Manual SERP teardown vs keyword-clustering tools: when grouping by SERP overlap misleads

Manual SERP teardown vs keyword-clustering tools: when grouping by SERP overlap misleads

Question: should you cluster keywords by SERP-overlap (tools that group terms sharing top-10 URLs) or read the SERPs yourself?

Evidence: SERP-overlap clustering is fast and usually right, but on one project it merged two intents that shared URLs only because one weak page ranked for both by accident. Manual teardown of 40 queries found that ~18% of "same cluster" pairs from a 3-URL-overlap threshold actually had divergent dominant intent (informational vs transactional). Raising the threshold to 4 shared URLs cut false merges but split genuine clusters.

Nuance: clustering tools are a strong first pass; the threshold is a precision/recall dial, not a truth setting. Audit the borderline clusters by hand before you commit one page to two intents.

Method note: 40 queries, tool clusters vs manual intent labeling, two overlap thresholds.

Caveat: single niche; intent labels are judgment calls.

Confidence: medium