Get cited in AI answers: what Google’s new pilot means for experts

To get cited in AI answers, your content has to add something to the answer itself, and Google has now attached money to exactly that standard. As the trade publication Digiday reported on Monday, 14 September 2026, Google is testing an “AI contribution pilot” in Search Console that pays publishers when their content contributes meaningfully to answers in AI Overviews, AI Mode and Gemini. According to the help text for the pilot, what counts is the contribution while an answer is being generated, whereas pages that merely confirm a finished answer or are linked to afterwards do not count. The pilot is aimed at publishers and websites. For speakers and experts it is a clear signal of the kind of content Google considers valuable in an AI answer. If you want to be named as an expert in AI answers, you should therefore use your talks and articles to publish insights of your own that an answer engine cannot find anywhere else.

Google’s AI contribution pilot: what Digiday reported on 14 September 2026

According to the report by Jessica Davies and Sara Guaglione at Digiday, Google has been testing the programme for some time without a public announcement. Digiday describes the model as “pay-per-value”, which means payment depends on the value Google assigns to a piece of content for the answer. Google has not explained how the amounts are calculated. Participating publishers see an “AI contribution” panel in Search Console with a monthly earnings figure and its history, and they can leave the pilot at any time.

Barry Schwartz of Search Engine Roundtable first noticed the pilot in Search Console in April 2026 and quotes the help text from the copy Digiday had seen. Speaking to Digiday, Google described the programme as an early-stage learning pilot meant to find out how to reward high-quality content on top of the traffic and tools it already provides to websites. The groundwork had been laid on 18 June 2026 by Markham Erickson, Vice President of Government Affairs and Public Policy at Google, who wrote on Google’s public policy site that Google was piloting a new way to partner with websites whose content helps keep generative AI responses fresh and factual.

According to Digiday, the pilot appeals more to small and mid-sized publishers and now reaches beyond news sites. One source described the early payouts as small compared with advertising revenue.

What counts as a contribution to an AI answer in the pilot

The most revealing part of the story is a distinction in the help text that separates content which shapes an answer from content that only comes into play afterwards:

AI contribution focuses on the phase of generation where web content can significantly influence what response is generated. Web content may also confirm facts or be linked to within Google's AI after a response is generated, but those instances don't qualify for AI contribution.

Google help text on the AI contribution pilot, quoted by Search Engine Roundtable

In the pilot, then, Google pays only when a page helps determine what the answer says. A page that merely confirms a statement gets nothing, even if it is linked below the answer. What is a question of money for publishers gives experts, in our view, Google’s clearest description so far of the content that counts in an AI answer.

Getting cited in AI answers starts with original expertise

Visibility in AI answers means that an AI system such as Google AI Overviews or ChatGPT names you, or cites your page as a source, when someone asks a question about your field. The work behind it is known as generative engine optimisation, or GEO. Generative engine optimisation for experts covers everything a specialist does so that answer engines find their content and attribute it to them.

In its guide AI features and your website, Google says there are no additional requirements and no special optimisations needed to appear in AI Overviews or AI Mode, provided a page is indexed and eligible to be shown in Search with a snippet. According to Google, both features may break a question down into several related searches across subtopics, a technique called “query fan-out”, which finds a wider set of supporting pages than a classic search. As early as 21 May 2025, John Mueller of Google advised on the Search Central Blog that site owners should focus on unique, non-commodity content, because users ask longer and more specific questions in the AI features.

The pilot puts a price on this advice. A page that repeats what a hundred other pages say can at best confirm an answer. The material an answer is built from comes from specialists who are the first to publish a figure of their own or a method from their practice.

How experts get named in ChatGPT and Google’s AI answers

The pilot concerns Google only. None of the providers discloses how ChatGPT or Perplexity select their sources, but in our observation every answer engine asked about experts needs names that it finds together with a topic on several independent pages. Which pages a talk produces, and how to check whether AI assistants mention you, is set out in our article Get mentioned in ChatGPT as a speaker.

Conference talks leave a citable trail

A talk at a specialist conference leaves behind pages written by someone else. The programme page lists your name with your talk title and abstract along with the date and venue of the conference. Later a recording or a write-up records what you said on stage. Each of these pages is a statement about your expertise that someone else has published.

Suppose you give six talks a year, every conference publishes a programme page with your abstract, half of them record the sessions, a third publish a write-up, and you publish an article of your own on each talk. The result is:

• 6 programme pages with your name, title and abstract
• 3 recordings that capture your arguments in your own words
• 2 write-ups in which an organiser quotes you
• 6 articles of your own that treat the topic in more depth

That makes 17 pages a year, 11 of them on other people’s websites. Whether an answer engine uses any of them is its own decision. By the standard of the pilot, these traces only count once they contain more than your name. An abstract with a survey of your own or a result from a project can help shape an answer, while an abstract made of buzzwords at most confirms what the answer already knows. What an abstract has to do is explained in What is a conference abstract?.

Writing abstracts and talks that contribute to AI answers

An answer engine is most likely to take up passages that make sense on their own and contain a clear statement with a figure and a source. For speakers, in our view, this means the abstract states the finding of the talk in one sentence that can be understood without the rest. The talk itself contains at least one insight you can support from your own work, such as a metric from client projects or an approach you give a name.

Mueller recommends supporting text with high-quality images and videos. In our view a recording works best when a transcript or written summary sits alongside it, from which your statements can be quoted. How to turn the talk into an article afterwards is shown in Turning a conference talk into content.

Five steps to get cited in AI answers more often

1. Define your own insight: choose a figure, a method or a case for your topic that only you can support from your own work.
2. Write the abstract around the finding: put the core statement in the first sentence so that it makes sense on its own on the programme page.
3. Turn recordings into text: ask the organiser for the recording and publish a transcript or summary alongside it.
4. Publish an article on the talk: develop the insight on your own website and link to the conference programme page.
5. Check your mentions: ask ChatGPT, Gemini and Perplexity at regular intervals about experts on your topic and note which sources they cite.