Guide
How to cluster keywords (and why most people over-cluster)
Some links here are affiliate links. If you buy through one, we earn a commission at no extra cost to you. It never changes where a tool places —here is how we score.
Clustering is the step that decides whether a content plan works, and it is almost always done with the wrong test.
The only test that matters
Two keywords belong on the same page when the pages currently ranking for them are largely the same pages. That is it.
It is not about how similar the words look. “Best CRM software” and “CRM tools” look different and usually share a SERP — one page serves both. “CRM pricing” and “CRM for nonprofits” look like siblings and return almost entirely different results — two pages, no argument.
Google has already decided which queries it treats as the same job. Clustering by SERP overlap is reading that decision instead of guessing at it.
The two mistakes
Over-clustering is splitting one intent across several pages because the keywords read differently. You end up with four thin pages competing with each other, none of which covers the intent completely. This is the more common failure and the more damaging one.
Under-clustering is forcing distinct intents onto one page — usually to hit a word count — so the page half-serves several searchers and fully serves none.
If you are unsure, over-clustering is the worse error. A single complete page beats four partial ones almost every time.
Where volume belongs
Not in the clustering step. Volume tells you which clusters to write first; it should never decide which keywords go together. Cluster by behaviour, prioritise by volume, in that order.
Check against your own archive first
Before publishing anything new, run your existing URLs against the map. The most common cannibalisation on any established site is a new page competing with an old one nobody remembered.
Updating the page that already has some authority beats publishing a rival to it, and it is faster.
Frequently asked questions
How many keywords should be in a cluster?
However many share a SERP - sometimes three, sometimes forty. A fixed target is the wrong instinct. If your tool produces evenly sized clusters, it is grouping by word similarity rather than by search behaviour.
Do I need a clustering tool?
For a few hundred keywords, no - a spreadsheet and manual SERP checks work. Past a thousand, checking overlap by hand stops being realistic and a tool pays for itself. Semrush, Keyword Insights and Frase all do this.
What is over-clustering?
Splitting one intent across several pages because the keywords look different. It is the most common clustering mistake and it guarantees the pages compete with each other, which is why sites with hundreds of thin pages often rank worse than sites with dozens of complete ones.
Should each cluster become a blog post?
No. Clusters map to intents, and different intents want different page types - a comparison table, a product page, a calculator, a guide. Deciding the format after you have named the intent produces better pages than defaulting everything to an article.