Google is testing direct payments to publishers whose content meaningfully contributes to AI-generated answers in Gemini, AI Overviews, and AI Mode.
The programme, called the AI contribution pilot, goes further than simply showing publishers where their content appears in Google’s AI products. Participating sites can accrue earnings when Google decides their material has significantly influenced the response being generated.
The critical detail is that Google is not describing this as payment for every citation, mention or link. That makes ordinary AI visibility checker data a poor proxy for how much a publisher might earn. Google appears to be trying to value a source’s contribution to the answer itself.
| What Google is testing | Payments when publisher content significantly contributes to an AI-generated response |
|---|---|
| Products covered | Gemini, AI Overviews and AI Mode |
| Who can participate | A limited group of publishers participating in the pilot |
| What does not automatically qualify | Simply being linked or used to confirm information after the response has been generated |
| Biggest unknown | How Google calculates the financial value of each contribution |
Google is paying for contribution, not every citation
Reporting from Digiday, subsequently highlighted by Search Engine Roundtable, gives a much clearer picture of how the pilot works than Google had previously disclosed publicly.
The reported programme documentation says participating pages can accrue earnings when their content contributes significantly to an AI-generated response. Publishers can see accumulated earnings and payment history through Google’s site-owner interface and can leave the programme.
More interesting is what apparently does not count. Content might be linked from an AI response or used to confirm facts after the response has been generated without qualifying for an AI contribution payment.
That creates a much narrower payment trigger than a simple citation counter. A publisher could theoretically receive plenty of visible exposure without every appearance creating revenue. Conversely, the reported rules suggest that the value calculation happens around the grounding and generation process rather than being determined by the number of links displayed to the user.
For publishers, this makes the eventual measurement model almost as important as the payment itself.
The payout formula is still a black box
Google has not publicly explained how it values one contribution against another. Participating publishers reportedly receive an earnings figure, but not the sort of page-level breakdown that would explain why one article generated value while another did not.
This is the pilot’s biggest weakness in its current form.
A recurring complaint from site owners already measuring generative AI visibility is that a top-line number is difficult to act on without the evidence underneath it. Knowing that visibility went up does not tell an editorial team which source, query or type of information caused the change. A monthly payment figure creates the same problem in financial form if publishers cannot trace value back to particular pages or content characteristics.
Without that attribution, publishers cannot reliably answer basic questions such as whether original reporting is earning more than evergreen guides, whether fresh updates are particularly valuable, or whether specialist factual content contributes more often than broad summaries.
It also makes experimentation difficult. Changing ten articles and seeing earnings increase next month proves very little if Google doesn’t show which contribution events changed.
Google had already signalled that a new content value exchange was coming
The payment pilot didn’t appear out of nowhere. In June, Google said it was exploring new partnership and value-exchange models across the information ecosystem.
In Google’s statement on supporting the information ecosystem, the company said it was piloting a new way to partner with websites whose content meaningfully contributes to the freshness and factuality of generative AI responses through grounding.
The AI contribution pilot now appears to put a financial mechanism around that idea.
For publishers, Google’s language around freshness and factuality is worth watching more closely than generic talk about AI optimisation. If the programme eventually rewards content according to the value it adds during grounding, the useful publishing strategy may look very different from chasing the highest possible number of AI mentions.
Fresh reporting, original datasets, specialist reference material and pages containing facts that cannot be reproduced from dozens of near-identical articles could become more economically useful inputs. That is a logical possibility rather than a disclosed ranking or payment formula, and publishers should not rewrite their editorial strategy around it until Google provides harder evidence.
Smaller publishers have a genuine decision to make
Digiday reports that at least dozens of publishers have been approached about the programme and that interest has extended beyond traditional news organisations. Smaller and mid-sized publishers reportedly appear particularly interested because they are less likely to negotiate large standalone licensing agreements with AI companies.
There is an obvious attraction. If Google is already using public web content to ground AI experiences, receiving incremental revenue for qualifying uses is better than receiving nothing.
The harder question is whether the payment is high enough to compensate for the value being transferred. Digiday reports that some publishing executives consider the early figures small relative to their advertising businesses, while others see participation as a useful first step towards a direct publisher-AI revenue model.
Publishers invited into the programme should therefore evaluate more than the first payment number:
- Rights: establish exactly what additional content rights participation grants Google.
- Attribution: find out how much evidence is available about which pages generated qualifying contributions.
- Unit economics: compare the payment with revenue that the same content historically produced through visits, subscriptions, advertising, affiliates or other conversions.
- Exit terms: understand what happens to content usage and accrued earnings after leaving the pilot.
- Negotiating leverage: consider whether accepting a standard programme affects the publisher’s ability to negotiate different licensing terms later.
A small recurring payment can still be useful. The mistake would be treating any payment as proof that the economic exchange is fair.
AI citations and AI contribution payments need separate measurement
The pilot also reinforces an increasingly important measurement problem: a citation, a mention, a recommendation and a compensated grounding event are not the same thing.
DIY AI’s LLM citation tracking methodology already treats citations separately from brand mentions and downstream outcomes. Publishers should add contribution revenue as another independent layer if Google’s pilot expands.
A visible citation tells you that a source was attached to an answer. It does not prove the content qualified for payment. A monthly earnings figure tells you that Google assigned financial value somewhere in the publisher’s content, but without better reporting it may not reveal which source created that value.
Combining those numbers into one “AI performance” score would hide the information publishers actually need.
The bigger story is Google establishing that AI grounding can have a price
Google says the programme is an early-stage learning pilot, so its current payment levels, access rules and reporting should not be treated as a finished commercial model.
The more consequential development is the precedent. Google is experimenting with a system in which a publisher’s contribution to an AI-generated answer can create a direct financial return rather than relying entirely on a user clicking through to the original website.
That could eventually create a new layer of publishing economics alongside advertising, subscriptions, affiliate revenue and conventional content licensing.
But it is not a functioning marketplace yet. Publishers still cannot see enough about how contribution value is calculated, why one use qualifies while another does not, or whether the resulting payments reflect the commercial value of the content being supplied.
The next meaningful step is not simply expanding the programme to more websites. It is giving publishers enough attribution to understand what Google believes their content is worth. Without that, the AI contribution pilot establishes a price without showing how Google arrived at it.
