glossary · topic cluster
What a topic cluster is, and how to build one
A hub page, the narrower pages that answer questions inside the same subject, and the links between them. What a topic cluster is, what search engines and AI assistants do with one, and which parts are worth your time.
A topic cluster is a set of pages covering one subject together: a broad hub page plus narrower pages answering the specific questions inside it, linked to each other so the set reads as one body of coverage.
reviewed 2026-09-02 · by the IT Master editorial team · how we check facts
What a topic cluster looks like in practice
A cluster has three parts: one hub page covering a subject broadly, several spoke pages each answering one question inside it, and links running both ways between them.
The hub is the page you would send someone who knows nothing about the subject. It defines the terms, sets out the choices and links to the detail. The spokes are the detail: one page per question people actually search, written to finish that question rather than to tease the hub.
A worked example. A security retailer's hub is a buyer's guide to CCTV recorders. The spokes: how much storage a four-camera system needs, NVR versus DVR, PoE power budgets, what happens when a drive fails, whether the recorder has to sit on the same network as the cameras. Each spoke stands on its own in a results page. Together they cover the subject.
The links are load-bearing, not decoration. Hub to spoke tells a crawler the set exists. Spoke to hub, and spoke to sibling spoke, is what a reader follows when one answer raises the next question. Internal linking is where most clusters come apart.
A cluster is a hub, its spokes, and the links between them. Remove the links and you have a folder of unrelated articles.
Why clusters matter more than they used to
The structure is not new. Content teams have built hubs and spokes for years. Two things changed what it is worth.
Search engines weigh sites as well as single pages. An article on a subject the domain has never covered starts cold. The same article sitting among a dozen related pages that already earn impressions starts with context, and the cluster gives a crawler a route to it that does not depend on someone else linking to it.
Then AI assistants changed the shape of the demand. An assistant answering one question rewrites it into several narrower searches, fetches a handful of pages and quotes passages from what it fetched. A cluster gives you a page per sub-question, so more of those narrow searches land on something you have already written. LLM SEO covers that retrieval step in detail.
There is a quieter benefit. A cluster map makes cannibalisation visible. When two of your pages compete for one query they show up as two spokes with the same job, which is a decision to make rather than a mystery to investigate.
An AI assistant answers from the few pages it fetches, so how many of the narrow sub-questions a site covers decides whether it is in the set that gets read.
How IT Master builds clusters
Clusters here start from measured demand rather than a brainstorm. Topics come from DataForSEO keyword data plus the queries a site already earns impressions for in Search Console, so a spoke exists because people ask that question, not because the phrase looked promising.
Each draft is then checked before it can ship. Fact-check, novelty against the competing pages, E-E-A-T critique and AI-tell detection all run on models from a different vendor than the writer, with up to three revision rounds. Across 186 Standard-tier runs, 51% of drafts passed every check first time; the rest were revised or rejected.
What passes publishes to the site with internal links into the pages already there, JSON-LD schema, an FAQ and a hero image, and IndexNow tells the search engines it exists. A nightly loop watches Search Console and refreshes spokes that are decaying. Articles are grounded in the customer's own first-party data where it exists, such as product records, datasheets and support Q&A, which is the part a competitor cannot copy.
One honest limit: the engine will not draw your subject boundary for you. That call comes from the business. How it works sets out the pipeline.
Common misunderstandings
Five ideas that waste months:
- That the hub must come first. It rarely does. Publishing three spokes, then writing the hub once you know what they cover, beats a 4,000-word hub that links to nothing.
- That a cluster is a folder. URL nesting is not structure. A directory of pages with nothing linking between them is a filing system, and a crawler that lands on one of them has nowhere to go next.
- That more spokes are always better. A spoke restating the hub in different words splits one query between two of your pages. Add a spoke when there is a question the cluster cannot currently answer.
- That a cluster replaces topical authority. The cluster is the shape; authority is what the coverage may earn once the pages are indexed and genuinely useful. Building the shape without answering the questions earns nothing.
- That one cluster is a strategy. Most sites need a few deep clusters around what they sell, not thirty shallow ones. Deciding which subjects to skip is most of the work, and no tool should make that call for you.
questions people ask
How many pages does a topic cluster need?
There is no fixed number, and a tool that quotes one is guessing at your subject. The useful test is whether someone with a question inside the subject finds the answer on your site instead of leaving for another one. Count the questions you cannot answer yet rather than the articles you have already published. A narrow service runs out of genuine questions quickly; a broad category keeps producing them for years. When the list of unanswered questions is short, the cluster is deep enough and the next page belongs somewhere else.
What is the difference between a topic cluster and a pillar page?
The pillar page is one part of the cluster, not another name for it. It is the hub: the broad page that frames the subject, defines the terms and links out to the detail. The cluster is the whole set, meaning the pillar plus the spokes plus the links between them. People use the terms loosely, and pillar page often gets applied to any long article, but a pillar with nothing linking into it and nothing linked from it is just a long article. The links are what make the set a cluster.
Do topic clusters help with AI assistants and AI Overviews?
Mostly through retrieval rather than any ranking bonus. AI Overviews and assistants break a question into several narrower searches, fetch a small set of pages and quote from those, so the number of sub-questions your site answers decides how often you are in that set. Consistency helps as well: when pages from one site disagree on a fact, an assistant has reason to prefer a source whose pages agree with each other. None of this buys placement. The free [AI visibility checker](/tools/ai-visibility-checker) shows what assistants currently say about a brand.
How do I map a cluster without guessing?
Start from what people search, not from what you want to write. Pull the queries your site already gets impressions for in Search Console, add keyword data for the subject, then group by the underlying question rather than by shared words. Two phrasings of one problem are one page, not two. Each surviving group becomes a spoke, and anything that only makes sense as background belongs in the hub. Re-check the map every few months, because the questions move and a spoke that answered last year's version drifts.
related
See what search engines and AI assistants find on your site
Free, no account. Type your address and we show you what is missing and what we would write first.