Cluster Building / Fan-Out for AEO: How Big Is It In AI Search For 2026?

What is cluster building?

Cluster building is the practice of organizing content into a central hub page supported by a set of interlinked pages, each answering one specific sub-question within the topic. The hub covers the broad subject. The supporting pages go deep on the questions people (and AI agents) ask next.

The reason it matters for AEO comes down to how AI search works. Engines use a retrieval step called query fan-out, where a single, complex prompt is split into several distinct sub-queries. One analysis of more than 60,000 AI-generated queries found that a single question to ChatGPT or Gemini routinely triggers 8 to 10 parallel, highly specific searches before an answer is returned.

Take a behavioral health example. Someone asks an AI assistant about alcohol rehab in their city. Behind the scenes, the engine is also searching for insurance coverage, levels of care, treatment approaches, family involvement, and aftercare. A single service page can’t answer all of that without overwhelming the reader and agent. A cluster can.

How does cluster building rank against other AEO signals?

The best evidence we have comes from Cyrus Shepard of Zyppy, who published a meta-analysis in May 2026 that distilled 54 experiments, patents, and case studies into a scored list of 23 factors. The top factors were URL accessibility (9.5), search rank (9.4), fan-out rank (9.3), preview controls (9.2), and query-answer match (9.2).

“Fan-out rank” is essentially what cluster building is designed to improve. It reflects how well your content covers the related sub-questions an engine generates when it breaks down a query, rather than only the single headline keyword.

So cluster building sits near the top of the list, but two other ranking ractors are higher:

1. Can AI crawlers reach your pages? If they can’t, nothing else matters.
2. Do you rank in search at all? AI engines pull from search indexes, so classic SEO is the entry ticket.

Compare that with tactics that get a lot of attention but little support. In the same analysis, LLMs.txt scored just 2 out of 10, with no credible evidence that it influences AI citations.

Does cluster building overlap with SEO?

Short answer, it’s highly correlated. Topic clusters (the hub-and-spoke model) have been an SEO best practice for years because they build topical authority and help search engines understand what a site is about. AEO doesn’t replace that work. It just elevates the importance of that ranking factor.

The top three factors in Shepard’s analysis are all SEO fundamentals. As Shepard explained on a podcast, the strongest correlation with appearing in AI answers is how well a URL ranks for the main query plus all of its fan-out queries. One summary of the study boiled the thesis down to “win SEO, win AI citations, but with extra steps.”

The “extra steps” are what separate a traditional SEO cluster from one built for AI citation:

• Mapping to sub-questions, not just keywords. Fan-out queries are generated on the fly, and many have no measurable search volume. You plan around the questions a patient or buyer would ask next.
• Answer-first structure. Each page, and each section, opens with a direct answer that an AI engine can lift cleanly.
• Tight internal linking. Links between hub and supporting pages make the relationship between them obvious.

Is cluster building more important than SEO ranking for AI citations?

Ranking for the exact query still matters, but it matters less than it used to.

An Ahrefs study of 863,000 keywords and 4 million AI Overview URLs found that only 38% of cited pages also ranked in the top 10 for the same query. This is down thirty-eight percentage points from the 76% in Ahrefs’ July 2025 version of the study. The remaining citations split evenly between pages ranking in positions 11 to 100 (31.2%) and pages beyond position 100 (31.0%). A separate BrightEdge analysis put top-10 ranking overlap even lower, at about 17%.

One caveat worth knowing: Ahrefs has noted that its parsing method improved between the two studies, so part of the drop reflects better measurement. The direction is clear, though, and Ahrefs points to the cause: AI Overviews are leaning more on sources that surface in fan-out query results than on the original results page.

The most useful answer to the “cluster vs. ranking” question comes from a September 2026 study by Floyi, which analyzed 535,239 Google rankings and 354,955 AI citations across 42 topical maps. It found that the two work together, in different ways:

• On queries you already rank for, position wins. Higher-ranking sites were cited more often, and broader topical coverage added no clear advantage within the same position range.
• On queries you don’t rank for, coverage wins. Broader coverage improved a site’s odds of being cited on queries it didn’t rank for, on all four AI engines studied. That matters because 26.8% to 43.7% of AI citations went to sites that didn’t rank for the query being answered but did rank elsewhere in the same topic map.

In plain terms: ranking wins the individual matchup, while cluster building gets you into far more matchups. The sites with the broadest coverage weren’t better on any single query. Their advantage came from volume, since they were doing about the same thing many more times.

What do studies show about cluster building and AI citations?

The Floyi study is the most rigorous public data on this so far. Its key findings:

• Coverage outperformed domain strength. Topical coverage tracked AI citations far more closely than Ahrefs Domain Rating on every engine: 0.51 vs. 0.09 on AI Overviews, 0.48 vs. 0.16 on AI Mode, 0.33 vs. 0.15 on ChatGPT, and 0.36 vs. 0.11 on Gemini. Backlinks still matter for SEO, but topical breadth tracked AI citations much more closely.
• The gap between broad and thin coverage is huge. Sites ranking for at least half of a topic map’s topics were cited in 30.8% of AI Overviews on average, compared with 0.14% for sites ranking for less than 5%.
• Healthcare is the most AI-exposed category. In the Health and Wellness maps, 98.4% of queries returned an AI Overview, the highest of any industry studied. The broadest-coverage health sites were cited in 54% of those answers, versus 19.5% in e-commerce.

This is correlational: the authors state that it doesn’t prove broader coverage causes AI citations. It’s also built from curated topic maps, not a random sample of the web.

A note on the numbers you’ll see elsewhere: many articles repeat figures like “clusters earn 3x more AI citations.” Some may be accurate, but most come from vendors without published methodology. Treat them as directional at best.

What this means for healthcare and behavioral health marketers

Health queries trigger AI answers more than almost any other category, so the cost of thin content is higher here than anywhere. A few practical takeaways:

1. Fix the basics first. Make sure AI crawlers can access your site and that your core pages rank. Clusters amplify SEO; they don’t replace it.
2. Map the sub-questions before you write. For each service line, list the 10 to 15 questions a prospective patient or family member would ask: cost, insurance, levels of care, what a typical day looks like, what happens after discharge.
3. Restructure before you publish more. Often the fastest win is giving each sub-question its own heading and a direct answer on pages you already have.
4. Keep clinical accuracy non-negotiable. Content rewritten purely to chase citations can drift from what your clinicians would actually say. Medical review protects both patients and your credibility.
5. Measure coverage, not just rankings. Track a fixed set of 20 to 30 representative prompts monthly and watch the trend in how often you’re cited.

Cluster building isn’t a shortcut. It’s the steady work of becoming the most complete, trustworthy source on the topics your patients care about. The research suggests that’s exactly what AI engines reward.