Search Intent Classification Tool

The four intents, in practice

The classic model still holds up. Informational queries want an explanation. Commercial-investigation queries want a comparison before a purchase. Transactional queries want to buy or sign up now. Navigational queries want a specific brand or page.

What matters operationally is the mapping: informational to guides and glossaries, commercial to comparisons, reviews and alternatives pages, transactional to product, category and pricing pages, navigational to your own brand pages.

• Informational — guide, tutorial, glossary, FAQ
• Commercial — comparison, review, "best X", alternatives
• Transactional — product, category, pricing, sign-up
• Navigational — brand, login, documentation

Classifying at scale

Modifier lists get you most of the way for free: "how", "what is", "why" signal informational; "best", "vs", "review", "alternative" signal commercial; "buy", "price", "discount", "near me" signal transactional. Run those patterns across the full list first and you will label the majority in seconds.

The remainder — ambiguous head terms with no modifier — is where judgement is required, and where the ranking pages are the deciding evidence. If the top results are all product pages, the query is transactional regardless of how it reads.

Mixed-intent clusters

If a cluster contains "what is keyword clustering" and "keyword clustering tool pricing", it is not one page. Mixed intent inside a cluster is the single most useful warning sign a clustering run produces: it means the grouping was tight on wording but loose on purpose.

Split the cluster by intent, give each half its own URL, and link them together. The guide sends readers to the commercial page; the commercial page cites the guide for credibility.

Feeding intent back into the plan

Intent labels should live in the same export as clusters and volumes, because they change prioritisation. Transactional clusters with modest volume often outperform informational clusters with ten times the demand, and a plan that sorts by volume alone hides that.

A workable order: transactional clusters that match existing products, commercial clusters that support them, then informational clusters that build topical coverage around both.