Search Intent Mapping and Page-Type Verdicts

On the last eleven estates we mapped, a median of twenty-two per cent of target URLs were the wrong page type for the result set they were pointed at: a buying guide filed against a query whose top ten is nine category pages, or a category page filed against a query Google answers with a comparison table. No amount of rewriting fixes a page-type mismatch. This engagement finds them before anyone commissions a word. The method: Each target query gets its top ten classified by page type, being category, comparison, tutorial, documentation, forum, tool, editorial or transactional, with the SERP features recorded alongside: shopping pack, People Also Ask, video carousel, local pack, AI overview. We pull the same set twice, six to eight weeks apart, because a single pull cannot tell you whether a page type is arriving or leaving. Four-label intent taxonomies are too coarse to build from and we do not use them. What you receive: A page-type verdict per cluster written as an instruction: build this as X, on this URL pattern, carrying these modules. The module list sits at wireframe altitude, meaning a price table, a spec block, a comparison grid, a code sample, a filter, and it is not a design. Then a mismatch register of existing URLs with a correction on each, being rebuild, retarget, merge or retire. Then a watch list of queries where the page type moved between the two pulls, with the direction of travel. Home and furniture: French SERPs in this category split along a line most content plans ignore. A query naming a product with a dimension or a material returns category and listing pages, and an editorial page will not enter that top ten however well it is written. A query opening with quelle, comment or combien returns editorial and video, and a category page will not enter that one. We map the boundary query by query, because it sits in a different place for canapes than it does for parquet, and clients regularly discover they have been funding the wrong side of it for two years. Where intent moves: Drift is directional. Product-adjacent retail queries move toward shopping surfaces over time and away from editorial. Developer queries move the other way when a tool ships a breaking version, with documentation displaced by forum threads for several months before it reclaims the top ten. Knowing which of the two you are watching decides whether you rebuild now or wait, and the report says which. Excluded: No keyword discovery, so bring the clusters or buy that as its own engagement first. No writing, no briefs, no design, no implementation of the corrections. We do not score pages against a content-optimisation tool, because matching a term list is not evidence that a page type is right and it has never once changed one of our verdicts. Not for you if: Your queries already sit in a single page type, which is common under a few hundred URLs and means this audit will tell you very little for the money. It is also wrong if the retire recommendations cannot be acted on. On most estates the mismatch register asks for deletions, and an organisation that cannot delete pages should not pay to be told which ones. How it runs: Two consultants, three weeks between the two SERP pulls, one working session at the midpoint so planning starts before the final document exists. The raw classification data goes into your Sheets, so your team can re-run the judgement later and disagree with ours on evidence rather than on memory.