphabet posted a $119.8B quarter in which the Google Network segment — the open web’s line item — declined, and advertising drew zero questions in investor Q&A against 28 mentions of TPUs. Where publishers built rather than blocked, the news was better: the IAB Tech Lab finalised programmatic signalling for pause, menu and four other CTV formats, beehiiv launched a programmatic layer for newsletters on an SSP it built itself, and Paramount decided prestige drama was worth trading for an email address. One thread ties it together: for the first time in a decade, the party with something to withhold is not always the platform.
01Once unimaginable — publishers are preparing to opt out of Google Search
USA Today Inc. CEO Mike Reed told Adweek the company is prepared to delist from Google within six to twelve months, having already signed licensing deals with Meta, Microsoft and Amazon. People Inc. CEO Neil Vogel told the WSJ that blocking Google entirely is “100% on the table.” Reddit is reportedly weighing cutting Google’s access to its content for AI training, straining a roughly $60 million-a-year agreement; Reddit’s stock fell 9% on the report. Reed: “Enough is enough.” The mechanism forcing the choice is that Google uses one crawler for both Search indexing and AI training — allow both, or disappear from Search. Google Extended exists as a nominal opt-out that publishers do not trust. The traffic math now justifies the threat: only about 25% of Google AI-mode sessions end in a click (Growth Memo, via NYT), Wikipedia’s human traffic is down 8% YoY, and Newsweek fell from ~100M readers in May 2025 to 23M a year later. The Verge’s Nilay Patel: “For publishers, Google Zero is already here.” The tooling arrived in parallel — beehiiv partnered with Cloudflare to let creators block Google’s crawler, and Cloudflare defaults new and free-tier sites to blocking multi-purpose crawlers from September 15.
Publisher takeaway
The headline is not “block Google” — it’s that USA Today can credibly threaten to, because three non-Google licensing deals are already signed. Leverage is downstream of diversification, and every publisher’s version of this starts with the same two pieces of work: knowing your Google dependency by revenue per session rather than traffic share, and having at least one non-Google licensing relationship, however small, to establish a price. The second thing to get right is technical: build per-crawler control so blocking is a dial you can turn selectively, in stages, rather than a single switch. Publishers who can throttle will negotiate better than publishers with a binary choice. And note the risk nobody is discussing — if Google eventually pays for usage, publishers who blocked in the interim may have less measured usage to invoice for. Turning the tap off can turn the meter off. Nobody has actually pulled the plug this week. The credible threat is the asset, and it is worth building deliberately.
02The EU fines Google €890M and orders structural change to Search, ads and Play
On July 24 the European Commission fined Google €890 million (~$1.01B) for two DMA breaches: €460M for self-preferencing its own shopping ads, hotel booking, transport and sports results over competing third-party services in search rankings, and €430M for Play Store anti-steering — preventing developers from “freely communicating, prompting, or linking users to less expensive purchase alternatives outside of Google Play.” The decision requires structural changes across advertising, search and app ecosystems. Google has indicated it will work to implement changes to how it presents its own services, shopping ads and content-related services. Google was designated a DMA gatekeeper in 2023. This lands eight days after the Commission’s July 16 specification decisions ordering Google to share Search data with rivals and open Android to competing AI assistants.
Publisher takeaway
The fine is a rounding error — €890M against a $119.8B quarter (see #3). The remedy is not, and the verticals named in it are publisher verticals: comparison shopping, travel, transport and sports are precisely where publishers built commerce and affiliate businesses that Google’s own modules outranked. This is the first regulatory remedy in years that maps onto a specific publisher revenue line. Three practical moves follow. First, instrument your EU search performance by vertical now — you need a clean before-and-after baseline, because when Google ships the remedy the publishers who can demonstrate a measurable change will be the ones regulators and buyers listen to. Second, re-model the verticals you abandoned; if self-preferencing is genuinely unwound in the EU, some unit economics you wrote off in 2022 change. Third, feed documented harm to your trade body — DMA enforcement runs on evidence from affected parties, and contributing costs an analyst’s afternoon while shaping the remedy you will operate under for years. Watch what Google ships in Europe, not what it pays.
03Alphabet’s record quarter — and the shrinking open web inside it
Alphabet reported Q2 2026 revenue of $119.8 billion, up from $96.4B a year earlier. YouTube advertising $11.1B (from $9.8B); Google Cloud up 82% to $24.8B. But Google Network — AdSense, AdMob and Google Ad Manager, the segment where Google sells other people’s inventory — declined year over year, in a quarter the company grew 24%. Per AdExchanger’s count of the earnings call, advertising-related words appeared 21 times in prepared remarks and zero times in investor Q&A; “TPUs” appeared 28 times. Where advertising did come up, it was as an AI story: CFO Anat Ashkenazi said AI Max “has become the core building block for advertisers to fully participate in our new AI experiences,” citing average improvements of ~50% in conversions or ROAS (Google’s own figure).
Publisher takeaway
Read the three facts together — record ad revenue, a contracting third-party network, and leadership spending its most scrutinised hour on silicon. This is not Google exiting advertising; it is Google’s growth relocating to surfaces it owns, sold through automated products where the buyer selects an outcome rather than a placement. AI Max does not need a publisher network to hit a conversion target, and the Network segment is the part of Google’s business that requires the open web to exist. Two consequences. Concentration risk has a new dimension: if a meaningful share of your programmatic revenue arrives through Google Network products, you are attached to a declining segment inside a growing company — and declining segments do not get the roadmap, the headcount or the pricing defence. Report it as its own board-level line, with a quarterly reduction target if it’s above 30%. And compete where optimisers can’t reach: guaranteed placements, sponsorships, editorial integrations and first-party audience products are bought by humans making choices, which is exactly the transaction automated buying is designed to remove. Hostility you can negotiate with; indifference from your largest revenue partner you cannot.
04A federal judge halted the $110B Paramount–Warner Bros. Discovery merger
On Monday, July 20, US District Judge Araceli Martínez-Olguín in Oakland granted a temporary restraining order blocking Paramount Skydance’s $110 billion acquisition of Warner Bros. Discovery, finding it “likely” violates antitrust law. The companies had expected to close July 22. The order was extended Thursday — the deal is paused until August 17, with a preliminary-injunction hearing on August 3. The action came from 12 state attorneys general led by California AG Rob Bonta, filed July 13 under Section 7 of the Clayton Act. The Writers Guild of America filed separately over wages and job losses.
Publisher takeaway
Every H2 forecast that assumed post-merger consolidation now has a live branch that has to be modelled both ways — upfront volume, the number of distinct buyers you negotiate with, and whether specific rights packages come to market at all. Do that before August 3, not after it. Operationally, the pause means two ad stacks, two sales organisations and two sets of programmatic decisions for at least another quarter — WBD handed its stack to AWS earlier this month while Paramount rebuilds its own, and neither integration is now settled. Add counterparty-change language to anything you’re signing that presumes the merged entity. But the structural lesson is the one to carry forward: a coalition of state attorneys general caught a signed $110 billion transaction 48 hours before closing — not the FTC, not the DOJ. The same multistate coalitions are writing minors’ privacy and data-sales law (see #7). If nobody at your company tracks state AG activity as a live channel, this week priced that gap.
05Congress moves on stealth AI crawlers
On July 23, Reps. Laurel Lee (FL-15) and Valerie Foushee (NC-4) introduced the bipartisan Stealth Bot Prohibition Act, requiring AI-powered web crawlers to disclose their identity and purpose when accessing a site, with FTC enforcement against bots that conceal what they are. The News/Media Alliance estimates deceptive bots exceed 50% of traffic on some publisher sites; CEO Danielle Coffey called the bill a “sorely-needed, common-sense solution.” Independent data supports the scale: automated systems now generate 57.4% of web requests versus 42.6% from humans, AI agent traffic grew 45% in Q2 2026 to 17.7 billion requests (from 12.2B in Q1), Meta’s training crawler grew 74% QoQ and its RAG crawler 163%. And the enforcement gap is measurable: 56.4% of news publishers block at least one AI crawler and 50.5% ban GPTBot — yet 39.5% of sites blocking GPTBot still serve it content.
Publisher takeaway
Identification is the precondition for every other remedy you want. Licensing, pay-per-crawl, telemetry standards and rate cards all assume you can tell which agent took your content; without mandated disclosure, all of them remain voluntary on the crawler’s side. That is why this narrow bill matters more than its scope suggests — and why the FTC deception hook is smart: misrepresenting your identity to gain access is well-trodden territory, a far shorter path than the training-data copyright cases grinding through the courts. It doesn’t resolve fair use; it just makes hiding illegal. Two things to do regardless of whether the bill advances. Reconcile what you block against what actually gets through — compare robots.txt and WAF rules to server logs by user agent and ASN; if you’re in the 39.5% still serving a crawler you formally block, that’s a fixable engineering problem this quarter. And separate bot from human traffic in every number you publish, because if more than half of requests are automated, every audience figure you give a buyer is contaminated by an unknown amount. Buyers will start asking. Have the answer built.
06The W3C Attribution API — and who ends up owning web measurement
A new draft of the W3C Attribution API published in July 2026 moved into wider industry review. It is browser-mediated measurement: the browser records ad exposure and conversion and reports an aggregated, privacy-protected result, rather than an ad tech vendor stitching a journey across sites. It is being developed in the W3C’s Private Advertising Technology (PAT) working group with Apple, Google, Microsoft and Mozilla participating, and is co-chaired by Aram Zucker-Scharff of the Washington Post, who stresses the work predates Privacy Sandbox and runs transparently. The objections: James Rosewell (Movement for an Open Web) warns it “could concentrate too much influence over measurement within browser platforms”; Don Marti questions the architectural resemblance to Google’s earlier Attribution Reporting API; and there is an open governance debate over whether advertisers, publishers, agencies and independent measurement providers have adequate say. Scott Messer makes the counter-case: attribution on the open web is what drives advertising investment toward it.
Publisher takeaway
Whoever defines what counts as a conversion, at what aggregation threshold and with what reporting delay, is defining the pricing engine for the open web. The specific risk for publishers is under-discussed: privacy-preserving aggregation suppresses results below a minimum count, so a publisher with modest volume per campaign may generate no reportable conversions at all — not zero performance, but no data. On a media plan, “unmeasurable” and “ineffective” are the same row. That is the evidenced objection worth filing, and it is filable: take your typical campaign sizes, work out how many would clear a plausible aggregation floor, and submit that analysis as a named publisher. Unlike Privacy Sandbox, this is a multi-vendor standards process with a public draft and a publisher co-chairing the working group — the procedural door is genuinely open, which is rare enough that not walking through it would be a choice. Meanwhile, keep building the first-party measurement you’ll own regardless: logged-in data, on-site conversion tracking and direct advertiser integrations don’t require a browser’s permission.
07Children’s privacy comes for everyone — by audience, not by intent
The FTC’s updated COPPA Rule reached its enforcement deadline on April 22, 2026. Federal liability remains “actual knowledge” that a user is under 13, and the FTC has usefully clarified that good-faith age gates or age checks, with the data not reused, won’t alone constitute a violation. The state layer is where the complexity is: age thresholds vary from under-16 to under-18; Maryland bans personal data sales and targeted advertising to minors outright; other states permit targeting only with consent under stricter conditions; separate age-verification mandates now apply to sites and app stores; and several states use a “should have known” standard rather than actual knowledge. On the programmatic side, publishers can no longer simply self-attest — platforms are expected to actively identify child-directed inventory. OpenX has built a dedicated marketplace for vetted child-directed inventory with external COPPA Safe Harbor oversight. The COPPA RTB signal remains binary, and many DSPs respond by refusing the traffic entirely.
Publisher takeaway
This is no longer a kids’-media problem — it applies by who is in your audience. Gaming, sports, music, entertainment and education sections all carry mixed audiences, and under a “should have known” standard, “we’re not a children’s site” stops being an assertion you can make without evidence. Do the audience audit by section and write the analysis down; a dated written assessment is the difference between a defensible position and a claim. Then implement good-faith age assurance — the FTC has effectively told you this is safe if the data stays walled off, so implement it, document the minimisation, and make sure no growth team quietly repurposes it. The uncomfortable part is that compliance currently costs revenue: flag inventory honestly and DSPs may drop it, which is a direct disincentive to correct signalling and a genuine perverse outcome. Sequence it deliberately — talk to your SSPs about vetted child-directed marketplaces before you flag inventory into a demand vacuum. And budget for the engineering, because fifty definitions means geo-conditional ad serving logic, not a memo from counsel. (Not legal advice; have counsel review the states you operate in.)
08Amazon’s million-dollar sellers revolt — and the complaints are familiar
Million Dollar Sellers (MDS), a group of merchants doing seven figures or more on Amazon, boycotted the platform this month. The concrete result: Amazon delayed a credit card transaction policy change. The grievance list spans fuel surcharges and the “DD+7” change delaying seller payment by a week — plus four advertising complaints: ads appearing in unexpected placements, an inability to fully exclude off-network sites (sellers can only “limit” them), no control over ad creative including AI-assembled ads, and no ability to manage ad types including Amazon’s new AI chatbot ads. AdExchanger notes the complaints rhyme with what advertisers said about Facebook’s Audience Network and continue to say about Google Performance Max.
Publisher takeaway
Read this from the other side of the transaction — you are the “off-network sites” the sellers want to exclude. Amazon’s off-network extension places retail-media demand on open-web publisher inventory, a demand source many publishers have been actively courting, and advertisers pushing for a full opt-out are pushing against your revenue. The reason they want out is that they cannot see where their money lands, which makes opacity in automated buying a demand risk rather than merely an advertiser grievance: when the buy side can’t verify placement, its first instinct is to restrict the entire surface rather than buy selectively. So make your retail-media placements independently verifiable — placement, page type, context, through your own logs or a verification partner — and you become the safe destination when advertisers start restricting. Second, ask your retail-media partners directly whether AI-assembled creative can run on your inventory and what approval rights you hold; if the answer is “none,” that should be a conscious decision by your standards team rather than a discovery after a complaint. And track MDS as a leading indicator: if sellers win off-network exclusion, a demand pool you may be forecasting as structural growth gets smaller.
09CTV’s dead air gets a standard: pause and menu ads become programmatic
On July 22 the IAB Tech Lab’s Advanced TV Working Group and Ad Format Hero Task Force finalised signalling for the CTV Ad Portfolio, covering six formats: Pause Ads, Menu/Home Screen Ads, Overlay Ads, Screensaver Ads, In Scene and Squeezebacks. It modifies existing frameworks rather than inventing new ones — AdCOM gains expanded enumerations for placement, on-screen position and playback method; VAST NonLinearAds now carries <MediaFiles>; OpenRTB menu/home-screen ads use the Native object while other formats use Video paired with NonLinear VAST delivery. Compliance is required across SSPs, exchanges, DSPs, SSAI/CSAI providers and measurement partners. Two days later, AdExchanger detailed Warner Bros. Discovery’s pause-ad business: standard pause ads on HBO Max since 2022 with soft gradient overlays, shoppable pause ads announced at May upfronts and expected later in 2026, and scene-level contextual targeting. VP Scott Rossman says standard pause ads “drive over 200% lift in website visits” versus benchmarks (Disqo data). Cited research: 67% of Gen Z and millennials prefer pause ads to a frozen screen; 51% took action after seeing one. WBD sells them mostly direct today and says the industry must align on standards before programmatic works.
Publisher takeaway
Pause ads existed for four years and stayed a direct-sold curiosity for one reason — there was no way to describe one in a bid request. That constraint was removed this week, and formats that get specified get bought while formats that don’t stay in direct-sales purgatory. The economics are attractive because the inventory is genuinely incremental: pause and menu ads monetise a moment that previously produced nothing, rather than cannibalising a mid-roll. If you sell CTV or in-app video, get engineering into the GitHub spec now — being signal-compliant before demand arrives is how you land in the first wave of buys rather than the third — and get a written support date from your SSP, because the standard only produces revenue when every hop in the chain carries it. For everyone else, take the pattern rather than the spec: the move is “find the dead moment in the user experience and make it a standardised, biddable unit.” Interstitials, load screens, app-resume states and end-of-article states are the web’s equivalent, and nobody has standardised them. (WBD’s 200% figure is vendor-supplied — a claim, not an audited result.)
10Streaming’s free-tier split: Paramount builds a “front porch,” Netflix refuses
Three positions, one week. Paramount is building an internally-named “free front porch” — more Paramount+ TV and film without a subscription, including titles like Landman and Dutton Ranch — on the condition that users create an account, enabling personalised ad delivery and an upsell path into the paid tier. It already runs Pluto TV as a pure free ad-supported service. Disney’s Adam Smith hinted the company may consider free content offerings. Netflix said no: co-CEO Greg Peters dismissed rumours of a free ad-supported live TV product, citing “cannibalisation of paid tiers.”
Publisher takeaway
Paramount’s version is not really a free tier — it’s a registration wall with prestige drama as the incentive, trading content margin for identity. In a market where third-party identifiers have collapsed, match rates are in single digits and browser-mediated measurement is still a draft spec (see #6), the logged-in user is the one asset whose value is rising. That is a strong validation of what most publishers are already doing, and an argument to be more aggressive about it: if Paramount will give away Landman for an email address, your content is not too valuable to trade for one. Set a registered-user target this quarter and give the growth team the content budget to hit it. Two second-order effects to plan for. Supply: Paramount’s front porch and any Disney equivalent create new impressions in the most premium video pool, so if you sell video against streaming budgets, model softer CPMs for 2027 and decide now whether you compete on price, context or data the streamers don’t have. And the cannibalisation question is your paywall question — Peters’ concern that free access degrades paid conversion is the exact argument inside every publisher’s subscription debate. Netflix and Paramount disagree because it is genuinely unresolved. Cohort your own registered-free users against subscribers and get your own answer.
11Beehiiv built its own SSP — newsletters get a programmatic layer
Beehiiv rolled out programmatic ads for newsletters this week across a platform of roughly 60,000 publishers, running on an SSP it built in-house since 2021 rather than plugging into an incumbent. The initial product is text-and-image native placements formatted to sit inside newsletter content; publishers control their own price floors; advertisers buy on CPC or CPM; targeting is contextual for now, with subscriber-level targeting in active development; tracking pixels handle measurement and attribution. CEO Tyler Denk plans to extend programmatic to publisher websites and dynamically inserted podcast ads, with a two-to-three-year goal of becoming “the go-to destination for advertisers to invest in sponsoring high-quality niche content at scale.” Beehiiv currently works directly with advertisers and “only a handful of agency partners.”
Publisher takeaway
Newsletter advertising has excellent fundamentals and terrible distribution economics — an agency planner cannot hand-allocate meaningful budget across 200 niche newsletters at $2,000 a deal, which is why the category has stayed small despite outperforming display on engagement. Programmatic access is the mechanism that fixes it, and the floor control is what makes it work in the seller’s favour: publishers set their own price and the inventory keeps its native format and contextual specificity, which is how scarce, high-engagement supply commands a premium instead of being commoditised the way open-web display was. The immediate consequence for you is that your newsletter inventory is about to have a comparable market price — which helps if yours is genuinely premium and exposes you if you’ve been charging on scarcity alone. Work out your true newsletter CPM now, before the benchmark exists. Then test programmatic on remnant inventory while keeping premium sponsorships direct-sold, and give the newsletter a rate card and a revenue owner, because it is inventory now rather than an audience-development line item. Note the direction of travel against everything else in this issue: email is the one channel Google’s crawler cannot intercept, and it is where new infrastructure is being built.
12Programmatic champions transparency — until a watchdog asks for a seat
Check My Ads, the nonprofit that documents where programmatic money lands, signed a master service agreement with a DSP to obtain a buying seat — for inventory analysis, supply-chain visibility and research on real campaign data. The DSP refused, saying it “weren’t able to get comfortable with the scope of protections extended to our vendor and supply partners.” A second DSP that had previously collaborated with the group also declined. Separately, AI ad-placement platform Thrad blocked campaigns targeting politics, abortion, mental health and LGBTQ topics. Check My Ads’ response: “If there are any DSPs out there that are confident in their supply and want to help us make programmatic accessible to small businesses, reach out.”
Publisher takeaway
A demand-side platform declined to sell access to a customer because it could not guarantee its suppliers would be shielded from that customer’s scrutiny. The product is a marketplace; the objection is that the buyer might inspect the merchandise. And the supply chain being protected is the one you sit in — some of it legitimate media that would survive any audit, some of it MFA and arbitrage competing for the same budget. Opacity covers both, and only one of them needs it. That makes voluntary disclosure a real differentiator right now: publish your supply paths, your sellers.json and ads.txt discipline, your reseller hops and your direct routes, because when the market’s default posture is refusing inspection, being inspectable costs you nothing if your supply is clean. Ask your SSPs what third-party auditing they actually permit — the answer tells you what you’re lending your brand to. And notice the shape of the whole week: Amazon’s sellers can’t verify placement (#8), 67% of video buyers lack confidence in the open exchange, and now a watchdog can’t get a seat. Everyone in the chain is on the outside of it. The separate Thrad item is a direct news-publisher revenue issue — sensitive-topic blocking has moved upstream from your inventory to the advertising itself.