Cannibalization audit and separation
Stop pages from competing with each other for the same query.
Cannibalization shows up when several URLs from one site chase one query and none of them consolidates the signals. We started by grouping every page in the catalogue by the query it appeared to target, then looked for clusters where multiple URLs collected impressions for the same term. Each cluster got one of three outcomes. Where two pages served genuinely different intents, we separated them by rewriting each around its own question so the overlap disappeared. Where a page existed only because a variant existed, and it had no distinct question behind it, we folded it into the parent and kept the variant as an option on that page rather than a page of its own. Where one page was clearly the strongest candidate, we made it the target and pointed the rest at it with internal links. The rule we applied throughout is simple. A variant earns its own page when someone searches for it specifically and needs information the parent page cannot carry. Colour rarely qualifies. Size, mechanism and material often do.
- Grouped catalogue URLs by the query each was competing for
- Separated overlapping pages so each answered a distinct question
- Folded variants with no distinct demand back into the parent page
- Chose one target page per query and pointed the rest at it



