Keyword Cannibalization

A long-standing traditional-SEO failure mode: multiple pages on one site target the same query, so they compete against each other and split ranking signals (links, relevance, click-through) instead of concentrating them on a single strong page. The classic fix is consolidation — merge or canonicalize the redundant pages into one.

The first-party grounding

Google’s own canonicalization doc (google-consolidate-duplicate-urls) names the mechanism directly: without a canonical signal, link signals “such as links” get split across the duplicates instead of consolidating to “a single, preferred URL,” and the duplicates waste crawl budget Google would rather spend on “new (or updated) pages.” The documented fixes, ranked by strength: redirects and a rel="canonical" annotation are each “a strong signal,” while merely listing the preferred URL in a sitemap is “a weak signal.” So “consolidate” has a precise first-party meaning — pick one canonical and point the rest at it, hardest signal first. Absent any signal, Google “will identify which version of the URL is objectively the best version,” i.e. it picks for you.

The LLM-era generalization

publishing-volume-hurts-seo argues the concept broadens under AI retrieval from keyword overlap to semantic / vector competition: near-duplicate pages now cannibalize each other in embedding space, so “your pages are not just competing for rankings — they are competing for embeddings.” The remedy is the same (consolidate), but the diagnosis widens — two pages need not share a keyword to cannibalize, only to occupy the same semantic region. This is one of the spoke’s clean traditional → AI-era bridges: an established SEO problem re-expressed in the vocabulary of authority-density and GEO.

Related

google-consolidate-duplicate-urls · publishing-volume-hurts-seo · authority-density · search-indexation · search-marketing