# Google's Pay-Per-Value AI Licensing Is a Price Without a Market *Google now pays publishers when their content shapes an AI answer. It also decides what that contribution was worth, in private. A number set by the only buyer is a payout, not a price.* Published: 2026-09-16 | Read time: 5 min read | Author: Smalk AI Research | Source: https://www.smalk.ai/blog/google-pay-per-value-ai-licensing-price-without-market --- Google's pay-per-value AI licensing pilot is the first time Google has agreed that publisher content has a price inside its AI answers. It is also the first time Google has reserved the right to set that price alone, score it in private, and display only the monthly total. Twenty-three years ago the same company ran AdSense for seven years without telling publishers its cut. The open web accepted it because Google was the only pipe. The agentic web is about to make the same mistake, with a worse deal. - Google is paying publishers when their content 'significantly' contributes to answers in Gemini, AI Overviews and AI Mode, via a Search Console widget that shows a monthly figure and no formula. - AdSense hid the revenue split from 2003 to 2010 but let advertisers set the price through an auction. The AI pilot hides the price itself, because there is no second bidder. - Publishers inside the pilot are right that precedent matters. They are wrong that a hedge priced by the hedger will grow into a market. - The missing bidder is the brand. Brands need to be inside the same answers Google is now paying to ground, and today they have no lane to fund the publishers who build them. ## What pay-per-value AI licensing is Pay-per-value AI licensing is a compensation model in which an AI platform pays a content owner only when it judges that the content materially shaped a generated answer, at a rate the platform sets internally. It differs from pay-per-crawl, which pays on ingestion, and from pay-per-query, which pays every time content informs a response. In Google's version, help text seen by Search Engine Journal draws the line at the generation stage: content that confirms a fact or appears as a link after the answer exists does not qualify. That last clause matters. Google is paying for the invisible input, the grounding, and explicitly not for the visible output, the citation. The unit publishers can count is excluded. The unit only Google can see is what gets paid. ## AdSense's black box vs the AI contribution black box In the open-web economy, Google's opacity covered the split, not the price. From AdSense for Content's launch in 2003 until May 2010, publishers did not know Google's cut. When Google finally disclosed 68% to publishers, it did so under pressure from Italy's antitrust authority. But throughout, advertisers set the underlying price through a live auction. The black box sat between a real market and the publisher. In the agentic-web economy, the AI contribution pilot removes the market entirely. There is no auction, no bid, no competing buyer. Google scores the value of a page to its own answer, pays a number, and shows the total. The old black box hid a percentage of a price the market had already discovered. The new one hides the price because no market has discovered it. That is not a transparency problem. It is a structure problem. ## What AdExchanger and Digiday reported, and what we verified AdExchanger's September 15, 2026 roundup led with the story, sourcing Digiday's September 14 report by Jessica Davies and Sara Guaglione. Digiday found that at least dozens of publishers, skewing small and mid-sized and reaching well beyond news, had been approached for an 'AI contribution pilot' that adds an earnings panel to Search Console. Google confirmed an early-stage learning pilot and pointed to a June 2026 policy post on grounding. Search Engine Roundtable first spotted traces of the program in April 2026. We could not verify participant numbers beyond 'dozens,' and neither could Digiday. Google's separate figures, more than 200 titles in its news AI scheme and over 2,800 publications in 33 countries under News Showcase, are Google's own and were restated without independent confirmation across coverage. One source described early payouts as 'peanuts' next to ad revenue; one exec called offers 'lowball' and said negotiations were ongoing. No payout amounts have been published. ## Why 'inside the tent' is the strongest case for taking Google's money ### Precedent beats principle: the case pilot publishers make Two publisher executives in the pilot told Digiday they would rather test direct payments than wait for referral economics to return. One framed it as the early shape of a marketplace for inference data, with Search Console evolving into the same feedback loop for AI that it already provides for search. Luke Stillman of Madison and Wall was blunter: publishers have little leverage, so take the new revenue while it exists and redeploy it into subscriptions and events. This is a serious argument. A payment rail that exists beats a fair one that does not. ### Why a hedge priced by the hedger cannot become a market David Buttle of Spur read the pilot as a strategic hedge: Google building the plumbing for a world where it might be forced to pay, while making sure that world does not arrive on usage-based terms. The evidence supports him. Google chose the one model that keeps pricing power entirely in-house, excluded the countable citation event, and, as one exec warned, gained an argument in future negotiations that compensation already exists. Precedent is real. But a precedent that establishes 'Google decides' is the wrong precedent to lock in. Markets form when a second buyer shows up, not when the first buyer adds a widget. ## What this means for brands and for publishers ### For CMOs, media buyers and agencies: you are the missing second bidder Google has now confirmed, with money, that specific publisher pages decide what its AI answers say. That is the exact inventory brands need to influence, and Google offers no way to buy into it; AI Mode ads run beside the answer, from your product feed, not inside the grounding. Treat the pilot as a market signal: the pages that shape AI recommendations in your category have a value Google is willing to pay for. Audit which ones they are, ringfence a 2026 test line for native placements on them, and stop assuming GEO alone will win a recommendation that is built on pages you do not own. ### For publishers: take the pilot, refuse the precedent Opt in if the money is real; there is no virtue in leaving revenue on the table. But log the terms, keep your generative AI performance report as independent evidence of contribution, and stay aligned with IAB Tech Lab's per-query position so 'Google decides' does not become the industry default. Then build the second revenue line the pilot cannot give you: brand demand for the pages AI engines rely on. The publisher who can prove a page shapes AI answers holds inventory, and inventory with two bidders has a price. ## Three signals that decide whether AI content gets a real price First, standards. IAB Tech Lab released CoMP v1.0 in March 2026, a protocol requiring commercial agreements before AI crawling, and CEO Anthony Katsur told a publisher audience in July 2026 that unpaid crawling is theft. If CoMP's per-query bid-response design gets adopters, Google's private scoring becomes the outlier. Second, distribution. The EU's Digital Markets Act is prying open the assistant layer: MacRumors spotted iOS code letting ChatGPT or Claude replace Siri, and Google revamped EU search results in September to comply. More engines competing for grounding content means more potential buyers. Third, demand-side formation. When brand budgets start flowing to the publisher pages that shape AI answers, the price Google assigns in private will have to compete with one set in the open. ## Conclusion Hold on to this: Google's pay-per-value AI licensing proves publisher content has a price inside AI answers, and proves the open web cannot let a single buyer set it in the dark twice. Smalk AI exists to be the second bidder: a Generative Engine Advertising network that places native ads for AI agents on the publisher pages AI engines actually rely on, funded by brands that need to be inside the answer, with those publishers paid on every campaign in the open. What to watch next: the first published payout figure from Google's pilot, and whether it lands above or below what brands will pay for the same page. ## FAQ ### What is Google's AI contribution pilot? Google's AI contribution pilot is an invite-only program that pays publishers when Google judges their content 'significantly' contributed to an AI answer in Gemini, AI Overviews or AI Mode. Participants see a monthly earnings figure in Search Console, without a breakdown of how it was calculated, and can opt out at any time. Digiday reported it on September 14, 2026. ### How does pay-per-value differ from pay-per-crawl and pay-per-query? Pay-per-crawl pays once when a bot ingests content. Pay-per-query pays every time content informs an answer, the model IAB Tech Lab favors because it scales with use. Pay-per-value pays only when the platform decides the content mattered, at a rate the platform sets. The first two are usage-based and countable by the publisher; the third is judgment-based and countable only by the buyer. ### Should publishers join Google's pay-per-value licensing pilot? Most publishers should join if the payout is material, since it is revenue for content they already produce. They should also document terms, keep independent contribution data, and avoid treating the pilot as a settled compensation model, because accepting a price set by one buyer can weaken leverage for better terms later. ### Why does Google's AI licensing pilot matter for brands? The pilot confirms that specific publisher pages determine what Google's AI answers say, and that Google values them enough to pay. Brands cannot buy into that grounding through Google today, so their route to being recommended inside AI answers runs through those publisher pages, not through ads placed beside the answer.