Keyword clustering
Keyword clustering groups searches that deserve one page. You do it by results-page overlap: if two queries return the same pages, they belong together. This matters when you have little time and need each page to pull its weight.
By Théophile Louvart, founder of Porteur · Updated 14 September 2026 · Markdown
What keyword clustering means for a small site
Keyword clustering groups queries that should be answered by one page. You are not listing synonyms. You are mapping searches that share results.
One cluster becomes one page. Put the head query in the title, and the other queries in headings and text. Two pages for one cluster fight each other.
This saves you time. It avoids thin lookalike posts. A single solid page can rank for many related queries, not just its head term.
How to cluster by results-page overlap
Collect candidate queries
Start from your seed, like "invoice reminder email". Add variations from Search Console, People Also Ask and your support inbox.
Fetch the top results
For each query, note the top ten URLs. Tools automate this by fetching the results and comparing them, but a small set can be done by hand.
Compare overlaps
If several URLs appear in the top ten for two queries, they belong in the same cluster. If there is little or no overlap, keep them separate.
Pick the head term
Choose the most representative query for the page title. It is often the shortest with clear intent, not always the highest volume.
Name the cluster and page
For example, cluster "invoice reminder email", "invoice reminder template", "how to write an invoice reminder" to one page: /guides/invoice-reminder.
A quick check: search "invoice reminder email" and "invoice reminder template". If many of the same domains hold top spots on both, that is one cluster.
Turn a cluster into one page
- Title: use the head query within about 60 characters.
- H1 and intro: restate the head query in plain words.
- Headings: cover the other queries as sections, one clear intent per section.
- Copy: answer completely, include examples and steps users search for.
- URL: keep it short and stable, like /guides/invoice-reminder.
- Internal links: link related pages in and out so the cluster has support.
A fixed page looks like this: title "Invoice reminder email templates", sections for timing, tone, subject lines and templates, and a short FAQ.
Measure the cluster and watch for cannibalisation
The cluster's volume is the sum of its queries. This is why one page written for a head term is often worth several times what its volume says.
- Before launch: add up the volumes of the queries you grouped.
- After launch: in Search Console, open Performance and filter to the page URL. Review Queries to confirm it collects the cluster.
- Track position: the page should appear for the head query and long tails. Rising impressions without clicks often means the title is weak.
- Check cannibalisation: filter by a key query, then compare Pages. If two of your URLs show, fold one into the other and redirect.
Traps to avoid
- Grouping by wording not results: "template" and "examples" may be one cluster even if the words differ.
- Mixing intents: "pricing" and "free trial" often do not overlap enough, keep them apart.
- Ignoring SERP features: if top results are calculators or videos, match the format or reconsider the page.
- Chasing head terms only: long tails in the cluster are easier wins and support the head.
- Over-splitting: five near-duplicate posts on the same need waste crawl and dilute links.
Questions
List candidate queries, fetch the top results for each and compare overlaps. Group the ones that share several URLs in their top ten, choose a head term, then build one page that covers the set.
"Invoice reminder email", "invoice reminder template" and "how to write an invoice reminder" often share the same top results. That is one cluster. One strong page at /guides/invoice-reminder can cover all three.
Tools fetch results for each query and compare overlaps. Pick one that shows you the SERP sources and lets you edit groups. For a small site, you can also cluster a shortlist by hand.
Classifications vary. For clustering, focus on intent pairs that affect overlap, like informational vs transactional, brand vs generic, local vs global. Overlapping intent usually signals one page.
As many as belong to the same cluster. One page per cluster. Put the head term in the title and work the rest into headings and copy.
Split only when the results-page overlap is weak. If the top results for two queries are mostly different, they deserve different pages with their own focus.
Check my site, free
Want to see which searches should live on one page, starting from your URL, in about thirty seconds, free, with three findings shown whole? Try Porteur.
- Free check, no card
- Read-only, your own accounts
- Readable by your agent
Read next
- GuideKeyword cannibalization: two pages, one search, and how to fix it
- GuideSearch intent: what the results page tells you to build
- GuideKeyword research from Search Console alone
- GuideKeyword difficulty: what the number means and when to ignore it
- GuideCompetitor keyword analysis by hand, with free data
- GlossaryLong-tail keyword
- Free toolKeyword ideas finder
- GlossaryKeyword mapping