Google is testing a new way of paying publishers whose content helps produce answers in its AI products. But there is an important detail that has received less attention: publishers can see the money Google says they have earned, but they still cannot see the detailed measurement behind that number.
That distinction matters.
This is not simply another story about Google paying websites for AI usage. The more useful question for publishers is how Google decides that one piece of content contributed enough to an AI answer to deserve payment.
Google has not publicly published a detailed formula for that calculation.
What Google’s AI Contribution Pilot actually does
Google confirmed in June that it was exploring new ways to work with websites whose content contributes to the freshness and factuality of generative AI responses through grounding. The company described this as part of its broader effort to support the information ecosystem. Google’s original announcement did not publish a payment formula or detailed eligibility criteria.
By September, reporting from Digiday established more about the experiment. Publishers invited into what is described as the AI Contribution Pilot can receive payments when their content “significantly” contributes to AI-generated responses. The reported products include Gemini, AI Overviews and AI Mode.
The pilot is separate from ordinary search traffic. A page being linked or cited after an AI response has already been produced does not automatically mean that page qualifies for a contribution payment. Search Engine Journal’s reporting on the pilot describes the distinction between content that helps generate an answer and content that is merely referenced afterward.
The number publishers can see
For publishers inside the pilot, Google Search Console adds an AI contribution panel showing earnings for the program and some historical information.
That gives publishers something they normally would not get from an AI system: a direct financial signal that their content has been considered valuable.
But the panel does not appear to expose the full calculation behind the earnings figure.
That creates an unusual situation.
A publisher can potentially see:
- that Google considers its content valuable enough to generate a payment;
- the earnings figure reported for the pilot; and
- some historical information about those earnings.
What the publisher cannot currently see in equivalent detail is the underlying measurement that explains why that particular amount was assigned.
The number publishers cannot see
This is the part I think publishers should pay more attention to.
There is a difference between knowing what Google paid and knowing how Google valued the contribution that produced the payment.
Google’s public material currently explains the principle: content must meaningfully contribute to AI-generated answers. But it does not publicly define a simple unit such as:
One contribution = X amount of money.
There is also no public indication that every citation, impression or appearance in an AI answer carries the same value.
That means a publisher looking at its earnings cannot independently reconstruct the calculation from the information Google has made public.
And that is the missing number: the measurable bridge between an AI contribution and the payment assigned to it.
A citation is not the same thing as a contribution
This distinction is easy to miss because traditional web publishing has trained publishers to think in terms of links, impressions and clicks.
AI search makes the relationship more complicated.
A page might be:
- visible in an AI search feature;
- linked as a source;
- used to verify a statement;
- used as part of the information that shapes the answer; or
- used in a way that Google considers significant enough to qualify for payment.
Those are not necessarily the same event.
Google’s own Search Console reporting makes this difference clearer. Its generative AI performance report, which rolled out globally on August 31, reports impressions for AI Overviews and AI Mode and lets publishers examine pages, countries, dates and devices. Google’s Search Console documentation explains what those measurements represent.
But an impression is not a contribution payment.
That means a publisher should not look at an increase in AI impressions and assume that Google will automatically increase its AI contribution earnings by the same amount.
Google itself says unique content matters
There is another useful clue in Google’s public guidance.
Google’s guide to generative AI features in Search encourages publishers to create valuable, non-commodity content, including material with a unique point of view and genuine first-hand experience.
Google specifically warns against simply recycling information that is already available elsewhere or could easily be produced by a generative AI system.
That does not tell us the payment formula. It would be wrong to turn Google’s SEO guidance into a claim that “unique articles earn more money” under the pilot.
But it does show the direction Google publicly says it wants its AI systems to value: content that adds something useful rather than another copy of information already everywhere.
So how much are publishers getting?
Reported payments vary dramatically, which is another reason not to treat any single figure as the normal payout.
The Information reported on September 30 that roughly 100 digital publishers were participating and described examples ranging from less than $1,000 over several months for some smaller sites to much larger sums for individual participants.
It also reported that several small and midsize publishers were receiving payments amounting to less than 0.1% of their advertising revenue.
Those figures are reported examples, not Google’s published payout schedule. They should not be used to predict what another website will earn.
The important point is that the payment range appears wide while the public explanation of the calculation remains narrow.
Why the black box matters more than the payout
A small payment can still be useful if a publisher understands what caused it.
For example, suppose a publisher learns that a particular type of original research consistently contributes to AI answers. That information could influence future editorial decisions.
But if the publisher only sees a payment total, it learns much less.
It cannot easily tell whether the value came from:
- the subject of the article;
- the originality of the information;
- the freshness of the reporting;
- the usefulness of a particular passage;
- the frequency with which the content was used;
- the importance of the information to an answer; or
- some combination of several signals.
That is why the transparency question is more important than simply asking whether Google is paying enough.
Takeaway: I would not judge this pilot by the payment number alone. For publishers, the more valuable development would be knowing what Google considers a meaningful contribution and being able to connect that measurement to specific pages or types of content. Until that becomes clearer, the earnings figure is useful as a signal, but limited as an editorial guide.
Publishers now have another AI measurement problem
This is happening at the same time Google is expanding its AI reporting tools.
Search Console now gives publishers more visibility into how their pages appear in generative AI search and Discover. Google also rolled out a control allowing websites to manage inclusion in AI Overviews, AI Mode and generative AI features in Discover. Google’s documentation for that control says publishers can manage whether their site’s links and content are included in these generative AI features.
So publishers increasingly have three different questions:
- Where is my content appearing?
- How is my content being used?
- What value does Google assign to that use?
Search Console answers the first question better than it used to. The AI Contribution Pilot begins to address the third financially.
The second question, and especially the connection between all three, remains much less transparent.
What publishers should do right now
There is no reason for publishers to start rewriting their entire editorial strategy around this pilot.
Google’s own guidance points in a more sensible direction: keep producing useful, original, reliable material for people rather than trying to manufacture content specifically for AI systems.
For publishers already invited into the pilot, the sensible approach is to study the terms, monitor the earnings and treat the payment data as an additional signal rather than a complete explanation of content value.
For everyone else, the bigger lesson is about the kind of content worth building.
Original reporting, firsthand testing, useful comparisons, specific observations and information that is difficult to find elsewhere have value beyond traditional search rankings. They can help a publication stand apart whether the visitor arrives through Google Search, an AI answer, Discover, social media or a direct visit.
That is also why publishers should not confuse AI visibility with business success. A page appearing in an AI answer does not automatically mean it generated a visit, a subscriber, a sale or meaningful revenue.
FaithfulBiz has looked at another side of this changing search environment in our guide on how to know if ChatGPT cites your website. The underlying issue is similar: being visible to an AI system and receiving measurable value from that visibility are two different things.
What Google still needs to explain
The AI Contribution Pilot is still an early experiment, so some of this uncertainty may simply reflect the fact that Google is testing the model.
Still, several questions would become much easier for publishers to answer if Google eventually published more detail:
- What exactly qualifies as a “significant” contribution?
- How is contribution measured?
- Does the value of a contribution depend on the type of query or answer?
- Can publishers see which pages generated their earnings?
- Can publishers see how often their content contributed without qualifying for payment?
- How are earnings calculated across different countries and types of publishers?
Those answers would turn the pilot from a payment experiment into something much more useful for publishers trying to understand the economics of AI search.
The bigger signal for publishers
Google’s experiment matters because it suggests that the old relationship between search engines and publishers is changing.
For years, the basic exchange was relatively easy to understand: a search engine sent a user to a website, and the website tried to turn that visit into advertising revenue, a subscription, a sale or another business outcome.
Generative AI changes that exchange.
An AI system can use information from several websites and produce an answer without requiring the user to visit every source. That makes contribution a potentially important economic concept alongside the familiar metrics of rankings, impressions and clicks.
Google is now experimenting with putting a price on that contribution.
But until publishers can see more of the measurement behind the price, they are still being shown the outcome without the complete calculation.
For now, that is probably the most important thing to watch as Google’s AI Contribution Pilot develops.
The money is visible. The measurement behind the money is not.
Conversation (0)