What Is Keyword Cluster?
Written byFlavio Amiel Founder, Roborank
A keyword cluster is a group of search queries that share the same underlying intent and can be satisfied by one page. Because search engines return overlapping results for those queries, a single well-structured page can rank for the whole group rather than needing a separate page for every phrasing or variation.
- SERP overlap is the empirical test for a cluster: if two queries return substantially the same top-ranking pages, search engines already treat them as one information need.
- A large analysis of 3 million random search queries found the average #1-ranking page also ranks in the top 10 for nearly 1,000 other relevant keywords, with a median around 400.
- Clustering assigns one page per intent, not one page per keyword — the discipline that prevents keyword cannibalization.
- A common practical threshold is ‘moderate’ overlap: keywords that share at least three of the same top-10 URLs are grouped together.
- The primary keyword of a cluster is usually the highest-volume query in the group, often called its parent topic.
How a Keyword Cluster Works
Keyword clustering rests on a fact about how modern search engines behave: they rank pages for intents, not for exact strings. When two queries mean the same thing to a searcher, the engine tends to return the same set of pages for both. That observable overlap in the search results is the ground truth a cluster is built on.
The practical test is SERP overlap. Take two candidate keywords, run each one, and compare the top-ranking URLs. If the results are substantially the same pages in roughly the same order, the engine has already decided those queries represent one information need — so one page can serve both. If the results are mostly different pages, the queries carry different intent and belong in separate clusters. This is why clustering is more empirical than it looks: you are not guessing at meaning, you are reading a decision the engine has already published in its results.
Grouping this way changes the unit of content. Instead of asking “how many keywords do I have?” you ask “how many distinct intents do I have?” A hundred keywords might collapse into eight clusters, which means eight pages, each targeting a whole family of related queries. The primary query in a cluster — usually the highest-volume one — becomes the parent topic the page is built around, and the rest inform its headings, subtopics, and phrasing.
How Overlap Thresholds Define a Cluster
Clustering tools differ mainly in how strict they set the overlap threshold — how many shared top-10 URLs two keywords need before they land in the same group:
- Soft (loose) — keywords share at least one top-10 URL. Widest net, but it can group queries that are only tangentially related.
- Moderate — keywords share at least three of the top 10 URLs. The common working default; it balances coverage against precision.
- Hard (strict) — keywords share nearly all top-10 URLs. Very tight groups reserved for near-identical queries.
The threshold you pick is a strategic choice, not a mechanical one. Loose clustering builds broad pages that risk being unfocused; strict clustering builds tightly-scoped pages but can fragment a topic into more URLs than it needs. Most practitioners settle around the moderate line and adjust per topic.
Example of a Keyword Cluster
The clearest evidence for why clustering works comes from a large-scale analysis of 3 million random search queries, reported by Search Engine Land. The study looked at top-ranking pages and asked a simple question: beyond the one keyword a page ranks #1 for, how many other keywords does that same page rank for?
The answer was striking. The average #1-ranking page also ranks in the top 10 for nearly 1,000 other relevant keywords, and even the median page — a more conservative figure — ranks for around 400 other keywords. In other words, the typical winning page is not a specialist that shows up for a single phrase; it is a generalist that satisfies a whole cluster of related queries at once. The study also found that ranking for two or three keywords with over 1,000 monthly searches on a single page is common, while dominating multiple very-high-volume keywords with one page is rare.
The implication for how you should structure content is direct. If one strong page naturally earns visibility for hundreds of related queries, then building a separate thin page for each of those queries works against the grain of how ranking actually happens. You would be splitting the relevance and authority that a single page accumulates across many weaker pages — and, worse, inviting those pages to compete with each other for the same results, which is the mechanism behind keyword cannibalization. The study reframes the goal of keyword research: not to find one keyword per page, but to find the cluster of queries a single, thorough page can own.
The most expensive mistake I see is building one thin page per keyword variation — a page for "cold brew recipe," another for "how to make cold brew," another for "cold brew coffee at home." Those are not three topics; they are one topic asked three ways, and Google knows it, because it returns nearly the same results for all three. When you split them, you don’t triple your coverage — you divide your authority across pages that then compete with each other, and the engine picks one to show and buries the rest. The right unit of content is the cluster, not the query. Check the actual search results before you spin up a new page: if the pages ranking for your "new" keyword are the same ones already ranking for a page you own, you don’t need a new page. You need to strengthen the one you have.
From Cluster to Page
Once a cluster is defined, the work shifts to execution. The cluster becomes the brief: its parent topic sets the page’s core focus, its member queries surface the subtopics and questions the page must cover, and its intent decides the format — a how-to guide, a comparison, a product page, a definition. Assigning each cluster to exactly one URL is the job of keyword mapping, and organizing related clusters into interlinked pillars and supporting articles is the job of the topic cluster model. Clustering is the first step in that chain: it tells you how many pages you actually need, and what each one is for.
Frequently Asked Questions
What is a keyword cluster in SEO?
How do you group keywords into clusters?
How is a keyword cluster different from a topic cluster?
Why does clustering matter for rankings?
The Bottom Line
A keyword cluster reframes research away from chasing individual phrases and toward grouping every query that means the same thing to a searcher. Because search engines already rank one page for hundreds of related terms, the winning move is to identify which queries collapse into a single intent and answer that intent thoroughly on one page — not to manufacture a separate page for each way a question can be typed.
Sources
- Study: Top-ranking page in Google ranks for a thousand other queries, too — Search Engine Land
Rank & Cash — the weekly SEO breakdown
One practical teardown a week on ranking in search and getting cited by AI. No fluff.