What is query fan-out?
Nati Elimelech·Updated 27.07.2026 Query fan-out is the technique where an AI engine splits one question into several searches and builds one answer.
Instead of one search on the question as phrased, the model splits it into a series of related searches across subtopics and data sources, and builds one answer from the results. The name comes from Google’s documentation for AI Overviews and AI Mode, but the behavior is not exclusive to Google: ChatGPT search also splits questions into several searches, in its own way. Google is careful to say AI Overviews and AI Mode may use the technique, not that they must, and that each may use different models and techniques, so the answers and links each shows will vary.
How query fan-out works
At Google the fan-out queries run at the same time, each against the regular index, and the model picks sources for the answer from the results. Google’s own example, from its optimization guide: for the question “how to fix a lawn that’s full of weeds”, the fan-out queries might be “best herbicides for lawns”, “remove weeds without chemicals” and “how to prevent weeds in lawn”.
How query fan-out relates to SEO and GEO
The documented opportunity is breadth: Google writes that fan-out lets it display a wider and more diverse set of supporting links than classic search, meaning a page can become a source through a query nobody typed. And there is an equally documented warning: Google writes that creating content for every search variation, fan-out queries included, primarily to manipulate rankings or AI responses in Google Search violates its scaled content abuse spam policy. The practical move: identify the entities and subtopics that surround the question, and strengthen the content that answers them. The full breakdown is in the query fan-out article.
The difference between query fan-out and keyword research
Keyword research guesses what people type, based on a search engine’s historical data. Fan-out shows what the engine itself searches, right now, in order to answer. The first is a sample of user behavior, the second a look into the mechanism that assembles the answer. And unlike a keyword, a fan-out query is not a target for a new page but a question your existing content needs to answer.
Questions about query fan-out
How do you actually see the sub-queries?▼
Google does not expose the fan-out queries it runs, and they are not reported in Search Console. ChatGPT is different: you can watch the searches it runs while answering, and there are tools that expose them systematically, including the free Chrome extension I built. That is an observation of another engine, not a look at what Google runs.
Should you build a page for every sub-query?▼
Building a page for every sub-query is exactly what Google warns against: it writes that creating content for every possible search variation, fan-out queries included, primarily to manipulate rankings or AI responses violates its scaled content abuse spam policy.
Google also adds that its systems recognize a page's relevance even without an exact match between the query and the content. Sub-queries teach you what the engine asks, not how many pages to open.
How many sub-queries does one search produce?▼
Google publishes no number for fan-out queries in AI Overviews or AI Mode. The only number it has named is "hundreds of searches", and that was said in the Deep Search announcement, which Google describes as the same fan-out technique taken to the next level, on the blog rather than in documentation. The numbers circulating in the industry, like 8 to 16, are third-party observations, not Google documentation.
Does ChatGPT run query fan-out too?▼
Yes, in its own way. OpenAI documents that ChatGPT search rewrites the question into one or more targeted queries sent to external search engines, and by OpenAI's own description the follow-up queries go out after reviewing the first results, meaning one after another rather than all at once like Google.
The name query fan-out itself comes from Google's documentation, and OpenAI does not use it. And the practical difference actually favors ChatGPT: you can watch its searches as it answers, while Google's cannot be seen.
Can you influence the sub-queries?▼
Google documents no lever for influencing fan-out, and writes there are no additional technical requirements and no special files or markup needed for appearing in AI Overviews or AI Mode: a page needs to be indexed and eligible to show with a snippet. What you can do: identify the subtopics and entities that surround the question, and make sure your existing content answers them in depth.