Account

What new tool do you want?

Web & SEO

Keyword Clustering Tool

Group a keyword list into topic clusters by shared terms — to plan pages, avoid cannibalization and build topic hubs. Runs offline in your browser.

How to use

  1. 1Paste keywords, one per line (from any keyword tool).
  2. 2Adjust similarity, then copy the clusters — each cluster is a candidate page.

How it's calculated

Keywords are reduced to meaningful word stems (stop words removed). Two keywords join the same cluster when the share of stems they have in common (Jaccard similarity) meets the threshold.

Frequently asked questions

Why cluster keywords?

Keywords with the same intent should be targeted by one page. Clustering shows which terms belong together and which need their own page.

How does this differ from SERP-based clustering?

Commercial tools compare Google results for each keyword. This tool uses word overlap — fast, free and private, but less precise for synonyms.

What similarity should I use?

0.5 is a good start. Raise it for tighter clusters, lower it to group more broadly.

Why stemming and stop-word removal happen before any comparison

Comparing 'best running shoes' and 'running shoes for women' word-for-word would miss that they share 'running' and 'shoes' as their real topical core, buried among words like 'best' and 'for' that carry almost no topical meaning. Stripping stop words (a, the, for, best, top...) and reducing remaining words to stems (running shoes keeps its core even if a keyword said 'runs' or 'runner') means the similarity comparison is measuring actual topic overlap, not incidental word choice.

Worked example: Keyword Clustering Tool

Clustering 'best running shoes', 'running shoes for women', 'trail running shoes', 'marathon training plan' and 'beginner marathon training' at a 0.5 similarity threshold produces two clusters: the three running-shoes keywords group together since they share the stems 'run' and 'shoe' at a high enough overlap ratio, while the two marathon-training keywords form a separate cluster around 'marathon' and 'train'. Raise the threshold to 0.7 instead, and 'trail running shoes' might split into its own cluster if its stem overlap with the other two running-shoe keywords falls just under the stricter bar — tightening the threshold trades fewer, larger clusters for more, smaller, tighter ones.

Why greedy first-match clustering isn't the same as finding the globally best grouping

The algorithm processes keywords in order and attaches each one to whichever existing cluster it matches best (if above the threshold), creating a new cluster otherwise — it never goes back to reconsider earlier assignments once later keywords reveal a different, perhaps better grouping. This is fast and fully private since it runs in the browser with no server round-trip, but it means keyword order can occasionally affect the final clusters in edge cases, unlike a more exhaustive clustering approach that considers every possible grouping before deciding.

Why word-overlap clustering misses synonyms that SERP-based tools catch

This tool's own FAQ is upfront about its real limitation: it clusters by literal shared word stems, so 'running shoes' and 'jogging sneakers' — synonymous in meaning but sharing no word roots at all — would never cluster together here, while a commercial tool that clusters based on actual overlapping Google results for each keyword would likely group them correctly, since Google's own ranking already recognizes the semantic relationship. Trading that precision for being fast, free and entirely private is the explicit design choice this tool makes.

Why a lone, unrelated keyword still gets its own one-item cluster

Every keyword that doesn't clear the similarity threshold against any existing cluster becomes the seed of a brand-new cluster of exactly one, rather than being dropped or lumped into a miscellaneous bucket — so pasting in a list where only three keywords actually share a topic among twenty will still return eighteen clusters, most of them containing a single orphaned keyword. That's useful signal, not noise: a cluster of one is the tool telling you that keyword doesn't yet have company, which for content planning often means either it deserves its own dedicated page or it needs more related keywords researched before it's worth targeting at all.

Related tools

More: Web & SEO