What is query fan-out, and why does it matter for which pages AI search cites?

· Edited by Chris Dolan

Query fan-out is what an AI search system does when it turns one question into several related searches, runs them together, and builds its answer from the combined results. Google uses the term for AI Mode and AI Overviews, and other AI search systems do something similar. Every statement below was checked against Google's, OpenAI's and Microsoft's own pages on 9 October 2026.

The practical point is this: the pages an AI answer cites are drawn from the results of those extra searches, and not only from the results for the question as it was typed. A page can be cited for a question it would never rank for directly.

What query fan-out is

Google defines it in its guide to optimizing for generative AI in Search, last updated on 10 July 2026: “A set of concurrent, related queries generated by the model to request more information” and fetch more search results for the user's question. Its example is a person asking how to fix a lawn full of weeds. The fan-out queries might cover the best herbicides for lawns, removing weeds without chemicals, and preventing weeds from coming back.

The term arrived with AI Mode. Google's launch post of 5 March 2025 described the new mode as “issuing multiple related searches concurrently across subtopics and multiple data sources” and then bringing the results together into one response.

Where it is used

The name is Google's, and the behavior is common to several systems.

  • Google Search. Google's page on AI features and your website says both AI Overviews and AI Mode may use the technique.
  • ChatGPT search. OpenAI's help article says: “ChatGPT search typically rewrites your query into one or more targeted queries” for its search partners, and may send further, more specific queries after seeing the first results.
  • The Gemini API. Google's developer documentation describes the same step for applications built on its models: “the model automatically generates one or multiple search queries and executes them”.

The details differ in each. What they share is that the question a person types is not the only search that gets run.

Why it matters for which pages get cited

The requirement for being cited by Google's AI features is no different from ordinary search. The AI features page says a page “must be indexed and eligible to be shown in Google Search with a snippet”, and that there are no additional technical requirements.

What changes is the set of pages considered. Because the system searches for the sub-questions too, a page that answers one of them well can be shown as a supporting link, although it would not appear for the broad question. Google says the AI features can show a wider and more varied set of links than a classic search for that reason. The section of our audit checklist on how AI search decides which businesses to name explains retrieval and generation in more detail.

What you can see, and what you cannot

For Google Search, the sub-queries are not shown to site owners. Digiday's 2025 explainer put it this way: “Google also doesn’t share which queries are being processed”. A tool that lists fan-out queries for a keyword is showing its own reconstruction, not a report from Google Search.

Two of the operators' own products do show something.

  • The Gemini API returns the search queries the model ran along with its answer, for a developer's own requests.
  • Bing Webmaster Tools has an AI Performance report with a view called grounding queries. Microsoft's launch post says it “Shows the key phrases the AI used when retrieving content” that was cited from your site, and that the data is a sample.

What Google says to do, and what to avoid

Google's guide draws a firm line here. It warns against creating separate content for every variation of how people might search, fan-out queries included. Doing that mainly to manipulate rankings or AI responses, it says, breaks Google's policy against scaled content abuse. It adds that a large number of pages does not make a site better or more relevant.

The useful reading of fan-out is therefore narrower than many articles suggest.

  • Cover the real sub-questions on the page where they belong. A page about fixing a weedy lawn that also deals with prevention answers two of the fan-out searches in Google's example.
  • Do not build a page for each guessed query. That is the practice Google's guide names.
  • Keep to the basics. Google's position is that a page eligible for ordinary search results is eligible for its AI features.

How to measure it on your own site

You cannot see the fan-out, but you can see its effect. Search Console's Generative AI performance report counts how often links to your site were shown in AI Overviews and AI Mode. Its help page defines the figure: “Impressions are how many times links to your site were shown to a user” in a generative AI feature. The report groups the count by page, country, date and device. It has no list of queries.

The page view is the one to read. A page that collects AI impressions while ranking for few ordinary searches may well be reaching people through sub-questions. Our guide to setting up Search Console covers getting access to the reports.

Sources

All eight were read on 9 October 2026.

  1. Google's Guide to Optimizing for Generative AI Features on Google Search, Google Search Central, last updated 10 July 2026.
  2. Expanding AI Overviews and introducing AI Mode, Google, 5 March 2025.
  3. AI Features and Your Website, Google Search Central, last updated 10 December 2025.
  4. Searching the web with ChatGPT, OpenAI Help Center.
  5. Grounding with Google Search, Gemini API documentation, last updated 23 September 2026.
  6. Explainer on query fan-out in Google's AI Mode, Digiday, 19 June 2025.
  7. Introducing AI Performance in Bing Webmaster Tools Public Preview, Microsoft Bing Blog, 10 February 2026.
  8. Generative AI performance report (Search), Google Search Console Help.