Query Fan-Out: How AI Mode and AI Overviews Search Before They Answer

Google introduced the term query fan-out when it launched AI Mode as a Labs experiment on March 5, 2025, describing a system that issues multiple related searches at once across subtopics and data sources. Its documentation on AI features says both AI Overviews and AI Mode may use the technique.

Google's AI optimization guide (published May 15, 2026, updated July 10, 2026) defines it as a set of concurrent, related queries generated by the model. In its example, a question like how to fix a lawn full of weeds might fan out into searches such as "best herbicides for lawns", "remove weeds without chemicals" and "how to prevent weeds in lawn".

The query fan-out technique works inside a loop Google describes for AI Mode: make a plan, run many searches at once, then adjust the plan based on what comes back. Its posts and documentation describe each step:

• Split the question. The model breaks the question into subtopics and writes a related search for each, "issuing a multitude of queries simultaneously", in the words of Google's I/O 2025 announcement.
• Search more than web pages. At AI Mode's launch Google said the searches can also tap real-time sources such as the Knowledge Graph, information about the real world and shopping data for billions of products.
• Adjust the plan. Google says AI Mode adjusts its plan based on what the searches find, so the set of searches is not fixed in advance.
• Retrieve through the normal ranking systems. Pages come back from Google's index through its core ranking systems, and the model writes one answer from what they say.
• Pick the links while writing. Google says its models identify supporting pages as the response is generated. A page only has to be indexed and eligible to show a snippet; there are no additional technical requirements.
• Fall back when unsure. When Google isn't confident an AI response is helpful enough, AI Mode shows a set of web links instead.

Deep Search uses the same technique at a larger scale, issuing hundreds of searches for one request, and in November 2025 Google said Gemini 3 lets fan-out run even more searches. What Google has not published is how many searches an ordinary fan-out runs. The counts in SEO articles, from 5 to 20 or more, come from third-party tests, and they disagree. Google also says only that AI Overviews may use the technique, not that every Overview does.

Query fan-out spread through Google Search as AI Mode did, and Deep Search took it further.

• March 5, 2025: AI Mode launches in Labs with query fan-out.
• May 20, 2025: AI Mode rolls out in the US without sign-up; Google previews Deep Search, which takes fan-out further.
• July 16, 2025: Deep Search launches for Google AI Pro and Ultra subscribers, issuing hundreds of searches for a single request.
• October 7, 2025: AI Mode reaches more than 200 countries and territories.
• May 19, 2026: Google says AI Mode has passed one billion monthly users.

If fan-out builds an answer from several searches, its sources can rank for any of them. That is what the data shows. In our 700-question software study, only 55 percent of the links in Google's AI Overviews came from domains in the organic top ten for the question asked. In our four-industry study about half did. ChatGPT shows the same pattern: checked against Google and Bing for the searches it ran, most of the pages it pulled up were on neither first page.

The implication is practical. Ranking for your head term is not enough to be in the answer, and not ranking for it does not keep you out. What matters is being the best page for the narrower questions the system asks on the way.

Optimizing for fan-out means being the best page for the narrower questions it asks, without making pages mainly to target them.

• Map the sub-questions. For each important buyer question, list the follow-ups a careful researcher would ask: cost, fit, alternatives, setup, risks.
• Answer each one clearly, once. One strong page or section per real question. Google warns that generating pages mainly to target fan-out queries can break its scaled content abuse policy.
• Build depth on your subject. Covering a topic thoroughly gives the system more of your pages that match its narrower searches. See topical authority.
• Get cited where you cannot rank. Some sub-questions will always be answered by review sites, publishers or forums. Accurate presence there is part of the job.

This is the planning work at the center of AI search optimization. For how AI Mode differs from AI Overviews, see Google AI Mode SEO. In software categories those sub-queries include comparisons, alternatives and integrations, which is why this planning anchors our SaaS SEO agency work.

Four query fan-out FAQs follow, from the plain definition to Deep Search.

What is query fan-out?

Query fan-out is a technique where an AI search system breaks one question into several related searches, runs them at the same time, and builds its answer from the combined results. Google uses it in AI Mode and AI Overviews.

Should I create a page for every fan-out query?

No, you should not create a page for every fan-out query. Google's own guidance warns that making pages mainly to target fan-out queries and manipulate rankings or AI responses breaks its spam policies. Cover the real questions buyers have, one page per question, and do it well.

How is query fan-out different from normal search?

Normal search ranks pages for the query you typed, while query fan-out also searches subtopics you did not type. That is why the sources in the answer can come from pages that do not rank for your original query at all.

What is Deep Search?

Deep Search is a more intensive version of AI Mode that Google launched for paying subscribers in July 2025. It uses the same fan-out technique at a larger scale, issuing many more searches before writing a report.

• Google: Expanding AI Overviews and introducing AI Mode (March 5, 2025)
• Google: AI Mode in Google Search, updates from I/O 2025 (May 20, 2025)
• Google Search Central: AI features and your website
• Google Search Central: Optimizing your website for generative AI features on Google Search
• Google: Search at I/O 2026 (May 19, 2026)
• Google: Gemini 3 in Google Search and AI Mode (November 18, 2025)
• Google Search Help: AI Mode in Google Search

Every High Salience engagement starts with the Category Salience Brief, which shows where AI answers to your buyer questions get their sources: your commercial questions run across Google and the AI surfaces, competitors side by side, and a ranked list of what to fix first.