A topic cluster is one page that covers a subject properly, surrounded by pages that each take one part of that subject further, all linked so a reader — and a crawler — can move between them without going back to a menu. That is the whole idea. Everything else is implementation detail.
It sounds close enough to a blog category that most teams assume they already have one. They almost never do, because a category is an accident of publishing history and a cluster is a decision made in advance.
The pillar is a page, not a landing page
The centre of a cluster should be the page you would send someone who said 'explain this to me properly'. It is long, because the subject is large. It is opinionated, because a page that refuses to make a recommendation is not an answer. And it links generously outward to the pages that go deeper on each part.
What it is not is a thin index page with six paragraphs and a list of links. Search engines have got good at spotting a page that exists to hold links rather than to answer a question, and readers were never fooled.
The supporting pages are where the specificity lives
Each supporting page should answer one question so completely that a reader has nothing left to search for. That usually means it is narrower than feels comfortable. 'How to price a service' is a pillar; 'what to do when a client asks for a discount at the last minute' is a supporting page, and the second one is far more likely to rank quickly.
The links are the part everyone skips
A cluster with no internal linking is just a folder. The links do three jobs: they tell search engines which pages belong together, they pass authority from whichever page earns attention to the ones that have not yet, and they keep a reader moving through your site rather than back to the results page.
The practical rule is that every supporting page links up to the pillar with descriptive anchor text, the pillar links down to every supporting page, and siblings link sideways where the connection is genuinely useful. Not because a plugin suggested it.
Three mistakes that turn a cluster back into a pile
- 1Choosing the subject by search volume alone. If you have no credibility on a subject, covering it completely still will not beat the sites that do. Pick something you have a right to win.
- 2Publishing the supporting pages and never finishing the pillar. This is the most common failure by a distance. The narrow pages rank a little, nothing connects them, and the subject never consolidates.
- 3Treating the cluster as finished. Coverage decays: competitors publish, the subject moves, your own best page ages. A cluster is maintained, not completed.
How large should a cluster be?
Small enough to finish. In practice, a pillar and five to nine supporting pages is enough to make a visible difference for most small and mid-sized businesses, and it is a quarter of work rather than a year of it.
The temptation is always to map thirty pages across four subjects, because the map looks impressive. Mapped work is not published work. Two finished clusters will beat four half-built ones every time, and you will know within a quarter whether the approach is working on your market.
What it looks like when it starts working
The first sign is usually not the pillar. It is a supporting page you had low expectations for, quietly picking up impressions for a dozen long questions you never targeted directly. That is the subject being recognised.
The second sign is that new pages in the same cluster rank faster than the early ones did. That is the compounding people mean when they say organic search pays back late. It is also why abandoning a cluster at month three is the most expensive decision available.
Worth remembering
- A cluster is a decision made in advance, not a category that emerged from your archive.
- Finish the pillar. A cluster without its centre never consolidates.
- Link up, down and sideways with descriptive anchor text — that is what makes it a cluster.
- Two finished clusters beat four mapped ones.
