Topic Clusters: The Hub-and-Spoke Model Explained
A topic cluster is a group of pages that covers one subject completely: a hub page owning the broad head term, and spoke pages each owning a single subtopic, with every spoke linking to the hub and the hub linking back to every spoke. The point is coverage, not volume. A site publishing ten unconnected posts about a subject has ten opinions; a site publishing a hub and nine spokes, properly wired, has a claim on the subject itself. Topic clusters are how that claim gets made in a form search engines and answer engines can read — a connected set of pages, each with one job, rather than a pile of overlapping articles.
Watch: the hub-and-spoke model on a real site — hub at the centre of each cluster, spokes on the ring, and the links that actually connect them.
What a topic cluster is
The topic cluster model has one rule that does more work than all the others: every subtopic gets exactly one home. One page owns it, links to the hub, and is linked back. Readers can enter anywhere and navigate the whole subject; search engines see a connected graph that says this site does not have a page about the topic — it has the topic.
That rule is also the diagnostic. When two cluster pages both try to own the same subtopic they compete with each other before they compete with anyone else: one query pulls two URLs, each carrying half the internal links and half the external ones. When no page owns a subtopic, the cluster has nothing to say to a question people are actually asking.
Neither failure is visible from the pages themselves, which is why the check is structural rather than editorial. You find them by reading the subject as a list of subtopics and setting the spoke pages against the subtopics they own — duplicates and holes surface in the same pass.
Why clusters beat lone pages
Modern ranking is topical. An engine deciding who deserves the head term looks beyond the single page to the site's demonstrated depth: how much of the subject is covered, how coherently it is connected, and whether authority flows through the structure. A meshed cluster concentrates that evidence; scattered posts dissipate it.
The mechanism is not mysterious, and it is documented. Google will generally only follow a link that is a real <a> element with an href attribute, and it treats the visible anchor text as a description of what sits on the other end — see Google's link best practices. Its own guidance on link architecture makes the same point about discovery: crawlers do not use your search box or your dropdown menus, so a page's reachability is decided by the links you actually place. A cluster is simply that principle applied deliberately to one subject instead of left to chance.
Topic cluster strategy: choose the subject, then the pages
A topic cluster strategy starts one level above the page. Before anything is written you decide which subject the site intends to own, and you accept that owning it means covering its subtopics rather than picking off the two keywords with the best volume. That decision sets everything downstream: what the hub says, how many spokes exist, and which queries the cluster is allowed to chase. Topic clusters are commissioned, not collected.
This is where most content clusters go wrong. They are assembled backwards — from a list of posts already published, grouped after the fact into something cluster-shaped. Grouping existing pages does not create coverage; it labels whatever coverage happened by accident. Built forwards, the subject is mapped first, its subtopics are counted, and the pages are commissioned to fill it. The practical test is unglamorous: write down the subject, list every subtopic a knowledgeable reader would expect to find, and mark which ones already have a page. What is left unmarked is the build queue — and a mapped keyword universe is how that queue stops being guesswork.
How to build a topic cluster
- Start from the keyword map, not the sitemap. The head term defines the hub; its subtopics define the spokes; every page gets exactly one home so spokes never cannibalise the hub.
- Publish the hub first where you can. Spokes pointing at a hub that does not exist have nowhere to send authority.
- Wire it deliberately. Hub to every spoke, every spoke back to the hub, and sibling links only where a reader genuinely benefits — with varied, descriptive anchor text rather than one phrase stamped everywhere.
- Give each spoke a fixed brief. Scope decided in advance is the cheapest protection against two cluster pages converging on the same subtopic later.
A worked example, start to finish
Take email marketing as the subject. The head term becomes the hub: one page covering the whole discipline at breadth. Demand data then names the subtopics rather than a brainstorm — deliverability, subject lines, list segmentation, automation flows, A/B testing, and compliance. Each becomes exactly one spoke, and the hub links to all six while all six link back.
What you do not do is publish both "email subject line tips" and "how to write email subject lines" as separate posts. That is one subtopic with two owners, and the cluster will split its own signal between them for as long as both exist. The fix, once it has happened, is a merge and a redirect rather than a rewrite of both. Equally, if compliance has real search demand and no page, the cluster has a hole — and a hole is a brief, not a crisis.
How to measure a topic cluster
Topic clusters are unusually easy to measure, because the structure they impose is machine-readable. Score the mesh rather than eyeballing it: five figures describe a cluster's structural health: whether a hub exists, what share of spokes the hub links to, what share of spokes link back, how much sibling interlinking there is, and how many pages belonging to the subject are orphaned from the cluster entirely.
Those five convert a vague ambition into a ranked list of links to add and pages to write, and because they can be re-scored after every change, progress is measured rather than asserted. That scoring is exactly what our Cluster Mesh module runs across a whole site.
Keeping a cluster healthy
A cluster is not finished when it is published. Topic clusters degrade in three predictable ways, and all three are cheap to catch on a schedule. Spokes go stale, because the specifics inside them age faster than the subject does. Spokes converge, because writers add pages without checking which subtopic is already owned — the two-owner failure arriving quietly, months after the build. And orphans accumulate, because a new page published outside the wiring is invisible to the cluster it belongs to.
The maintenance pass is the build pass run again: re-read the subject as a list of subtopics, re-check ownership, re-score the mesh. What changes between passes is usually the subtopic list itself, as new demand appears and old questions stop being asked.
Frequently asked questions
What's the difference between a hub page and a pillar page?
Mostly vocabulary: both name the broad page that owns a topic's head term and links out to the specific subtopic pages. What matters is the contract rather than the label — the hub covers the topic wide, the spokes cover it deep, and the links run both ways.
How many spokes does a topic cluster need?
Enough to genuinely cover the subject's subtopics that have real search demand — commonly five to fifteen. The number is dictated by the keyword map, not by a quota. A subject with four real subtopics does not need a fifth page invented to hit a target.
What is a topic cluster in SEO?
It is a structure, not a format. The pages inside it can be guides, comparisons or FAQs of any length; what makes them a cluster is that each one owns a distinct part of the subject and the links between them say so. A set of pages is not a cluster because it shares a tag or a category — it is a cluster when the ownership is unambiguous and the wiring is real.
How do you build a topic cluster?
The order is the part people get wrong: publish the hub before the spokes wherever you can. Spokes pointing at a hub that does not exist yet have nowhere to send authority, and they tend to drift wide to compensate, which is how two of them end up claiming the same subtopic. If the hub has to come later, write the spokes to a fixed brief so their scope is decided in advance rather than negotiated after publication.
How do you find subtopics for a topic cluster?
From demand, not from a brainstorm. The queries people already run against the subject are the subtopic list: question forms, comparisons, and the narrower terms sitting beneath the head term. Search Console queries and the questions a results page surfaces for the head term will name subtopics you would not have guessed, and each one arrives with evidence that somebody is asking.
Do topic clusters help SEO?
They help by making coverage and internal linking legible at the same time. The structure adds no ranking power by itself; what it does is concentrate signals that would otherwise scatter, so links, anchor text and demonstrated depth all point at one arrangement of pages instead of a pile of unrelated posts. A badly built cluster helps nothing — the gain comes from the coverage being real.
How do you measure a topic cluster?
Fix the cheapest failure first. Orphans cost nothing to repair and return the most, because a page receiving no links from its own cluster is invisible to it. Missing spoke-to-hub links come next, then hub-to-spoke, then sibling interlinking, which is the refinement rather than the foundation. A cluster scoring well on coverage but badly on orphans is not a good cluster with a small flaw; it is a cluster that has not been wired.
What is a spoke page in SEO?
A spoke page is a cluster page that owns one subtopic in depth and points back to the hub. It is written to be the best answer to its own narrow question rather than a summary of the subject — that job belongs to the hub. A spoke that drifts wide starts competing with the page it exists to support.
Go deeper
Related reading: hub pages · the Cluster Mesh module · internal linking strategy. Or start at the learning hub.
See where your site stands. The free AI-powered SEO audit is your first telemetry read — request your free audit or book a strategy call.
See where your site stands. The free AI-powered SEO audit is your first telemetry read.
Request your free auditSEO pricing calculatorBook a strategy call
Keep exploring this topic
This guide is part of our Website Structure for SEO cluster.