AI Search & GEO

What Is Query Fan-out? A Plain-English Explanation

By BacklinkPrice Team · Nov 18, 2026 · Updated Nov 20, 2026 · 9 min read
Illustrated cover for What Is Query Fan-out?

Query fan-out turns one question into several related searches, run at the same time, to build one AI answer. This guide explains how it works, with real examples and checks you can run.

Query fan-out is the way an AI search engine turns one question into several smaller searches, runs them at the same time, and builds one answer from what they find. Google uses the term for AI Overviews and AI Mode, and ChatGPT does a similar rewrite before it searches the web.

Key takeaways

  • Google says AI Overviews and AI Mode may use query fan-out, running several related searches across subtopics and data sources to build one answer.
  • ChatGPT searched the web in only 14.4% of 66.6 million free responses in an Ahrefs study from August and September 2026, and those responses averaged about 1.6 fan-out queries each.
  • A page can be cited for a sub-question it answers well, even if it does not rank for the exact question the user typed.
  • Deep Search in Google AI Mode can issue hundreds of searches for one research question, Google said in May 2025.
  • No tool shows Google's private sub-queries, so you measure coverage by checking which related questions your page answers, not by reading a fan-out log.

What query fan-out means

Query fan-out is the process where one user question becomes several related searches that run at the same time. The engine then combines the results into a single answer with links to the pages that supplied each part.

Google Search Central defines fan-out queries as "a set of concurrent, related queries generated by the model to request more information and fetch additional relevant search results to address the user's query" in its 2025 guide.

This mechanism sits at the center of generative engine optimization (GEO), which is the work of making pages easy for AI systems to find, read, and cite.

Query fan-out also explains why answer engine optimization (AEO) focuses on direct answers. Assistants pull passages that answer a sub-question, so a page that states the answer first is easier to lift than one that buries it.

Classic search runs one query against an index and returns a ranked list of links. Query fan-out adds a planning step before that list, so the answer draws on several searches instead of one.

How query fan-out works

Query fan-out works in four steps: plan the subtopics, run the sub-queries at the same time, retrieve the best passages for each, and combine them into one answer with links.

At Google I/O in May 2025, Google said AI Mode breaks a question into subtopics and issues "a multitude of queries simultaneously on your behalf". The model first decides what a complete answer needs, then writes a sub-query for each part.

Each sub-query returns its own set of pages, and the system pulls the passages that answer it best. SEO Birdy noted in 2026 that sources are not only web pages, since Google also names the Knowledge Graph, real-time information, and shopping data as inputs.

The model then merges the passages into one response and attaches links to the pages that support each claim. A follow-up question starts the loop again, with the earlier answer as extra context for the next fan-out.

Scale varies with the question. Google said in May 2025 that Deep Search, a research mode in AI Mode, uses the same technique at a larger scale and can issue hundreds of searches for one request.

Query fan-out matters because an AI answer can cite a page for a sub-question, even when that page does not rank for the exact question the user typed.

An Ahrefs study of 1.4 million ChatGPT prompts in 2026 found that ChatGPT cited only 49.98% of the URLs it retrieved. Retrieval is not the same as citation, so being read does not guarantee being shown.

Cited pages also matched the prompt more closely. In the same study, their titles scored 0.60 for similarity to the prompt, against 0.48 for pages that were retrieved but not cited.

Source type also matters. In the same study, Ahrefs reported in 2026 that pages from ordinary web search were cited 88.46% of the time, against 12.01% for news, 1.93% for Reddit, and 0.51% for YouTube.

Citations rarely turn into visits. Pew Research Center found in 2025 that users clicked a standard result on 8% of visits when an AI summary appeared, against 15% without one, and clicked a link inside the summary on only 1% of visits.

ChatGPT does not search on every prompt. Semrush reported in 2026 that ChatGPT ran a live web search on about 34.5% of queries as of February 2026, down from about 46% in late 2024.

Each citation is a separate win. An AI citation is the link an assistant shows next to a claim, and it is the unit you should count, not the number of pages retrieved.

Real examples of query fan-out

Real fan-out examples show one broad question splitting into cost, comparison, timing, requirements, and proof sub-queries, each answered by a different page.

Google gives a lawn example in its guide to generative AI features, as WPPoland reported in 2026. For "how to fix a lawn that is full of weeds", the fan-out queries might include "best herbicides for lawns", "remove weeds without chemicals", and "how to prevent weeds in lawn".

A roof cost question shows the same pattern. SEO Birdy illustrated in 2026 how "how much does a roof replacement cost" can split into insurance, permits, financing, timing, and hiring sub-queries before one answer is written.

Parent questionSub-query the engine may runPage that can be cited
How to fix a weed-filled lawnBest herbicides for lawnsA herbicide guide
How much does a roof replacement costRoof replacement insurance coverAn insurance explainer
Best pediatric dentist in AustinPediatric dentist reviews from parentsA reviews page

The dentist example from SEO Birdy in 2026 works the same way. A parent asking for the best pediatric dentist for an anxious 5-year-old may trigger sub-queries on age, insurance, sedation, reviews, and Saturday hours.

How to check or measure query fan-out

You cannot read Google's private sub-queries, so you check fan-out coverage by listing the related questions behind a keyword and marking which ones your page answers.

Google Search Console does not show the sub-queries either. As of November 2026, its generative AI report shows impressions in AI Overviews and AI Mode, but it reports no clicks, no click-through rate, and no queries for those appearances.

  1. List the related questions from autocomplete, People Also Ask, and your Search Console queries.
  2. Group them into aspects such as cost, timing, comparisons, requirements, and proof.
  3. Give each aspect its own headed section that answers in the first sentence.
  4. Mark each sub-query covered, partial, or missing, then write the missing sections first.
  5. Re-check after publishing, using AI Overview and AI Mode impressions in Search Console as the outcome to watch.

A worked example shows what a coverage score looks like. SEO Birdy reported in 2026 that Google's SEO Starter Guide scored 33% on a Fan-Out Map for "seo starter guide", with 110 sub-queries across 15 aspects: 19 covered, 34 partial, and 57 missing.

Coverage is only half the job, since the page must also be visible to AI systems at all. AI visibility explains how assistants find and choose pages, and why a page that is not indexed cannot be cited.

Good and bad query fan-out

Good fan-out coverage means one page answers several related sub-questions in self-contained sections. Bad coverage means the page repeats the head keyword and answers nothing else.

SignalGoodBad
Sub-question coverageAnswers cost, timing, comparisons, and requirementsRepeats the main keyword only
Section styleDirect answer first, then detailAnswer buried in paragraph 6
FactsNamed figures, dates, and sourcesVague claims with no source
Page countOne page per intent, with linked sectionsA new thin page for every sub-query

Spinning up a thin page for every sub-query is a risk, not a tactic. Google treats creating separate pages for every fan-out variation, mainly to manipulate rankings or AI responses, as scaled content abuse as of November 2026.

ChatGPT leans toward official sources when it does search. In Ahrefs October 2026 study, 6.8% of fan-out queries used the site: operator, and 75.9% of the most common site: searches pointed at a brand site.

That is why an official page that states prices and facts plainly can win a citation over a roundup. A brand mention in AI search also helps, since assistants often check what other pages say about a brand.

The same split shapes LLM SEO, which tracks how assistants choose sources across engines, not just how one page ranks on Google.

Common mistakes and myths

Common mistakes come from treating fan-out as a keyword list to chase, instead of a set of questions to answer.

  • Writing a separate thin page for every sub-query. Google treats that as scaled content abuse when the aim is to manipulate rankings or AI answers, so map sub-queries to sections first and pages second.
  • Assuming a top Google ranking guarantees a citation. Ahrefs found in 2025 that only 12% of AI-cited URLs were in Google's top 10, so track citations separately from rankings.
  • Believing Google publishes its fan-out queries. It does not, and Search Console shows no query list for AI features, so any tool that claims to show Google's exact sub-queries is estimating.
  • Believing more sub-queries always mean a better answer. ChatGPT averaged about 1.6 fan-out queries per searching response in Ahrefs' 2026 study, and most prompts triggered no web search at all.
  • Believing every question fans out into 8 to 12 sub-queries. No fixed count exists: Ahrefs measured about 1.6 for ChatGPT, while Google's Deep Search can issue hundreds.
  • Hiding the answer in long prose. Each section should answer its sub-question in the first sentence, with figures and dates after it, so a passage can stand alone.

Frequently Asked Questions

What is query fan-out in simple terms?

Query fan-out is when an AI search engine turns one question into several smaller searches, runs them at the same time, and combines the results into one answer. Google uses it in AI Overviews and AI Mode, and ChatGPT rewrites prompts into extra searches in a similar way.

Does query fan-out replace keyword research?

No. Keyword research still tells you which topic deserves a page and what people search for. Query fan-out tells you what that page must cover to answer the whole question, so use the sub-questions as the page outline, not as a reason to drop keyword research.

Can I see the fan-out queries for my keyword?

No, not Google's exact ones. Google does not publish the sub-queries behind an answer, and Search Console reports impressions without showing the queries. You can model the fan-out from autocomplete, People Also Ask, and your own Search Console queries, then check which ones your page answers.

How many sub-queries does a question produce?

There is no fixed number. In an Ahrefs study from August and September 2026, ChatGPT averaged about 1.6 fan-out queries per response that searched the web, and only 14.4% of responses searched at all. Google's Deep Search can issue hundreds of searches for a research question.

Is query fan-out good or bad for small sites?

It helps small sites that answer a sub-question well, since a citation does not require a top ranking for the head term. It hurts thin pages that repeat a keyword without answering anything. A clear, factual section on one sub-topic can earn a citation that the whole page could not.

Sources

If you want the wider picture, read our guide to generative engine optimization (GEO) next.

BacklinkPrice Team
Written by
BacklinkPrice Team
Link building and backlink pricing research

The BacklinkPrice Team researches backlink prices, publisher sites and link building tools. We check prices and plans against each provider's own pages and date them, so you know how current the numbers are.

More from BacklinkPrice Team

Related guides