Query fan-out is how AI search tools split one complex question into many smaller searches, then combine the results into a single answer. It changes content strategy because you can be cited for one sub-question without ranking for the main query. Topic clusters now need to cover the likely sub-questions, each answered clearly on a page or section built for it.
Topic clusters, a central pillar page linked to supporting pages on related subtopics, aren't new. What's new is that AI systems effectively build their own cluster on the fly for every complex question. If your content maps neatly onto the sub-questions they generate, you have many chances to be cited. If it's one sprawling page or a pile of near-duplicates, you have few. This is part of our wider guide to AEO, GEO and ranking in AI answers.
What is query fan-out, exactly?
It's the process of breaking a question into related sub-queries, running them in parallel and merging what comes back. Google has described this technique as part of how AI Mode works, and other search-backed assistants behave in a similar way.
Take a question a business owner might type:
"Should my 15-person real estate agency in Sydney build a custom CRM or use HubSpot?"
A fan-out system might quietly search for things like:
- custom CRM cost in Australia
- HubSpot pricing for small teams
- CRM features real estate agencies need
- custom CRM vs off-the-shelf pros and cons
- how long a custom CRM takes to build
- integrating a CRM with property listing portals
- data migration risks when switching CRM
Each of those is a separate retrieval. The final answer cites the pages that best answered each piece. Your page doesn't have to answer the whole question; it has to be the best answer to one of the pieces.
Why does fan-out change how you plan content?
Because visibility shifts from ranking for a few head keywords to owning the sub-questions buyers have along the way. A small business can't outrank big publishers for broad terms, but it can be the clearest source on specific, practical sub-questions in its niche.
Three implications:
- Coverage beats volume. Ten pages that each answer a distinct sub-question well are worth more than thirty overlapping pages.
- Sections matter, not just pages. Fan-out retrieves passages, so a well-labelled section inside a broader page can be cited for its own sub-question.
- Links between pages matter more. Internal links show search engines how your pages relate, which helps them understand you cover the whole topic.
How are fan-out-ready clusters different from old-style clusters?
Old clusters were built around keyword variations; fan-out-ready clusters are built around the questions a buyer asks on the way to a decision.
| Old-style topic cluster | Fan-out-ready cluster | |
|---|---|---|
| Starting point | A high-volume keyword | A real, complex customer question |
| Supporting pages | Keyword variations ("best CRM", "top CRM", "CRM software") | Distinct sub-questions (cost, timeline, comparisons, risks, integrations) |
| Page structure | Long intro, keyword in headings | Question-style headings with direct answers |
| Risk | Near-duplicate pages competing with each other | Gaps where a sub-question has no answer |
| Internal linking | Mostly spokes linking to the pillar | Pillar to spokes, spokes to related spokes, all with descriptive anchors |
| Success measure | Rankings for the head term | Citations and rankings across many sub-questions, plus enquiries |
How do you map a cluster for query fan-out?
Start from the hardest real question your buyers ask, list every sub-question it implies, and give each one a clear home. Here's the method, step by step.
- Pick a decision question. Not a keyword, a decision. "Custom CRM or off-the-shelf?" "Which website platform for my clinic?" Use real wording from sales calls and enquiries.
- Generate sub-questions. Ask yourself what someone needs to know to decide: cost, time, alternatives, risks, local factors, who it's not for, how to start. AI tools are useful here for brainstorming, but check the list against real customer questions.
- Group by intent. Cost questions together, comparison questions together, process questions together.
- Decide: page or section? Use the rules in the next section.
- Check existing content. You may already answer some sub-questions badly or in three places. Merge before you write anything new.
- Write the pillar as a navigable overview that answers the decision question briefly and links to each supporting page.
- Interlink deliberately. Every supporting page links to the pillar and to its closest siblings, using anchors that describe the destination.
Our own CRM content follows this shape: a broad comparison of custom CRM versus HubSpot, Zoho and Salesforce sits alongside a dedicated guide to how much a custom CRM costs to build, so the cost sub-question has its own clear answer rather than a paragraph buried in a comparison.
When should a sub-question get its own page?
Give it its own page when it has distinct search demand, enough to say for a genuinely useful answer, and a different intent from its neighbours. Otherwise, make it a clearly labelled section.
Use these decision rules:
- Own page if people search for it on its own ("custom CRM cost India"), it needs 1,000+ words to answer properly, or it needs its own table, calculator or examples.
- Section if it only makes sense in context, can be answered in 100–300 words, or would produce a thin page.
- FAQ entry if it's a quick clarification with a short, factual answer.
- Neither if it's a near-duplicate of something you already cover with a word swapped.
That last rule is where many sites get into trouble. We noindexed 167 templated city pages on our own site for exactly this reason: they were near-identical pages with a place name swapped, the pattern Google's spam policies call scaled content abuse. More pages is not more coverage if the pages say the same thing.
How should the pages in a cluster be written?
Each page should answer its own sub-question in the opening lines and stand on its own if it's the only page a reader, or an AI, sees. Fan-out means a visitor may arrive at any spoke without having read the pillar.
- Open with a direct answer to the page's question.
- Use question-style subheadings that mirror follow-ups.
- Include something specific: a range, a timeline, a table, an example.
- Restate essential context briefly instead of relying on "as we said in the previous article".
- Link to the pillar and siblings where a reader would naturally want more.
Our guide on writing content AI assistants quote goes deeper into making each passage self-contained.
How do you measure whether a cluster is working?
Track impressions and clicks across all the cluster's queries, citations for a sample of sub-questions, and enquiries from cluster pages. Head-term rankings alone will understate the value.
- In Search Console, filter by the cluster's URL folder or a list of its pages and watch total impressions and the number of distinct queries over time.
- Add four or five sub-questions from each cluster to your monthly AI visibility checks.
- In analytics, compare enquiries that start on cluster pages with the rest of the site.
Frequently asked questions
Is query fan-out only a Google thing?
Google has described it publicly for AI Mode, but other search-backed assistants also break complex questions into several searches before answering. Planning around sub-questions helps across all of them.
How many pages should a topic cluster have?
As many as there are distinct, useful sub-questions, and no more. For a small business that might be one pillar and six to twelve supporting pages per core service. Quality and distinctiveness matter far more than hitting a number.
Do I need to rewrite my existing blog to suit fan-out?
Usually not from scratch. Start by mapping existing posts to sub-questions, merging duplicates, filling clear gaps and adding question-style headings with direct answers. Restructuring often delivers more than writing new posts.
Can a single long page cover a whole cluster?
It can work for narrow topics if every section is clearly labelled and answers one sub-question directly. For broader topics with real search demand on each sub-question, separate pages tend to serve readers and search engines better.
How does internal linking help with AI search?
Internal links help search engines discover your pages and understand how they relate. A clear structure signals that you cover a topic thoroughly, which supports both rankings and the chance of being retrieved for related sub-questions.
Plan content around the questions buyers ask
We can map clusters from your real customer questions, merge overlapping pages and build the internal linking that ties them together. If you'd like your content reorganised for AI search, see our services or request a fixed-price quote.