A shopper used to type “running shoes flat feet” and scroll. Now they ask an AI assistant for “the best running shoes for flat feet under $120”, and the assistant does not run one search. It runs several, reads the results and writes a single answer. If your category page answers only the head query, it may never be part of that answer.
TL;DR: Query fan-out is the technique AI search uses to split one prompt into several related sub-queries and combine the results. For ecommerce it means category pages are judged on how many sub-queries they cover: price, material, fit, comparison, delivery, returns. Find the sub-queries your page already matches, fill the gaps with attributes, a comparison table and concise answers, and keep classic rankings in place, because those still drive citations.
The stakes are real. Pew Research found in 2025 that when Google showed an AI summary, users clicked a traditional result in 8% of visits, against 15% when there was no summary. Visibility now depends on being inside the answer as well as under it.
What is query fan-out in AI search?
Query fan-out means an AI search system turns one prompt into several searches, runs them at the same time and merges what it finds. Google describes the technique in its AI Mode announcement as issuing multiple related searches concurrently across subtopics and multiple data sources, then combining the results into one response.
Take the prompt “best running shoes for flat feet under $120”. As an illustration, the system might look for stability running shoes, shoes with arch support under $120, reviews from runners with flat feet, wide-fit options, and return policies. These sub-queries are our example, not a published list, and that is the point: nobody outside the system knows the exact set, and it changes.
How can you see which fan-out queries your category pages already match?
You cannot read the sub-queries directly, but you can approximate them. Search Console is the first place to look: filter for long, question-like queries that already produce impressions for a category page, because shoppers phrase sub-queries the same way. People Also Ask boxes add more. A third method is to run the same prompt in an AI search mode several times and write down which angles the answer covers.
Expect the results to move. Surfer’s analysis of 10,000 keywords produced 33,000 fan-out queries, and only about 27% of them repeated across runs of the same prompt; 66% appeared just once. Optimize for the types of sub-queries, not for exact strings.
Search Console only shows the queries you already get impressions for, so the gaps are invisible there. JetOctopus, an SEO tool for eCommerce sites, keeps 16+ months of Search Console history next to crawl and log data, includes a Fan-Out Queries report, and its AI Page Optimizer checks product and category pages against the queries AI systems fan out into.
Why does fan-out matter for ecommerce visibility?
It changes what you are competing for. Two sets of data show how.
· Clicks are shrinking where AI summaries appear. In Pew’s 2025 study of 900 U.S. adults and 68,879 Google searches, users clicked a traditional result in 8% of visits with an AI summary and 15% without one. Clicks on links inside the summary happened in about 1% of visits, and sessions ended in 26% of visits with a summary against 16% without.
· Sub-query coverage correlates with citations. In Surfer’s analysis of 173,902 URLs, pages that ranked for the main query and at least one fan-out query were 161% more likely to be cited in AI Overviews. And 67.82% of cited pages did not rank in the top 10 for the query or any fan-out query, which means a page can be cited without a top-10 position.
Surfer is a vendor and the finding is a correlation, not proof that covering sub-queries causes citations. The practical reading is modest but useful: coverage of related questions helps, and it is not limited to pages already in the top 10.
What should a category page contain to cover more sub-queries?
Treat every category page as the answer to a cluster of questions, not one keyword. The table maps common sub-query types to page elements, using running shoes as the example category.
| Sub-query type | What the shopper asks | What the page needs | Running shoes example |
|---|---|---|---|
| Budget | What fits my price range? | Price bands, filter-friendly price copy | Best picks under $120 and under $80 |
| Material and specs | What is it made of? | Structured attributes in the product grid and copy | Mesh upper, drop in mm, weight |
| Fit and compatibility | Will it work for me? | Sizing guidance, use-case notes | Arch support level, wide sizes, flat feet |
| Comparison | How does A compare to B? | Short comparison table of top products | Cushioned vs stability models |
| Delivery and returns | What if it does not fit? | Shipping times and return window in plain text | 30-day returns, free exchanges |
| Reviews and proof | Do people like it? | Rating summaries, review snippets | Average rating and common complaints |
A short before-and-after shows the idea. A thin intro says “Our running shoes collection offers great quality at low prices.” A fuller one says which shoes suit flat feet, names the price bands, states the return window and links to a comparison of cushioned and stability models. The second version answers four sub-queries in about 80 words, and a shopper can scan it as easily as a model can.
Add Product schema so attributes are machine-readable, and answer four to six common questions in a short FAQ block at the bottom of the page. Keep each answer specific: a number, a material, a condition.
Which category pages should you fix first?
Start where a gain is most likely to matter commercially. Three filters are enough for a first pass.
· Revenue. Take the three to five categories that bring the most orders. Improvements there pay back fastest.
· Impressions without clicks. In Search Console, find category pages with plenty of impressions for question-style queries and a weak click rate. They are already visible for related questions but do not answer them well.
· Thin category copy. Pages that consist of a product grid and a two-line intro give an AI system little to quote, however good the products are.
Leave the long tail of small categories for later. A handful of well-covered pages teaches you more than fifty lightly edited ones, and it gives you a template for the rest.
A step-by-step workflow for a category page
1. Pick the three categories that bring the most revenue.
2. Collect 20–30 question-style queries for each from Search Console and People Also Ask.
3. Group them by sub-query type: budget, specs, fit, comparison, delivery, proof.
4. Check which groups your page does not answer.
5. Add the missing attributes, a comparison table, an FAQ block and Product schema.
6. After four to six weeks, review impressions for the new queries and check whether the page shows up in AI answers for your test prompts.
Conclusion
Query fan-out does not replace SEO. It adds a second test: after ranking, a page must also answer the questions around the main query. Start with your three most important categories, map the sub-queries, and fill the gaps with specific facts.
Keep expectations realistic. The research behind fan-out is young, the sub-queries change from run to run, and the best evidence is correlation. What you can control is how completely your pages answer the questions shoppers ask around a purchase, and that is worth doing whichever way the technology changes.
FAQ
How many sub-queries does AI search generate for one prompt?
There is no fixed number, and the set changes between runs. In Surfer’s analysis, 10,000 keywords produced 33,000 fan-out queries, and only about 27% repeated across runs of the same prompt, while 66% appeared once. That is why it is better to cover types of sub-queries (budget, fit, comparison) than to chase exact phrases.
Do I need to rank in the top 10 to be cited by AI Overviews?
No. In Surfer’s data, 67.82% of cited pages did not rank in the top 10 for the main query or any fan-out query, and the share was 45.86% for the top three visible citations. Pages that ranked for both the main query and a fan-out were 161% more likely to be cited. This is a correlation from a vendor study, not proof of cause.
How do I find fan-out queries for a product category?
Combine three sources: question-style queries with impressions in Search Console, People Also Ask boxes, and repeated runs of the same prompt in an AI search mode. Run each prompt at least three times and keep the angles that repeat; that is our recommendation, since the sub-queries vary. A fan-out report in an SEO platform can add a fourth source.
What should a category page include to match more sub-queries?
Structured attributes (material, size, compatibility), price bands, a short comparison table, delivery and return terms, four to six specific FAQ answers and Product schema. For a running shoe category that means arch support level, drop and weight, a cushioned-versus-stability comparison and the return window, each stated as a fact.
Does fan-out optimization replace classic SEO?
No. Pew found users click a traditional result in 8% of visits with an AI summary against 15% without, so organic positions still deliver clicks, and Surfer’s data links ranking for fan-out queries with a higher chance of citation. Keep your base rankings and layer sub-query coverage on top.
