SEO

Keyword Clustering: How to Group Keywords by the Pages Google Ranks

Keyword clustering groups search terms that one page can rank for, based on shared Google results and search intent. The two methods, a worked example with real search results from 12 November 2026, a spreadsheet formula to do it by hand, and the mistakes that cause cannibalization.

Alexander ChuaCo-Founder, Growigami November 12, 2026 9 min read

Keyword clustering is grouping search terms so that each group can be targeted by one page. Two keywords belong in the same cluster when Google ranks many of the same pages for both, or when the searcher wants the same thing. The point is practical: it tells you how many pages to build and which keywords each page should carry, so you do not write three articles that compete for one query.

This guide covers the two ways to cluster, a worked example using real search results we pulled on 12 November 2026, a spreadsheet formula to do it by hand, and the mistakes that most often lead to pages competing with each other. If you already know which pages you need and want to organize them, read our guide to topic clusters instead. That is the next step.

What is keyword clustering?

It is the step between keyword research and the content plan. Keyword research gives you a list of hundreds of terms. Clustering turns that list into a list of pages, each with one primary keyword and a few secondary ones.

The term dates from 2015. According to Wikipedia’s entry on keyword clustering, the Russian SEO Alexey Chekushin introduced it that year, and the first SERP-based clustering tool, Just-Magic, launched in Russia the same year.

Keyword clustering is often confused with two related ideas:

TermWhat it decidesOutput
Keyword clusteringWhich keywords one page should targetA list of pages, each with its keywords
Topic clusterHow related pages link to each otherA pillar page plus linked subtopic pages
Keyword cannibalizationWhether two existing pages compete for one queryA merge, redirect or retarget decision

What are the two methods of keyword clustering?

There are two, and they answer slightly different questions.

SERP-based clustering compares the search results. For each keyword, you take the top 10 results, then check how many URLs each pair of keywords shares. Wikipedia describes the minimum number of shared URLs needed to group two keywords as the “clustering level”. A low level such as 1 or 2 creates a few large groups, and a high level such as 9 or 10 creates many tiny ones, because two keywords rarely share almost all of their top 10. The logic is that if Google already ranks the same pages for two queries, one page can rank for both.

Semantic clustering groups keywords by meaning, using language models or shared words, without looking at the results. It is faster and cheaper because it needs no search data. Keyword Insights shows where it fails: two vape queries that look alike had only 11.8% similarity in their search results, so a semantic tool would have put them on one page while the SERPs called for two.

SERP-basedSemantic
Groups keywords byShared URLs in the top 10Similar meaning or shared words
Needs search dataYes, one results page per keywordNo
Catches different intents behind similar wordsYesOften not
Best forDeciding which pages to buildFirst pass on a very large list, brainstorming topics

Wikipedia also describes three strictness levels for SERP-based tools. Soft clustering compares every keyword only with the highest-volume keyword in the group. Moderate requires every keyword to share URLs with at least one other keyword in the group. Hard requires every keyword to share the same matching URLs with all the others. Hard clusters are smaller and safer; soft clusters are bigger and need more checking by hand.

A worked example with real search results

To show how this works, we searched four related keywords on 12 November 2026 through a web search API set to US English, and compared the top results. The results come from a search provider, not a Google rank tracker, so treat them as an illustration of the method rather than the exact Google page.

Pair of keywordsShared URLs in the top 10What the results showDecision
keyword clustering / keyword grouping1Both return definitions and how-to guidesSame page, despite the low count
keyword clustering / keyword clustering tool2The tool query returns free tools and “best tools” listsSeparate page
keyword clustering / topic clusters0Topic cluster guides from Semrush, HubSpot and MozSeparate page
keyword grouping / keyword clustering tool1Guides on one side, tools on the otherSeparate page

Two lessons come out of this small test.

First, the count is an input, not the answer. “Keyword grouping” shared only one URL with “keyword clustering”, which would split them at any normal threshold. But the results for both are the same kind of page (what it is and how to do it), and several of the “keyword grouping” results say outright that it is another name for clustering. A person reading the results would put them together. A tool set to a threshold of 3 would not.

Second, intent shows up in the type of page that ranks. “Keyword clustering tool” shares two URLs with the main term, but most of its results are tool pages and lists of tools. A guide will struggle to rank there, and a tool page will struggle to rank for the guide query. That is two pages, and it is why this article targets the guide query and links to a free cluster template rather than trying to be both.

How to cluster keywords by hand

For a list of up to about 50 keywords, a spreadsheet is enough.

  1. Clean the list. One keyword per row in column A. Remove exact duplicates and plurals that are obviously the same query.

  2. Pull the top 10 for each keyword. Search each one in a private window, or export the results from a SERP tool, and paste the 10 ranking URLs across columns B to K of that keyword’s row.

  3. Count shared URLs for each pair. To compare row 2 with row 3, use this formula, which works in Google Sheets and Excel:

    =SUMPRODUCT(--ISNUMBER(MATCH(B2:K2, B3:K3, 0)))

    It returns how many of row 2’s URLs also appear in row 3. Build a small grid of these counts for the keywords you are unsure about.

  4. Set a threshold and group. Start at 3 or 4 shared URLs and group the pairs that meet it.

  5. Read the results for every group. Check that the pages ranking are the same type (guides, tools, product pages, comparisons) and answer the same question. Split a group if they do not, and merge two groups if they clearly do, as in the example above. This is where search intent decides.

  6. Pick one primary keyword per group. Usually the one with the most search volume, unless a slightly smaller term matches what your page will actually offer.

  7. Check your existing pages. Before you plan a new page, search your own site for each cluster. If a page already targets it, update that page instead.

  8. Log the result. Each cluster gets one URL, one primary keyword and its secondary keywords. Our SEO cluster template has tabs for seed keywords, clusters and the content map.

For larger lists, a clustering tool runs steps 2 to 4 for you. Semrush’s Keyword Strategy Builder groups keywords from up to five seed terms into topics and clusters with a primary and secondary keywords for each page. KeyClusters and Keyword Insights cluster an uploaded keyword file using Google results. Keyword Insights, for example, sells clustering in credits, so check the price on the vendor’s own page before you upload a list of thousands.

What are the most common keyword clustering mistakes?

  1. Grouping by shared words. “CRM for startups” and “startup CRM” are the same query. “CRM pricing” and “free CRM” share a word and nothing else. Words are a hint; the results decide.
  2. Mixing intents in one cluster. A definition, a tool and a pricing query on one page means the page is the second-best result for all three.
  3. One page per keyword. The opposite mistake. Ten near-identical pages for ten variants split links and signals between URLs, which is the definition of keyword cannibalization.
  4. Ignoring the pages you already have. Clustering a fresh list without checking the live site is how teams publish a second page on a topic they already rank for.
  5. Trusting the threshold blindly. As the worked example showed, synonyms can share few URLs and different intents can share several. Read the results for every cluster you will build a page for.
  6. Never rechecking. Results change. Re-run the clusters for your main pages every few months, and before any large content push.

Why clustering matters more for B2B

B2B keywords are small. A term that matters to a SaaS company may have 50 or 100 searches a month, and many of its variants show 0 or 10 in keyword tools. When volumes are that low, picking the primary keyword by volume alone is noisy, and the shared-results test becomes the more reliable signal of which terms belong together. It is also how you build topical authority: a set of distinct pages that each own one question, linked into a cluster, rather than many thin pages that overlap.

It matters for AI search too. AI assistants quote passages, and a page written for one intent gives them a clear passage to quote. Our SEO and GEO service clusters the keyword list before any page is written, then maps each cluster to one URL and checks it against the pages the site already has.

Frequently asked

What is keyword clustering?

Keyword clustering is grouping search terms so each group can be targeted by one page. Keywords go in the same group when Google ranks many of the same pages for them, or when they share the same search intent. The result is a map of one page per cluster, which avoids writing several pages that compete for the same query.

What is a keyword cluster, with an example?

A keyword cluster is one primary keyword plus the variants a single page can rank for. For example, 'keyword clustering', 'keyword grouping' and 'how to cluster keywords' all ask what the practice is and how to do it, so one guide can target them. 'Keyword clustering tool' returns tool pages and lists of tools, so it belongs to a different page.

How do you create keyword clusters?

Collect your keyword list, pull the top 10 Google results for each keyword, and count how many URLs each pair of keywords shares. Pairs that share at least your chosen number of URLs, often 3 or 4 out of 10, go in one group. Then read the results for each group to confirm the intent matches, pick one primary keyword, and assign the group to one URL.

What is the difference between keyword clustering and topic clusters?

Keyword clustering decides which keywords one page should target. A topic cluster is a set of pages on related subtopics that link to a central pillar page. You cluster keywords first to find the pages you need, then organize those pages into topic clusters for internal linking.

Can you do keyword clustering for free?

Yes, for small lists. Search each keyword, paste the top 10 URLs into a spreadsheet row, and use a formula that counts shared URLs between rows. It takes a few minutes per keyword, which is fine for 30 to 50 terms. For hundreds of keywords, a SERP-based clustering tool does the same comparison automatically.

Is keyword clustering still useful with AI search?

Yes. AI answers quote passages from pages, and a page built for one intent has clearer passages to quote than one stretched across several. Clustering also stops you from publishing near-duplicate pages, which split links and signals between URLs whether the visitor comes from Google or an AI assistant.

Alexander Chua

Co-Founder, Growigami

Decade running growth at B2B SaaS. Co-founded Growigami to ship the work every week.

Start with one process.

We learn the business and build your first go-to-market process on us this month.