Keyword clustering tools: turning a list into a page plan

The short version

Clustering is the step between having a keyword list and having a content plan: it decides which keywords belong on the same page. Get it wrong in one direction and you publish five thin pages competing with each other; get it wrong in the other and you write one bloated page that serves nobody.

Dedicated clustering is cheap relative to the mistake it prevents. Keyword Insights prices it at 1 credit per keyword, on a Basic plan of $58 a month, or $46.40 prepaid annually, carrying 10,000 credits a month. If your list is under a few hundred keywords, a spreadsheet and an afternoon will do the same job.

What a cluster actually is

A cluster is a set of keywords that a single page can satisfy. "Best keyword tool", "top keyword research tools" and "which keyword tool should I use" are one page. "Keyword tool pricing" is probably another. "What is keyword difficulty" is definitely another.

The reliable test is not how similar the words look — it is whether the same page would satisfy someone typing each of them. Two phrases sharing three words can want completely different pages, and two phrases sharing none can want the same one.

How the tools decide

There are two broad approaches. The first compares the actual search results for each keyword and groups terms whose results overlap, on the logic that if Google returns the same pages for two phrases, Google considers them the same need. The second compares meaning directly using language models.

Keyword Insights describes applying natural language processing "to find the semantic relationship between the clusters", and does not publish the threshold or algorithm in any detail you could evaluate. That is normal for this category and worth knowing: you are buying an output to inspect, not a method to audit.

What Keyword Insights costs to do this

Clustering costs 1 credit per keyword, which makes pricing a job trivial: five thousand keywords costs five thousand credits. The Basic plan's 10,000 credits a month therefore covers ten thousand clustered keywords, which is more than most small sites will ever hold.

The same credits also pay for other jobs at very different rates — 100 credits for a content brief, 200 for an article through the writing assistant, 1200 for one through the AI writing agent — so an allowance that looks enormous for clustering empties quickly if you use the writing features. There is no free tier; a $1 seven-day trial carries 5,000 one-off credits, enough to cluster 500 keywords and see whether the groups match your judgement.

What the general tools give you instead

Most keyword tools offer grouping of some kind, usually by shared term rather than by shared intent. Semrush's Keyword Magic Tool groups suggestions by topic, and Ahrefs lists keyword clusters among the features its Standard plan adds over Lite. These are useful for navigating a big list and they are not the same job: grouping by common word is a text operation, while clustering by intent is a judgement about search results.

For a list of two or three hundred keywords the gap barely matters, because you can read the whole list. It starts mattering somewhere around a thousand, which is also roughly where a dedicated tool starts being cheaper than your own time.

Doing it without buying anything

1. Sort the list alphabetically and read it. At a few hundred keywords this alone reveals most of the groups.
2. For any pair you are unsure about, search both phrases and compare the top five results. Heavy overlap means one page; no overlap means two.
3. Label every keyword with the kind of page it wants — a definition, a comparison, a how-to, a product page. Most bad clusters mix two kinds.
4. Name each group as the page title you would actually write. A group you cannot title is not a cluster yet.
5. Keep a list of the keywords that fit nowhere. They are usually either a new section or a sign the list needs another source.

How to tell a good cluster from a bad one

Whatever produced the groups, machine or spreadsheet, the same three checks catch most errors. First, can you write a single honest page title for the group? If the title needs an "and" joining two unrelated ideas, the cluster is two clusters.

Second, does the group mix kinds of page? A definition, a comparison and a how-to have different shapes, and a group containing all three usually wants splitting even if the words look similar. Third, is there a keyword in the group that a reader would be annoyed to land on this page for? One badly placed keyword is a sign the boundary is in the wrong place, and it is the cheapest error to fix before anything is written.

When it is worth paying

There is one more argument for buying rather than doing it by hand, and it has nothing to do with speed: a tool applies the same rule on a bad day as on a good one. Hand-clustering drifts, and a plan built over three sessions often contains three different ideas of what counts as one page.

Buy a clustering tool when the list is genuinely too big to read, when you are planning several sites at once, or when you need the grouping to be repeatable rather than a product of your mood on a Tuesday. Agencies hit all three; individual site owners often hit none.

The longer note on Keyword Insights covers its plans and limits properly. If you do not have a keyword list yet, clustering is the wrong purchase — the eight keyword research tools with their entry prices is the page for finding keywords in the first place.