Query fan-out is what happens between your customer's prompt and ChatGPT's answer. When ChatGPT decides a question needs fresh information, it does not search for the prompt word for word. It writes one or more web searches of its own, reads what comes back and builds the answer from that. Those searches, not the prompt, decide which pages it ever gets to see.
That makes fan-out queries worth studying for anyone working on AI visibility. Ahrefs did exactly that in a study published on October 12, 2026, analyzing 66.6 million free, logged-out GPT-5.6 Luna responses and their 15 million fan-out queries from August and September. The short version: most responses contain no fan-out queries at all, and when ChatGPT does search, it increasingly goes to whoever owns the information.
For brands that means two jobs: cover the specific sub-questions ChatGPT searches for, and make sure your own site holds the facts it looks up about you.
What query fan-out actually is
ChatGPT answers many prompts from what it already knows. When a prompt needs current or specific information, such as a price or a product spec, it searches the web first. Query fan-out is the step where the model turns the prompt into one or more search queries, each aimed at a different part of the question.
Ask whether a stability running shoe is worth it for flat feet, and ChatGPT might look for a recent review, the official product page with the price, and a comparison with a rival model. You never typed those searches. The model did, often adding words like "price", "review" or "official", or a site: operator that searches one website only. Your page has to match one of those searches to be read at all.
What Ahrefs found in 15 million fan-out queries
Most answers contain no fan-out queries
Only 14.4% of the 66.6 million responses contained fan-out queries, roughly 1 in 7, with about 1.6 fan-out queries each on average. The study's author draws a blunt conclusion:
With about 1.6 queries, your page has to match one of those one or two searches. If it doesn't, ChatGPT never sees it.
Louise Linehan, Ahrefs
These were free, logged-out sessions, which Ahrefs notes search less than paid or reasoning models. Chris Long of Nectiv, testing via the API, measured 7.61 fan-outs per prompt. The count varies, the principle does not: each fan-out is a narrow search, and your page either matches it or stays out of the answer.
When ChatGPT searches, it often goes straight to the brand
Of all fan-out queries, 6.8% used the site: operator to search a single website. Among the roughly 160 most-searched sites, three in four of those site: searches (75.9%) targeted a brand site, meaning a company website like apple.com or ikea.com rather than a retailer, government site, forum or review site. Consumer brands took 8 of the 10 most site: searched domains.
Price and review searches are rising
After "2026", the most common words in the fan-outs were "official" and "price". "Official" appeared in 13.1% of English fan-out queries in September, up from 10.1% in August. "Price" went from 6.04% to 8.88% of fan-outs, a relative rise of 47% in one month, while "reviews" grew by a third.
Citations are shifting to original sources
In separate Brand Radar citation data, ChatGPT's Reddit citations dropped 88% in six months, and PubMed climbed from 64th to 4th among its most-cited domains. Ahrefs reads this as ChatGPT moving from sites that summarize information to the sources that produce it.
Cover the sub-questions behind a topic
With only one or two narrow searches per answer, a broad page that touches everything loosely is a weak match for any of them. What tends to match is a page, or a clearly headed section, that answers one sub-question directly. The Ahrefs study suggests grouping fan-out queries by five intents, and they make a good checklist for every topic you want to be found for:
- Pricing: what it costs, what is included and how versions differ.
- Specs: dimensions, materials, ingredients and compatibility.
- Reviews: what customers and independent testers say, downsides included.
- Comparisons: how you stack up against named alternatives, and who each option suits.
- How-to: setup, use, maintenance and troubleshooting.
Write headings that read like the question and answer it in the first sentence or two. Keep facts current and show when they were last updated, since "2026" topped Ahrefs' list of the most common fan-out words. Solid keyword research helps too: the long-tail questions people type into Google tend to overlap with the sub-questions a model splits a prompt into.
Make your own site the place ChatGPT searches for your brand
When ChatGPT needs a fact about your brand, it often searches your domain directly. If the answer is not there, it has to come from somewhere else, or not at all.
A separate Ahrefs test of 20,000 US prompts shows how much this counts. 84.4% of factual questions about pricing, product features or company details cited at least one official source, and 63.4% of their citations went to official websites. Ahrefs' advice: put those facts on your site, in plain text, on a page ChatGPT can find.
Pages to check first
- A pricing page with real numbers in the page text, not only in a PDF, an image or a configurator.
- Product and spec pages that state the facts in sentences or tables, using the model names people search for.
- Honest comparison pages for the alternatives buyers weigh you against.
- Reviews and testimonials on your own domain, with more detail than a star rating.
- Company facts: who owns the business, where it operates and how support works. "Company", "brand" and "owner" all appear in Ahrefs' top 20 fan-out words.
For webshops most of this already sits in your product data; the work is getting it onto crawlable, readable pages. That is where ecommerce SEO and AI visibility meet.
Where your own site is not enough
Official sources dominate factual questions, not comparisons. For "A vs B" or "which is cheaper" questions, the share of answers citing an official source dropped to 68.3%, and only 46.4% of citations went to official websites. Ahrefs also found that publications with OpenAI licensing deals more than doubled their combined share of citations among ChatGPT's 200 most-cited domains, from 7.5% in March to 15.8% in July, and still held 12.5% in late September. Independent reviews and press coverage still carry real weight.
How to analyze query fan-out in your industry
One fan-out query tells you little; the pattern across many prompts tells you a lot. Ahrefs does this in its Brand Radar tool, which records fan-out queries for ChatGPT and Perplexity: filter the AI responses report on your topic, platform and market, then filter the "Fan-out queries" column for terms like site:, "official", "price", "review" or "best".
Your industry can look very different from the averages. For "running shoes", Brand Radar returned 1,474 US ChatGPT responses with fan-out queries: "best" appeared 206 times, "review" 117 times, "vs" 105 times and "official" only 65 times. Ahrefs concludes that a running shoe brand would get more from review coverage and comparison pages than from its product pages alone.
Whatever tool you use, the process is the same:
- 1List the prompts real buyers ask in your category, from sales calls, support tickets and search data.
- 2Collect the fan-out queries ChatGPT runs for them, in your market and language.
- 3Group them by intent, then count the most common terms and site: domains, yours and your competitors'.
- 4Map each group to a page. No page means a content gap; fan-outs aimed at review sites mean a coverage gap.
- 5Repeat monthly. Ahrefs stresses that its numbers are a snapshot of one model between August and September 2026, and that OpenAI changes how ChatGPT retrieves information constantly.
Frequently asked questions
What is query fan-out?
Query fan-out is the process in which an AI search tool like ChatGPT turns a single prompt into one or more web searches before answering. The model writes those searches itself, often adding terms like "price" or "review", or a site: operator. Only pages that match those searches are considered for the answer.
How many searches does ChatGPT run per prompt?
It depends on the model and setup. In Ahrefs' study of free, logged-out sessions, 14.4% of responses contained fan-out queries, with about 1.6 each on average. Other studies, using the API or logged-in accounts, found far more fan-outs per prompt.
Does query fan-out change keyword research?
It adds a layer rather than replacing it. You still research what people type into search engines, but you also look at the sub-questions a model generates from their prompts, and make sure a page answers each one directly.
Find out what ChatGPT searches in your market
Query fan-out makes AI visibility something you can inspect: the searches are specific, and you can check whether your pages answer them. LASEO maps the fan-out queries in your industry as part of our answer engine optimization work, then closes the gaps on your own site and in the coverage around it. Get in touch and we will start with the prompts your buyers actually ask.