How to Cluster SEO Keywords With SERP Overlap

When you decide whether several SEO keywords belong on one page, compare the pages that rank for them in search results. SERP means “search engine results page.” This method, called SERP-overlap clustering, uses the pages that appear for both keywords to help you judge whether one page could meet both searchers’ needs.

Use overlap as one input when deciding whether keywords belong on one page. Similar results suggest that Google currently finds some pages relevant to both searches, but you still need to check what searchers want and whether one page can meet those needs clearly.

Compare the same kind of search results

For each keyword, collect a fixed number of organic results, such as the top 10. Organic results are the standard web-page listings, separate from ads and special features such as maps or AI Overviews. Keep the search engine, country, language, and device the same. Also collect the same number of results at roughly the same date. Google says search results can vary with time and location. Device and language can also affect them, so mixing conditions makes the comparison harder to interpret. See Google’s Search Console performance report guidance.

Compare result URLs, not just site names. Two queries can return different pages from the same website, which does not necessarily mean one page serves both searches. If you remove tracking parameters or account for redirects, apply the same rules to every URL and avoid merging distinct pages.

A shared-URL count is the simplest measure: count how many result URLs appear in both lists. If two top-10 lists share four URLs, they have four shared results, or 40% of each list. A different measure, Jaccard similarity, divides the shared URLs by the total distinct URLs in both lists. In this example, the lists contain 16 distinct URLs, so the Jaccard similarity is 4/16, or 25%. Label the measure you use; the two percentages describe different things.

Build candidate groups, then check intent

Start with keywords that could plausibly relate to the same page. Capture a consistent set of organic results for each, then compare pairs and flag those with notable URL overlap. Keep a record for each keyword and its result URLs. Include the overlap measure, what searchers likely want, and the page you might assign.

A rule such as “three shared URLs in the top 10” can signal that you should review the keyword pair. It is a workflow choice, not a Google standard or a proven universal cutoff. Tool providers also compare different numbers of results and use different clustering methods, so their cluster labels and scores are not automatically comparable. For example, Semrush describes its own keyword clustering approach; treat its output as a starting point for review.

Next, inspect the overlapping pages and ask what each searcher is trying to do. A person looking for instructions may need a guide, while someone using similar words to find a product may need a category or product page. Check whether the result pages have the same general format and whether one useful page could answer both queries without becoming unfocused.

Watch for chain-linking. If keyword A overlaps with B, and B overlaps with C, that does not establish that A and C belong on one page. Review the whole group for a shared goal and a coherent page type before assigning it to a URL. If combining the topics would make the page confusing, separate them. If splitting them would create repetitive, low-value pages, keep them together or leave a keyword unassigned until the site has a useful page for it. Google’s people-first content guidance supports making pages for readers’ needs rather than creating pages just to capture search traffic.

Check that your URL data is usable

A comparison is only as reliable as the URLs it compares. In a search-data check on November 11, 2026, eight Google queries returned 65 organic rows. Sixty-three rows contained Google goto URLs that did not show the destination URLs directly. A string comparison of those fields showed no shared URLs across four query pairs, but that result did not establish zero destination overlap: the underlying page URLs were unresolved.

The check also found a row whose raw domain field said google.com, even though the title and breadcrumb indicated a Moz result. Those clues did not verify the full destination URL, so the destination was not guessed. Inspect raw rows when results look implausible. Ask the data provider for resolved destination URLs or use its documented matching method before drawing conclusions from URL overlap.

This was one snapshot of eight queries, not evidence about Google results or provider data in general. It did not test whether any clustering approach changed rankings. Keep the date, query set, and data limitations beside any reported result.

Use Search Console for your own pages

After you form a candidate cluster, check how your site’s pages appear for those queries. Google Search Console lets site owners examine query and page performance, which can reveal whether one URL already appears for several related searches or whether different pages appear. Google explains these uses in its Search Console common tasks.

Search Console reports data for your website, not the full set of competing results for every query. Use it to assess your existing coverage, then use consistent search-result data and editorial judgment to decide whether to combine, split, or leave pages as they are.

What overlap does and does not tell you

Repeated URL overlap is a practical signal that some pages currently rank for both queries in the conditions you checked. When the search goals and page types also match, that supports considering one page for both.

Overlap does not prove that the queries have identical intent, that Google will rank your page for both, or that one page is right for every site. Results change with context, and a small overlap can reflect only a few pages that address both needs. A low overlap also does not automatically require separate pages: examine what searchers need and whether your site can serve those needs well.

Google’s guide to how Search works describes how Google crawls, indexes, and serves pages. A keyword cluster is a planning hypothesis; it does not guarantee that Google will crawl, index, or show a particular page.

Keep the AI-search implication narrow

Google’s current guidance says that its AI features use the same foundational SEO practices as Search and require no separate AI optimization. A clear keyword map can help you organize pages around related searches, but the evidence here does not establish that SERP-overlap clustering itself improves visibility in AI Overviews or AI Mode. Google’s AI features guidance explains the relevant eligibility and SEO principles.

Key Takeaways

• Compare consistent results: For every keyword, use the same search engine, market, language, and device. Keep the result depth and approximate date consistent too.
• Count URLs carefully: Match destination pages rather than raw wrapper links or site domains, and state whether you report a shared count or a similarity percentage.
• Review intent before grouping: Combine keywords when their search goals and page types fit a single useful page; overlap alone does not settle the decision.
• Treat thresholds as review rules: A cutoff such as three shared top-10 URLs is a provisional workflow choice, not a Google rule.
• Validate the page map: Use Search Console to see which of your pages appear for which queries. Search Console does not show every competing result.

What I'm building

Delegate tasks. Get software.

Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.

Take a look at vroni.com