The case for and against publishers blocking Google

Full analysis

Google's AI Overviews and AI Mode are answering queries without sending the click, and search referral traffic is down 30 to 40% at some publishers. Digiday's Sara Guaglione and Seb Joseph, with hosts Kamika McCoy and Tim Peterson, spent an episode debating whether publishers will actually block Google's crawlers in response. Their answer: mostly no, and not soon. The math still doesn't work for anyone who lives on ad revenue.

Reversibility and what's really being decided. For any single publisher, blocking Google is Type 1, hard to reverse and expensive to test. The 2014 Axel Springer run makes that plain: traffic fell about 40% in two weeks and they folded. The question that actually matters for ad-tech operators is whether the open-web inventory pool that programmatic buyers depend on is quietly shrinking, and who gets squeezed as it does. That's the frame. The forcing functions are calendar-real: Cloudflare turning on AI-crawler blocking by default for new sites on September 15, 2025, the Digiday Publishing Summit the same week, and the UK CMA's opt-out mechanism heading to the Our Media Strategies event in October.

The council

The Market Analyst. Two supply chains are diverging, and buyers should price it. Google's open-web inventory, the stuff flowing through AdSense and Ad Manager, is getting thinner and lower-intent as zero-click eats the top-of-funnel visit. Guaglione's People Inc. figure, Google search dependency cut from roughly 60% to 20%, is the direction of travel for every serious publisher. That 40 points didn't vanish, it moved to email, direct, and apps, where Google's ad stack doesn't reach. For an operator: the audiences advertisers want are migrating off the surfaces the largest ad platform monetizes. Liz Reid told the IAB in January 2025 that Overviews surface more mid-tail publishers, which Peterson correctly read as Google saying it needs no single publisher. True. But a platform that no longer needs any one publisher also stops being where the premium audience lives.

The Skeptic. The block thesis is mostly theater and everyone on the call knows it. Seb Joseph's line does the work: at every technological shift in 15 to 20 years, we ask whether Google will do right by the open web, and it never happens. What has to be true for a block to bite? A large coalition acting at once. That runs straight into antitrust exposure, the WFA advertiser-boycott mess is the template, plus the politics of keeping a Fox-affiliated outlet inside the same tent as everyone else. Guaglione's earnings-call reporting is the giveaway: executives pressed by analysts admit they're "not that close." The public rhetoric is a negotiating posture, not a plan. Nobody blocks alone, and nobody can legally block together.

The Operator. Tuesday morning, blocking Google is one config change and a revenue cliff by Friday. That's why it won't happen at scale, and it's the wrong thing to watch. The real work is already underway: cutting Google dependency the slow way, the way People Inc. went from 60% to 20% by building email, events, and direct. At 90 days, what an ad-tech operator actually sees is attrition. Open-web session volume erodes quarter over quarter, RPMs on the remaining Google-sourced traffic soften as intent thins out, and your SSP fill assumptions drift. The CMA opt-out is the mechanism that actually changes behavior, because it lets a publisher stop AI scraping without losing search rank. That removes the suicide part of the decision. Watch adoption of that lever. The block headlines are a sideshow.

The Customer / End User. Two customers here, and they want opposite things. The advertiser wants the premium audience wherever it goes, and Kamika McCoy flagged where some of it is going: brands are building their own blogs and answer-engine-optimized content to grab the zero-click real estate publishers are losing. So advertisers are becoming publishers, which scrambles agency planning about where branded content sits in the funnel. The publisher, meanwhile, wants optionality, not exodus. Nobody on this call wants to leave Google. They want leverage over it. Publishers want to renegotiate the terms, and right now they have almost no cards to play.

The CFO. Run the payback and the block never clears. You'd trade a known 20 to 40% of traffic today for a regulatory outcome that's uncertain, years away, and captured by everyone who didn't sacrifice. Peterson floated the "Lehman Brothers martyr" idea, that CNN, Disney, NBCUniversal, or the New York Times could absorb a block to force regulatory action. The Times has the direct-subscriber base to survive it. But a CFO doesn't volunteer the company as a martyr so the industry gets a remedy. The rational spend is diversification: subscriptions, newsletters, events, and licensing deals with OpenAI or Anthropic. Microsoft's content marketplace, run by Julia Bazer out of Bloomberg, is the same idea, and Guaglione's caveat matters, it's "in development" and pays nobody's bills this year.

The tensions

Is the CMA opt-out a real lever or a false comfort? The Operator and Market Analyst treat it as the one structural concession that changes the calculus, because it splits "stop AI scraping" from "lose search rank." The Skeptic notes U.S. publishers don't have it and Google fought it everywhere it could. Whether it crosses the Atlantic via DOJ remedy or FTC action is the hinge.

Attrition versus event. The Operator says the story is slow erosion nobody will livestream. Peterson and the CFO entertain a deliberate martyr block that forces regulation. These are genuinely different futures for open-web supply, and buyers should model both.

Does Google even care? The Market Analyst says a platform that needs no single publisher has already conceded that premium audience is leaving. The Skeptic says that's exactly why publishers have no leverage and the block is empty. Both read the same Liz Reid quote and land in opposite places.

Synthesis

This hinges on two beliefs. First, whether the CMA-style opt-out spreads to the U.S., which is the only thing that makes reducing Google exposure survivable rather than suicidal. Second, whether publishers can rebuild enough direct audience fast enough that the 20% Google residual stops being a hostage. The council leans clearly: no dramatic block, steady attrition. Publishers keep the rhetoric loud and the crawler open, because the earnings-call admissions and the Springer scar both point the same way.

For ad-tech operators the actionable read covers both the buy and sell side. Open-web inventory sourced from Google search is getting thinner and lower-intent while the good audience migrates to email, direct, and app surfaces Google's stack doesn't monetize. DSPs and SSPs leaning on Google-originated supply should model that compression now, and identity and measurement vendors should note that the audience they most want to graph is moving to logged-in, first-party surfaces. What to verify: CMA opt-out adoption rates once it's live, whether any U.S. remedy echoes it, and whether the OpenAI and Anthropic licensing dollars scale past pilot.

The Prediction

Prediction: Through the end of Google Alphabet's Q1 2027 earnings call (late April 2027), no U.S. publisher with meaningful ad revenue will fully block Googlebot search crawling as a standing policy; the visible action will instead be adoption of scrape-without-derank opt-out tools (Cloudflare defaults, CMA-style mechanisms) and direct-audience diversification.

Confidence: High. Springer's 40% cliff and executives' own "not that close" admissions both point one way.

Why: The 2014 Axel Springer block dropped traffic about 40% in two weeks and got reversed, and Guaglione's reporting says large-publisher executives told analysts on recent earnings calls they are "not that close" to blocking, so the public rhetoric is negotiating posture, not intent. A unilateral block is a revenue cliff a CFO won't volunteer for, and a coordinated one carries antitrust exposure nobody wants after the WFA boycott saga. The route that actually spreads is the opt-out lever, because it lets a publisher stop AI scraping without losing search rank, removing the suicidal half of the trade. The opposite outcome, a real standing block, requires a company to torch known revenue for an uncertain regulatory payoff that free-riders capture, which is exactly the martyr move rational operators avoid.

Revisit by 2027-04-30: We're right if the notable publisher moves through April 2027 are opt-out adoption and diversification, with no ad-dependent U.S. publisher maintaining a full Googlebot search block. We're wrong if a top-tier ad-supported U.S. publisher (think CNN, NBCUniversal, People Inc.) blocks Google search crawling as standing policy and holds it past a fiscal quarter.

One wrinkle to keep an eye on: the CMA lever. If a U.S. antitrust remedy imports the scrape-without-derank right, the cost of reducing Google exposure drops enough that the slow attrition speeds up, and open-web supply compression stops being a 2028 problem for the buy side.