Query fan-out

What Google says, source by source

Google has described query fan-out in at least four places. The wording differs a little each time, so each quote below carries its date.

5 March 2025. Announcing AI Mode as an experiment in Labs, Google wrote:

"It uses a “query fan-out” technique, issuing multiple related searches concurrently across subtopics and multiple data sources and then brings those results together to provide an easy-to-understand response."

20 May 2025. From the post on Google I/O, when AI Mode began rolling out in the U.S.:

"AI Mode uses our query fan-out technique, breaking down your question into subtopics and issuing a multitude of queries simultaneously on your behalf."

Search Central documentation, "AI features and your website", last updated on 10 December 2025:

"Both AI Overviews and AI Mode may use a "query fan-out" technique — issuing multiple related searches across subtopics and data sources — to develop a response."

AI Mode help page, retrieved on 15 September 2026:

"AI mode uses a “query fan-out” technique, dividing your question into subtopics and searching for each one simultaneously across multiple data sources. It then brings those results together to provide an easy-to-understand response."

Two details in these sentences often get lost. The documentation names AI Overviews as well as AI Mode, so fan-out is not limited to AI Mode. And it says "may use": Google does not promise the technique for every query.

None of the four sentences says which data sources are meant. The March 2025 post lists what AI Mode can draw on besides web content: "fresh, real-time sources like the Knowledge Graph, info about the real world, and shopping data for billions of products."

The number Google does not give

None of the four sources says how many searches one question produces. "Multiple related searches", "a multitude of queries" and "searching for each one" describe a process without a count. Any number of sub-queries attributed to Google can be checked against these pages.

The I/O post does contain two numbers, and both belong to something else. The first is about Deep Search, which Google announced for AI Mode in the same post. Deep Search "uses the same query fan-out technique but taken to the next level", and Google writes of it: "It can issue hundreds of searches, reason across disparate pieces of information, and create an expert-level fully-cited report in just minutes."

The second is about buying tickets with the agentic features of AI Mode, where AI Mode will "kick off a query fan-out, looking across sites to analyze hundreds of potential ticket options". Those are hundreds of options, not hundreds of searches.

What a page is compared against

On a classic results page, pages are ranked for the query the person typed. With fan-out, Google splits the question into subtopics and searches for those. A page can fit one of these searches well without being a good answer to the question as a whole.

Directly after the fan-out sentence, the Search Central page continues: "While responses are being generated, our advanced models identify more supporting web pages, allowing us to display a wider and more diverse set of helpful links associated with the response." Google does not say how often those links differ from the pages that rank for the typed question.

The same page states that there are "no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary". Fan-out adds no rules of its own. When it is used, what changes is the search a page is measured against: a subtopic rather than the question as typed.

What Search Console shows

Search Console reports what the user asked. The Performance report groups data "by the search query users typed". For AI Mode, the help page on counting treats a follow-up question as a new query: "All impression, position, and click data in the new response are counted as coming from this new user query."

The generative AI performance report, available to all websites since 31 August 2026, shows impressions in AI Overviews and AI Mode grouped by page, country, date or device. It has no query dimension. None of the Search Console help pages retrieved for this entry mentions the searches that fan-out generates, and neither report shows which subtopic a page was found for.

What follows for a page

• Answer each subtopic a page covers in its own section, under a heading that names it, so that one section can serve one search.
• Do not write for a single phrasing of the question. The searches are derived from the question, and people phrase the same question in many ways.
• Keep the basics Google names for supporting links: the page must be indexed and eligible to be shown in Search with a snippet.
• Measure links in AI answers separately from rankings for the typed query. Google says it identifies further supporting pages while the response is generated.

Common mistakes

• Treating fan-out as an AI Mode feature only. Google's documentation names AI Overviews too.
• Quoting a count of sub-queries as Google's. Google has published none for ordinary fan-out.
• Using "hundreds of searches" for fan-out in general. Google wrote it about Deep Search.
• Looking for the generated searches in Search Console. The reports group data by the typed query, or by page, country, date and device.
• Assuming fan-out runs on every query. The documentation says the features "may use" it.

Related terms

• AI Mode: Google's conversational search experience, for which Google announced the technique in March 2025.
• AI Overview: the generated answer above classic results, which the documentation says may also use fan-out.
• Passage Ranking: ranking individual sections of a page rather than the page as a whole.
• Content Chunking: structuring a page into self-contained sections that each answer one question.