Keyword clustering tool
Paste a keyword list and group it into topic clusters in your browser. Free, no signup, no credits. It clusters by shared words, and the guide below explains when you need SERP data instead.
Paste keywords to cluster them.
4 steps, no signup
- 01
Paste your keywords
One per line, or CSV as keyword,volume. Exports from Search Console, Ahrefs, Semrush or Keyword Planner work once trimmed to those two columns.
- 02
Set the similarity threshold
Higher means tighter clusters that share more words. Lower means fewer, broader groups.
- 03
Review the clusters
Each group is named after its highest-volume keyword (or the shortest if there are no volumes), with a total volume.
- 04
Export to CSV
Download cluster, keyword and volume rows to plan one page per cluster in a spreadsheet.
Short answer: paste one keyword per line, optionally followed by a comma and its search volume. The tool groups keywords that share words (after light stemming and ignoring words like "how" and "the"), names each cluster after its biggest keyword, totals the volume and exports a CSV. It runs in your browser and does not look at Google results.
What this keyword clustering tool does
A keyword export gives you hundreds of lines like "running shoes", "best running shoes for women" and "women's running shoes". Many of those belong on one page. The keyword clustering tool groups them so you can see the topics in a list, total each group's search volume, and decide how many pages you actually need.
It runs entirely in your browser with no signup, credits or row limit. Many clustering tools need an account or charge per keyword because they pull live search results. This one does not, and the trade-off is explained honestly below.
How it clusters
- Tokenise. Each keyword is lowercased and split into words. Common words like "for", "the", "to" and "how" are dropped.
- Stem. A light English stemmer reduces endings, so "shoes" matches "shoe" and "running" matches "run".
- Compare. For each pair, it calculates Jaccard similarity: shared stems divided by all distinct stems across the two keywords.
- Group. Keywords are processed from highest volume to lowest. Each joins the existing cluster whose head keyword it is most similar to, if that score meets your threshold. Otherwise it starts a new cluster.
- Name. The head (first, highest-volume) keyword names the cluster. Without volumes, shorter keywords go first, so the head term leads.
Comparing against the cluster head, rather than chaining any two similar keywords, stops a single long-tail phrase from linking two unrelated topics together.
A worked example
Press Load sample and the tool clusters 20 running keywords at the default 0.50 threshold. It returns five clusters:
| Cluster | Keywords in it |
|---|---|
| running shoes | best running shoes, trail running shoes, running shoes for women, running shoes for flat feet, how to clean running shoes |
| running watch | best running watch, cheap running watch, garmin running watch, running watch with gps |
| marathon training plan | marathon training schedule, half marathon training plan, beginner marathon training plan |
| couch to 5k | couch to 5k app, couch to 5k plan |
| clean white sneakers | how to clean sneakers |
Four of those are exactly what you would group by hand. The fifth shows the main weakness of word-based clustering: "how to clean running shoes" joins the running shoes cluster because it shares two words with it, even though the intent (cleaning) matches the sneakers cluster. Raising the threshold would split it out, along with some keywords you want kept together. That is the point to check the results page in Google, as described below.
Lexical vs SERP clustering
| Lexical (this tool) | SERP overlap | |
|---|---|---|
| Groups keywords that | Share words | Return the same ranking URLs in Google |
| Data needed | Just your list | A live search results lookup for every keyword |
| Cost | Free, instant | Usually paid per keyword or credit |
| Catches synonyms | No ("cheap flights" vs "budget airfare" stay apart) | Yes, if Google ranks the same pages |
| Separates intent | Only when the words differ | Better: "apple pie" and "apple pie recipe" may still differ |
| Best for | First pass, big lists, site structure drafts | Final decisions on which keywords share a page |
The sensible workflow is both: cluster lexically to turn 2,000 lines into 150 groups, then search the head term of any group you are unsure about and look at what ranks. If the same pages rank for two clusters, merge them.
Why clusters matter for SEO
Google's guidance on helpful content warns against producing lots of content on many topics in the hope that some of it performs, and asks whether a page leaves readers feeling they have learned enough. One thorough page per cluster fits that better than ten thin pages that each target one wording. Google's spam policies also describe doorway pages, near-identical pages made for slightly different searches, as spam.
Tips for better clusters
- Remove brand and location modifiers first if they are not part of your plan, or they will form their own clusters.
- Clean the list: delete duplicates and irrelevant terms before clustering, not after.
- Try two thresholds. 0.5 for page-level groups, 0.3 for broad topic hubs.
- Watch intent words: "buy", "price", "how to" and "vs" usually mean different pages even inside one cluster.
Troubleshooting
- Everything is in one cluster: the list shares a seed word. Raise the threshold.
- Every keyword is alone: lower the threshold, or the list may genuinely be unrelated.
- Volumes show as 0: make sure the line is
keyword,volume. Thousands separators like "1,200" are fine; if the keyword itself contains a comma, wrap it in quotes. - Non-English keywords: grouping works, but stemming and stopwords are English only, so set a lower threshold.
Related tools
Once you have clusters, the AI Overview checker gives a rough read on which cluster heads are likely to show an AI answer. The keyword density checker shows how a draft page uses its cluster's terms, and the SERP snippet preview helps write the title and description for the page that targets it.
Privacy
Your keyword list stays in your browser. The last list is saved in this browser's local storage for convenience, never on a server, and Clear deletes it.
Last checked and updated: . Everything on this page runs in your browser; nothing you enter is sent anywhere.