Free keyword clustering tool
- No signup
- Free
- 5 AI runs a day
Keyword clustering turns a flat list into page jobs. This tool groups queries that share meaning signals, suggests a primary keyword and page type, and exports a CSV you can brief against. It is not live Google SERP overlap. It is a practical first map.
How to use the clustering tool
Paste at least ten keywords, one per line, or upload a CSV with a keyword column. Optional volume, difficulty, CPC, and current URL columns are detected from common Ahrefs, Semrush, and Search Console style headers. Cap is five hundred unique keywords.
Move the tightness slider left for broader groups and right for smaller, stricter groups. Cluster in the browser for a free local pass. Turn on AI labels when you want clearer cluster names, intents, and H1 suggestions. AI runs count toward the daily limit.
Read the verdict, then the topical map cards, then the table. Start-with flags are a simple volume-versus-difficulty suggestion, not a traffic promise. Export CSV when you are ready to put the plan in a sheet.
What the groups mean
Each cluster is meant to become one URL unless you deliberately split for a good reason. Mixed-intent flags mean the group may contain both how-to and hire-me language. Split those before you write. Unclustered rows are often long-tail ideas that can live as FAQs or supporting sections later.
Page types are labels for briefing: service, location, product category, blog guide, comparison, tool, or FAQ. They are suggestions based on wording and optional AI labelling. Your site architecture can override them when a real template already exists.
| Signal | What I do with it |
|---|---|
| Primary keyword | Becomes the working title target for that URL |
| Intent | Stops a blog brief from stealing a commercial cluster |
| Page type | Picks the template and proof blocks |
| Start-with | Order of drafting when capacity is limited |
| Unclustered | Park for FAQs or a later pass |
SERP overlap versus wording clusters
SERP-overlap clustering asks whether Google already shows the same results for two queries. That needs a SERP API and budget. Wording clusters are cheaper and still useful for first drafts of a topical map. For money pages, spot-check the top results yourself before you merge two clusters into one URL.
If two keywords share words but the results diverge, keep two pages or tighten the brief. If two keywords look different in a spreadsheet but share the same results, consider merging. This tool will not make that SERP call for you in the MVP.
From clusters to a topical map
Group service and location clusters under commercial hubs. Put guides under supporting content. Link from guides to the money page that should win the hire or buy query. Use the cannibalization checker after you publish so two new URLs do not fight for the same head term.
When a current URL column is present, treat existing pages as candidates to update before you invent a new slug. Updating a URL that already has impressions is usually cheaper than launching a duplicate.
Briefing writers from the export
Every brief should name the primary keyword, the page type, the words that belong on this URL, and the words that belong elsewhere. Paste the cluster keyword list into the brief. Add the H1 suggestion as a draft, not a law. Add proof the page must include: prices, process, photos, FAQs, or comparison criteria.
Do not force every keyword into a heading. Secondary phrases can live in body copy and FAQs. The cluster is a membership list, not a stuffing checklist.
Add internal link targets in the brief. A new guide should name the service URL it supports. A new service page should name the two guides that will point to it after publish. Clustering without linking leaves orphan pages.
When legal or brand review slows publishing, ship the money page first and the supporting guides next. The start-with list exists for capacity constraints, not as a creative ranking.
Revisit clusters after Search Console shows queries you did not expect. New phrases often belong in an existing cluster as FAQ lines instead of new URLs.
How I turn clusters into a content calendar
I export the CSV and sort by start-with and volume. The first five pages are usually a mix of one money page and a few supporting guides, not five disconnected blog ideas. Commercial clusters get briefs with proof requirements. Informational clusters get outlines that link upward.
I open the cannibalization checker after anything ships that might overlap an existing URL. Clustering prevents some collisions. It does not see what already ranks in Search Console. Both tools belong in the same workflow.
When AI names a cluster poorly, I rename it in the sheet. Model labels are a draft. Human editors own the final page job. If a cluster mixes near-me and how-to language, I split before anyone writes.
Location modifiers often deserve their own location or service-area URL when the business truly serves that place. Stuffing ten towns into one thin page is how local templates fail quality checks. Clustering helps you see the town list before you commit to a bad template.
Comparison clusters need criteria tables and honest trade-offs. Tool clusters need a working interactive element or they are just articles with a gadget promise. FAQ clusters can live as sections on a stronger parent when they are too thin alone.
I keep unclustered long-tail rows in a parking sheet. Many become FAQ blocks on the primary page later. Forcing every rare phrase into its own URL is how sites create orphan thin pages.
Volume zeros are normal on niche lists. Do not delete those rows. Difficulty zeros usually mean the export lacked a KD column. The start-with sort then leans on volume alone, so use judgement.
For ecommerce, product category clusters should map to indexable category templates, not to blog posts that try to rank for buy intent. Blog guides support. Categories convert. The page-type label exists to keep that straight in the brief.
When stakeholders want one URL per keyword, I show them a cluster of twelve near-duplicates and ask which single page should earn the internal links. One strong page beats twelve starved ones.
Re-cluster after a major site redesign or a new market launch. Old maps drift. A fresh paste of the keyword universe catches intents you now serve that the old spreadsheet ignored.
Share the topical map view with developers when new templates are required. Service versus location versus comparison is an information architecture decision, not only a writing decision.
I date the export filename and keep the raw keyword paste beside it. When someone asks why a page exists six months later, the cluster membership is the answer, not a vague memory of a workshop.
If two clusters feel close, I search both primaries before merging. Shared words are a hint. Shared results are the decision. This tool starts the map. Judgement finishes it.
When freelancers inherit a messy blog, I cluster the target universe first, then map existing URLs onto those clusters. Orphan posts that match nothing commercial become merge or prune candidates instead of endless rewrites.
Keep the tightness slider honest. Broad groups are fine for early discovery. Tight groups are better when you are about to brief writers and cannot afford mixed-intent pages.