Keyword clustering by SERP overlap, explained

A keyword list is not a content plan. Before you write anything, you need to know which keywords one page can cover together and which need pages of their own. Getting that wrong leads either to thin pages competing with each other or to one page trying to answer unrelated questions. Keyword clustering is how you decide.

What is keyword clustering?

Keyword clustering is the process of grouping keywords by the page that should target them. Each cluster becomes one page: a main keyword (the pillar of the cluster) plus secondary keywords that the same page should also answer.

Done well, it prevents keyword cannibalization, where several of your pages compete for the same searches, and it gives each page a clear job.

Why cluster by SERP overlap instead of similar words?

Because similar words don't always mean the same search, and different words often do. The search results themselves show how a search engine understands each query.

• Different words, same intent: "cheap flights to Rome" and "low cost Rome airfare" share few words but usually return the same kind of pages.
• Same words, different intent: "python tutorial" and "python snake facts" share a word and have nothing else in common.
• Subtle splits: "CRM software" and "what is CRM" look close, but one leads to comparison pages and the other to explanations, so they usually need separate pages.

Text similarity, including embeddings, is still useful as a fallback when there are too few results to compare. But when results exist, they are the better evidence.

How does SERP-overlap clustering work?

You collect the top 10 results for every keyword, count how many URLs each pair of keywords shares, and put keywords in the same cluster when they share at least a set number.

The numbers above are an illustration of how the comparison reads, not measured data. The third keyword shares words with the first, but its results are about diagnosis rather than shopping, so it needs its own page.

What overlap threshold should you use?

Three shared results out of ten is a common default and a good starting point. Raise it for tighter, more specific clusters; lower it for broader ones.

Whichever you choose, spot-check a few clusters by searching the keywords yourself. If a cluster mixes clearly different needs, raise the threshold; if obvious synonyms end up apart, lower it.

Why compare each keyword with the pillar?

Comparing every new keyword with the cluster's pillar, rather than with any member, stops clusters from chaining into one large group.

Chaining works like this: keyword A shares three results with B, and B shares three with C, but A and C share none. If a keyword can join through any member, A and C end up on the same page even though their results have nothing in common. Repeat that across a long list and unrelated topics merge.

1. Sort keywords from strongest to weakest, for example by how many results they have, preferring shorter phrases on ties.
2. Take the strongest unassigned keyword as the pillar of a new cluster.
3. Add every remaining keyword that shares at least the threshold with that pillar.
4. Repeat with the next strongest unassigned keyword until none are left.

Where does search intent fit in?

Intent tells you what kind of page a cluster needs. SERP overlap groups keywords; intent decides whether the page should be a guide, a comparison, a product page or a local landing page.

• Informational: "how", "what", "why". Usually a guide or explainer.
• Commercial: "best", "vs", "review". Usually a comparison or roundup.
• Transactional: "buy", "price", "near me". Usually a product, service or booking page.

Read intent from both the wording and the types of pages that rank. If the top results are all product category pages, a blog post is unlikely to fit, however well written.

How do you turn clusters into pages?

1. One cluster, one page. Use the pillar keyword in the title and H1.
2. Secondary keywords become sections or phrases, used where they fit naturally, not as a list.
3. Check existing pages first. If you already have a page for a cluster, improve it rather than writing a new one.
4. Group clusters into topics. Related clusters form sections of a topical map, with internal links between them.
5. Prioritize. Start with clusters closest to what you sell and those where the results look beatable.

How does Postly Rank's Keyword Clustering work?

Postly Rank Keyword Clustering uses the SERP-overlap method with pillar comparison. The results it compares come from the Join Postly Index, the search index built by its own crawler, not from Google.

1. Paste keywords or upload a CSV or Excel file, and optionally choose a country and language.
2. Choose how many of the top 10 results keywords must share to be clustered: 2 to 5, with 3 as the default.
3. Each keyword is searched in the Join Postly Index. The strongest keyword (most results, then fewest words) becomes a pillar, and a keyword joins the cluster whose pillar shares enough of its results.
4. Keywords with too few results to compare fall back to word overlap or text similarity.
5. Each cluster shows its pillar, its keywords and how each joined, its size, the dominant intent and a suggested page type.
6. Download everything as Excel or CSV, or press "Write article" to open the SEO Writer with the pillar as the main keyword and the rest as secondary keywords.

The Free plan gets the full tool, including both downloads, with a smaller list size. Large lists take a while to process; you get a notification when the run is done.