Summary

This keyword clustering tool groups a pasted keyword list by lexical overlap: the significant words phrases share, not their live search rankings. Pick Loose, Balanced, or Tight sensitivity to control how much overlap counts as a match, and each cluster surfaces its shortest keyword as a starting pillar suggestion. It runs entirely client-side, so nothing you paste leaves the browser, and it is meant as a fast first pass before you verify clusters against real SERP overlap data.

A free keyword clustering tool that groups by shared words

Paste a keyword list and this keyword clustering tool groups the phrases that share enough significant words, right in your browser. No SERP data pulled, no signup, no black box.

Try the tool

Keyword clustering tool

Paste your keyword list below, one phrase per line. The tool groups phrases that share enough significant words, then flags the shortest phrase in each group as a starting pillar suggestion. Nothing leaves your browser.

Works on raw exports from Google Keyword Planner, Ahrefs, Semrush, or Search Console. One phrase per line.

How it works

What the clustering actually measures

Shared words, not shared rankings

Each keyword is broken into its significant words. Common connectors like “best” or “for” are dropped first. Two keywords land in the same cluster when enough of their remaining words overlap.

You set the overlap threshold

Loose groups by a shared topic, roughly a fifth of the significant words in common. Balanced needs about a third and is the default. Tight only merges near-duplicates, at least half the words shared.

Shortest keyword becomes the pillar

Inside each cluster, the shortest phrase is flagged as a starting pillar suggestion, a rough head-term heuristic. Confirm it against real search volume before you write a brief around it.

Use it in practice

A fast first pass, not a final answer

Export a raw keyword list from Google Keyword Planner, Search Console, or an Ahrefs or Semrush report, then paste it above with the volume and CPC columns stripped out. Start on Balanced, the default sensitivity, and switch to Tight if everything collapses into one giant cluster, or Loose if nothing groups at all. Treat every cluster as a hypothesis: check what actually ranks for the pillar keyword before you commit a content brief to the group.

  • Works from any plain keyword export, no formatting required
  • Adjustable sensitivity instead of one fixed rule
  • Runs client-side: paste, read, close the tab
Colored dot stickers on a cork board grouped into clusters connected by string

Common questions

Is this free?
Yes. The clustering runs in your browser with plain JavaScript. No account, no usage limit, no export paywall.
Does it use Google Search Console or live rank data?
No. It clusters by the words shared inside the keyword phrases themselves, what is usually called lexical overlap, not by which pages actually rank for them. SERP overlap clustering, which compares the ranking URLs for each keyword, is more accurate but needs live rank-tracking data this tool does not pull.
Why did two clearly related keywords end up in separate groups?
The tool compares word stems, not full meaning. “Clean running shoes” and “running shoe cleaning tips” share the same idea but not enough matching stems at the Balanced setting, so they land in separate groups until you switch to Loose.
What does the pillar suggestion actually mean?
Inside a cluster, the shortest keyword is flagged as a starting pillar, since shorter phrases tend to sit closer to the head term. It is a heuristic, not a volume-based recommendation. Confirm it against your keyword research tool before writing a brief.
Is my keyword list stored anywhere?
No. Everything runs client-side in your browser. Nothing is sent to a server, aside from the anonymous tool-run ping we use for our own usage stats.
How many keywords can I paste in?
There is no hard cap, but the tool compares every pair of keywords, so a few thousand lines will feel slower than a few hundred. Most content teams cluster in batches of 100 to 500 anyway, close to what one editorial brief can absorb.
Can this replace a paid SERP-clustering tool?
For a first pass on a small list, yes. For a 500-article programmatic build, we would still run a SERP-overlap check before finalizing the clusters. Lexical overlap is a fast starting filter, not a replacement for checking what actually ranks.

Clustering is the easy part

The harder part is turning each cluster into a page worth publishing. EsyBlog writes about how that process works end to end, cluster to brief to editorial review, without turning one topic into ten thin articles.

See how EsyBlog builds topic clusters