Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

SEO › SEO Content & Strategy

Topic Clusters

Topic clusters explained: defining scope from keyword clusters, the hub and supporting pages, the linking pattern, build order and measurement.
Guideintermediate8 min readUpdated

The model

Three elements:

  • The hub (also called a pillar page): one page that covers the subject broadly — what it is, why it matters, its parts, how they fit — and owns the broad query. It links to every supporting page and is the page external links to the subject should land on.
  • Supporting pages: one per specific keyword cluster within the subject, each covering its part in depth and owning its own query. Each links back to the hub and to the siblings it relates to.
  • The links: hub to every supporting page; every supporting page to the hub; supporting pages to each other where the content connects. The internal linking guide covers the mechanics; the point here is that the pattern is deliberate, not incidental. The hub's other name invites confusion with editorial content pillars, which are themes rather than pages, and the comparison of content pillars and topic clusters sets out how the two relate.

Why it works, in documented terms: Google's starter guide describes internal links as how it discovers pages and understands their relationships, and anchor text as a signal of what the linked page is about. A hub linked from twenty relevant pages with descriptive anchors is clearly the site's main page on the subject; a supporting page linked from the hub and its siblings is clearly part of that coverage. The industry's shorthand for the result is topical authority; the mechanism is relevance and link structure.

Defining a cluster's scope

The scope comes from research, not from the subject's dictionary definition:

  1. Take the keyword clusters for the subject area from the clustering guide, with intents and demand.
  2. Identify the broad cluster — the head term and its variants — which the hub will own.
  3. Group the specific clusters by sub-topic. Each group that one page can serve becomes a supporting page; a sub-topic with many clusters of its own becomes a sub-hub with its own supporting pages.
  4. Check intent across the cluster. Informational clusters usually belong together; a transactional cluster in the same subject (the product or service page) is linked from the cluster but is a different page type, and the hub links to it as the next step rather than absorbing it.
  5. Set the boundary. Clusters that belong to an adjacent subject go to that subject's cluster and are linked across. A cluster that tries to own everything adjacent becomes unfocused.

The output is a cluster definition: hub (URL, head term), supporting pages (URL, cluster, format), sub-hubs if any, adjacent clusters to link, and the next-step page. Where a tool proposed the sub-topic groups by meaning, check them against intent and the results pages before fixing the boundary; how embeddings group similar text explains why groups built on meaning and groups built on rankings can differ.

The hub page

A hub is substantive. It explains the subject at the level the broad query wants — which the search intent guide reads from the results page (SERP) — and it links to each part with a sentence of context, not a bare list. Its structure usually follows the subject's parts, with a section per supporting page, and it carries the subject's definitions, its frameworks, and the questions whose answers are short enough not to need their own page. A hub that is a list of links is a category page and ranks like one. Where definitions span several clusters, a glossary can hold them in one place, and the tutorial on building a glossary links each entry to the guide that teaches it.

This site's cluster hubs are an example of the pattern: each section hub explains its subject and introduces its guides, and each guide links back and across.

Supporting pages

Each supporting page owns exactly one keyword cluster, in the format the cluster's results page demands, written to the brief for that cluster. It links to the hub with an anchor that names the subject, and to the sibling pages its content touches with anchors that name their topics. It does not restate the hub; it goes deeper on its part and refers up for the rest. Where supporting pages share one shape and differ by a variable, such as a role or a city, the programmatic SEO guide covers building them from a template and data without producing thin pages.

Build order

Clusters are built in an order that produces value early:

  1. The hub first, even in a shorter form, so that supporting pages have something to link to and the broad query has an owner.
  2. The supporting pages in priority order from the keyword map: those closest to the business and those with the best chance first.
  3. Links as each page publishes: from the hub (which is updated with each new section), from the siblings that already exist, and from any older content on the site that mentions the topic.
  4. The hub expanded as the supporting pages accumulate, so it reflects the full coverage.
  5. External links earned for the hub through the link building methods, since it is the page that external references naturally point to and the one that passes the most to its cluster (Backlink).
Illustrative example — not a real result

A cluster for a home-battery installer

Keyword clusters around home batteries split into a broad cluster ("home battery storage"), informational clusters (how they work, sizing, costs, lifespan, pairing with solar, grid export, safety), a commercial cluster ("best home battery") and a transactional cluster (installation quotes). The hub owns the broad query and explains the subject with a section per part. Seven supporting guides own the informational clusters; a comparison page owns the commercial cluster; the installation service page is the next-step page, linked from the hub and from the costs and sizing guides. The hub is published first in a shorter form, the sizing and costs guides follow because they are closest to the quote, the rest over the quarter. Each guide links to the hub, to its related siblings, and to the service page where the content leads there. The installer's existing blog posts on the subject are consolidated into the new pages with redirects.

Measuring a cluster

A cluster is measured as a unit: total impressions and clicks across its pages, the hub's position for the broad query, each supporting page's position for its cluster, the links earned by the hub, and the conversions attributed to the cluster's pages. The SEO KPIs guide sets up the reporting; reading the cluster as a whole shows whether coverage is working even when single pages fluctuate.

Clusters and site architecture

A topic cluster is a content structure; the URL hierarchy in the site architecture guide is the technical one. They usually coincide — the hub at /subject/ and the supporting pages beneath it — but they need not: a cluster can span a blog and a resources section as long as the links hold. Where the site is being designed, aligning them is simpler for everyone.

Retrofitting clusters onto an existing site

Most sites arrive at the model with years of content already published. The retrofit:

  1. Inventory every indexable page with its ranking queries from Search Console, its page type, its inbound internal links and its external links.
  2. Assign each page to a subject from the keyword research, or to "none" — the pages that serve no cluster and no other purpose.
  3. Find the hub for each subject: the page that best covers the subject broadly, if one exists. Often there is none, and the site's deepest page on the subject is a supporting page masquerading as a hub.
  4. Map the gaps and the overlaps: clusters with no page, clusters with several pages (consolidation candidates per the cannibalisation guide), pages with no cluster (refresh, merge or remove).
  5. Build or designate the hub, then wire the links: hub to supporting pages with contextual sentences, supporting pages to hub and siblings. Old pages that mention the subject get a link to the hub.
  6. Consolidate the overlaps with redirects and remove or redirect the pages that serve nothing, so the cluster is the site's whole coverage of the subject rather than a layer on top of an archive.

The retrofit is usually where the largest early gains are, because the content already exists and the work is structure, links and consolidation.

Maintaining a cluster

A cluster is a living structure. Quarterly: new keyword clusters from the research review become new supporting pages or sections; the hub is updated to introduce them; results-page changes that merged two clusters prompt a consolidation, and ones that split a cluster prompt a new page; supporting pages that have lost impressions are refreshed under their URLs; links are checked so that every supporting page still has its hub link and at least two sibling links, and the hub links to everything. The measurement in the section above shows which clusters earn the maintenance.

When clusters are built to grow visits, organic search for website traffic sets them beside snippet, indexing and refresh work and judges the result on engaged non-brand visits.

A new page published outside any cluster often stays low in the results; diagnosing new content that does not rank covers that and the other common causes.

Before linking a whole archive into clusters, the internal linking test design shows how to measure what added contextual links do for under-linked pages against an unlinked control group.

Common mistakes

  • Hubs that are lists. No content, no ranking, no authority to pass.
  • Supporting pages that restate the hub. Cannibalisation inside the cluster.
  • Clusters defined by opinion rather than research. Pages for parts nobody searches, none for the parts they do.
  • Links left to chance. A cluster without the pattern is an archive.
  • Everything in one cluster. A subject so broad that nothing is covered in depth.
  • Publishing supporting pages before the hub. Nowhere to link up to.
  • Ignoring the next step. The cluster educates and never leads anywhere.

The content briefs tutorial writes the brief for each page in the cluster; the E-E-A-T guide sets the standard each page has to meet (E-E-A-T).

Frequently asked questions

Is topical authority a Google ranking factor?

Google does not use the term or document a factor by that name. What it documents is that it tries to understand whether a page and a site are relevant and reliable for a subject, and its guidance on helpful content asks whether the site has a primary purpose or focus. Covering a subject thoroughly and linking the coverage together is the practical way to be that site; the industry name for the outcome is topical authority.

How big should a cluster be?

As big as the subject's keyword clusters require. A narrow subject may be a hub and four pages; a broad one, a hub with sub-hubs and dozens of pages. The size follows the research, not a template.

What is the difference between a hub and a category page?

A category page lists; a hub explains and lists. Hubs carry substantive content on the subject as a whole and own the broad query. A blog category archive with a title and a list of posts is not a hub, and rarely ranks.

Terms used on this page

Frameworks

Where this page is used

Sources

  1. SEO Starter Guide — Google Search Central

Related