A shopper asks ChatGPT for "the best sunscreen for sensitive skin." Behind that one sentence, the model quietly fires off ten or twelve smaller questions before it writes a single word back. Your product either answers most of them or it answers almost none. That hidden step decides whether you make the recommendation list.
TL;DR: AI shopping engines use query fan-out to break one shopper question into 8-12 sub-queries, then assemble an answer from whichever pages satisfy those sub-queries. Product pages with five or six attributes lose because they only match the surface query. Pages with deep, structured attribute data match the sub-queries too. Surfer SEO found you are 161% more likely to be cited when you rank for fan-out sub-queries, and 67.82% of AI-cited pages do not rank in the top 10 organic results at all.
Most retailers still think about AI search the way they thought about Google ten years ago: one query, one ranked list, win the top spot. That mental model is now wrong, and it is quietly costing you placements you will never see in any analytics dashboard. The mechanism behind the disappearance has a name, and Google uses it openly.
What is query fan-out in AI shopping?
Query fan-out is the process where an AI search system takes a single user query and decomposes it into multiple parallel sub-queries, runs each one against its index, and then synthesizes the results into one answer. Google introduced the term at I/O 2025, describing AI Mode as "breaking down your question into subtopics and issuing a multitude of queries simultaneously".
The numbers are bigger than most merchants assume. A typical AI Mode or ChatGPT shopping question fires 8 to 12 sub-queries. A deep-research request can fire hundreds. According to iPullRank's analysis, AI-search queries average 70 to 80 words of intent versus 3 to 4 words for classic search, because the engine is expanding your short question into a far richer set of needs you never typed.
The practical effect: the engine is no longer matching your page to the words the shopper used. It is matching your page to the words the shopper meant.
How fan-out plays out for a real shopping query
Take "best sunscreen for sensitive skin" and watch it decompose. A modern engine does not search for that phrase. It generates a fan of sub-queries that might look like this:
- Is mineral or chemical sunscreen better for sensitive skin?
- Which sunscreens are fragrance-free and non-comedogenic?
- What SPF is recommended for daily facial use?
- Are these formulas reef-safe?
- Which sunscreens are safe for rosacea or eczema?
- What is the texture and finish under makeup?
- Are there pediatric or baby-safe options?
Each sub-query gets answered from whichever pages contain that specific information. Your sunscreen might be the perfect product for sensitive skin. But if your product page lists a title, a price, two marketing sentences, and a hero image, it answers exactly one of those sub-queries and silently fails the other six. The engine assembles its recommendation from the pages that answered six of seven. You are not in that set.
This is why a product can be objectively excellent and still be invisible. The fan-out does not reward the best product. It rewards the best-described product.
Why thin product pages fail the fan-out
The core problem is an attribute gap. Most e-commerce product pages carry five to eight structured attributes. The sub-queries an AI engine generates routinely demand thirty or more: material composition, use case, compatibility, certifications, ingredient flags, size and fit specifics, care instructions, comparison dimensions, and audience suitability.
When AI search cites organic pages, 67.82% of those cited pages do not rank in the top 10 organic results for the main query (Surfer SEO). Ranking for the fan-out sub-queries, not the headline keyword, is what earns the citation.
That stat comes from Surfer SEO's study of 173,902 URLs, which also found that pages ranking for both the main query and its fan-out sub-queries are 161% more likely to be cited than pages ranking for the main query alone. The retrieval happens at the passage level, not the page level. The engine pulls the specific paragraph or attribute that answers each sub-query, so a page that buries its reef-safe certification in an image or omits it entirely simply does not exist for that sub-query.
For commerce specifically, this reframes optimization. The job is no longer to rank a page; it is to make every dimension of a product independently retrievable.
Query fan-out vs traditional SEO: what actually changed
The shift from page-level ranking to passage-level retrieval changes what "optimized" means. The table below maps the old model against the fan-out model so you can see where existing SEO work helps and where it leaves you exposed.
| Dimension | Traditional SEO | Query fan-out era |
|---|---|---|
| Unit of competition | The page | The passage or attribute |
| What you optimize for | One head keyword | 8-12 decomposed sub-queries |
| Winning signal | Top-10 organic rank | Coverage across sub-queries |
| Content depth needed | Enough to rank | Enough to answer every sub-intent |
| Product data required | 5-8 attributes | 30+ structured attributes |
| Failure mode | Page ranks too low | Page answers too few sub-queries |
Traditional SEO is not useless here. Ranking still feeds the retrieval pool. But ranking for one keyword is now a partial credit score on a test with twelve questions. This is the heart of generative engine optimization and answer engine optimization: you are optimizing for synthesis, not for a blue link.
How to measure your fan-out coverage
You cannot fix what you cannot see, and standard rank trackers do not show fan-out coverage. They report your position on the head query and miss the seven sub-queries you failed. The signal you want is your AI product found rate: across a realistic set of decomposed shopping queries, how often does your product actually surface?
This is the same external fan-out methodology that monitoring tools run when they test discovery across personas. The approach is straightforward to replicate manually for a single product: write out the eight to twelve sub-queries a shopper question would generate, then check whether your product page contains a clear, structured answer to each one. The gaps you find are the sub-queries you are losing today. Platforms built for this, including Paz.ai, an agentic commerce optimization platform, run that fan-out externally across ChatGPT, Google AI Mode, and Perplexity and report coverage at the sub-query level, but the manual version still surfaces the biggest holes.
What to Do This Week
- Pick your top revenue product and map its fan-out. Write down the 8-12 sub-queries a real shopper question about it would decompose into. Be honest about the ones your page does not answer.
- Audit attribute depth. Count the structured attributes on that product page. If you are under 15, you are almost certainly failing multiple sub-queries. Aim for 30 or more, covering use case, materials, compatibility, certifications, and audience.
- Move buried facts into structured, retrievable text. Certifications, ingredient flags, and fit details locked inside images or PDFs are invisible to passage-level retrieval. Put them in structured product data the engine can read.
- Check coverage, not just rank. Run your sub-queries through ChatGPT and Google AI Mode and note which ones surface your product. Track that found rate over time, not your keyword position.
- Prioritize by sub-query value. Some sub-queries carry more buying intent than others. Fix the high-intent gaps first using a product data enrichment workflow rather than rewriting everything at once.
Frequently Asked Questions
What is query fan-out?
Query fan-out is when an AI search engine takes one user query and splits it into 8-12 parallel sub-queries, runs each against its index, and synthesizes a single answer. Google described the mechanism for AI Mode at I/O 2025. It applies to ChatGPT and Perplexity shopping answers too.
Why does query fan-out matter for e-commerce?
Because the engine matches your product to the decomposed sub-queries, not to the words the shopper typed. A product page with thin attribute data answers only the surface query and fails the rest, so it gets left out of the recommendation even when the product itself is a strong fit.
How many attributes does a product page need for AI shopping?
Most pages carry 5 to 8 structured attributes. Fan-out sub-queries routinely demand 30 or more, spanning use case, materials, certifications, compatibility, and audience suitability. Pages closer to 30+ attributes answer more sub-queries and surface more often.
Is traditional SEO still useful in the fan-out era?
Yes, but it is incomplete. Ranking feeds the retrieval pool, yet 67.82% of AI-cited pages do not rank in the top 10 for the main query, per Surfer SEO. Coverage across sub-queries, not a single top rank, is what earns the citation.
How do I know which sub-queries I am losing?
Write out the sub-queries a shopper question would generate, then check whether your product page answers each one clearly. Tracking your found rate across those decomposed queries, rather than your position on the head term, reveals the specific gaps.
Does query fan-out apply to Google, ChatGPT, and Perplexity?
Yes. Google named the mechanism for AI Mode, and the same decompose-retrieve-synthesize pattern drives shopping answers in ChatGPT and Perplexity. The sub-query sets differ by engine, which is why coverage should be measured across all three rather than assumed from one.
The retailers who adapt to this will stop obsessing over a single keyword rank and start asking a harder question: of the dozen things an AI engine wants to know about my product, how many can it actually find? That number, not your organic position, is what determines whether you make the list when a shopper asks the machine what to buy.
How AI-ready are your products?
Check how ChatGPT, Google AI, and Perplexity evaluate any product page. Free score in 30 seconds.
