Query fan-out
A retrieval technique in which one user question is expanded into several related searches across subtopics and data sources, so an answer is assembled from many small results rather than from one ranked list.
For twenty years the model in your head was one query, one results page, one position. Query fan-out breaks that model, and Google says so in its own documentation rather than leaving it to be inferred.
What Google says
The definition comes straight from Search Central: "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" (Google Search Central, page last updated 2025-12-10, checked 2026-08-06).
The next sentence is the one worth reading twice. Google states that while responses are being generated, its models identify more supporting web pages, which allows it to display a wider and more diverse set of helpful links than a classic web search would.
Wider and more diverse. That is Google describing an expansion of the source pool, not a contraction, and it is the opposite of how most people first hear about this feature.
What it changes
One question stops behaving like one question.
Someone asks which home battery suits a three-bedroom house with an electric car. Under classic search, that is one query against one ranked list, and roughly ten results compete for it.
Under fan-out, it becomes several searches at once: typical household consumption, battery capacity ranges, EV charging load, installation constraints, price. Each of those pulls its own results, and the answer is assembled from across them.
Your page is no longer competing for the query the user typed. It is being considered against sub-questions the user never typed and you never saw.
The practical consequence
Two things follow, and only one of them is the obvious one.
The obvious one is that a page which answers a specific narrow question completely can be pulled into an answer about a much broader question. That is the "wider and more diverse" part working in your favor, and it is why a well-made specific page is worth more here than it was under a rank-for-the-head-term strategy.
The less obvious one is that context does not travel. A sub-search retrieves your page on its own, with none of the surrounding conversation. If your page assumes the reader read the previous section, or arrived from your category page, or knows what your product does, none of that is present at the moment your page is being evaluated.
So the requirement is not more pages. It is more self-contained pages. Each one should name its subject, state its answer in plain text near the top, and stand up alone. That is the same discipline described in how to write an answer capsule an engine can extract.
What it does not license
The tempting misreading is to build a page for every sub-question you can imagine.
That is how a content plan becomes a doorway. Google's spam policies define doorway abuse as "sites or pages ... created to rank for specific, similar search queries" that lead to intermediate pages less useful than the destination, and scaled content abuse as many pages "generated for the primary purpose of manipulating search rankings and not helping users" (Google Search Central spam policies, page last updated 2026-05-15, checked 2026-08-06).
Fan-out is a reason to write clearly, not a reason to write more. The engine is generating the sub-questions. You do not need to guess them and mirror them one to one.
Where the honest description stops
This is the part that separates a definition from a sales pitch.
Google names the technique. It does not publish how sub-queries are generated, how many are issued for a given question, how their results are weighted against each other, or how the final source set is chosen. None of that is documented.
So any tool that shows you "your fan-out queries" is showing you a model's guess at what Google might have asked. That can be a useful brainstorming device. It is not a readout, and anyone presenting it as one is describing internal behavior nobody outside Google can observe.
The same caution applies to fan-out as an explanation. It is easy to attribute any change in visibility to it, precisely because it is invisible. An unfalsifiable explanation is not an explanation.
What is checkable instead
The mechanism is opaque. The inputs are not.
Google states the eligibility floor plainly: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet ... There are no additional technical requirements." The same page lists practices that continue to matter, including crawling allowed in robots.txt and by your CDN, content available in textual form, internal links that make pages findable, and structured data that matches the visible text.
Every one of those is a yes-or-no you can establish today about a specific URL. None of them requires knowing what the fan-out asked.
And note the boundary between surfaces. Google states that "AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary," and that AI Overviews "often don't trigger" at all. Being absent from one tells you little about the other, and being absent from a query that produced no overview tells you nothing.
Measuring around it
There is no fan-out report. Google states that appearances in AI features are "included in the overall search traffic in Search Console" and reported "within the 'Web' search type," with no separate AI breakdown.
What is left is sampling: run the questions your buyers would actually ask, record which engines return or name your pages, repeat it on a schedule. That produces a number with a method and a date rather than a rank, which is the correct shape for measuring a probabilistic system. Search prompt is the unit, and how to measure your AI citation rate is the method.
The honest limit
Readiness does not make an engine choose you. Fan-out widens the pool of pages that could be pulled in, and that is genuinely good news for specific, well-structured pages, but the selection inside that wider pool is still Google's and it is still probabilistic.
Citedon measures whether engines can reach, read, and name your pages, and reports that as evidence. It cannot promise placement in an AI Overview or in AI Mode, and it does not query the AI Overview box itself: the Gemini check runs against a Gemini model through Google's API with web search grounding enabled, which is a different surface and we say so.
How to check yours
Take the page you would most want pulled into an answer, and read it as if you arrived with no context at all. Does it name its subject in the first line? Is the answer in text rather than in layout? Would it stand up alone if it were the only page a machine saw?
Run a free scan on that URL to see how ChatGPT, Perplexity, Gemini, and Claude read it today, or read why is my site not appearing in Google AI Overviews for the full diagnostic order.