<b>The canonical collision check</b>

<b>The canonical collision check</b>

Scaled sites generate near-duplicate pages by accident: filter combos, sort orders, pagination, and entity aliases. Catch collisions before the crawler does.

Run this pre-ship audit:

☐ Sort/filter parameters: every variant self-canonicalizes to the clean URL
☐ Pagination: page 2+ canonical points to itself, not to page 1 (page 1 is not a duplicate of page 2)
☐ Entity aliases: "new-york" and "new-york-city" resolve to one canonical, the other 301s
☐ Trailing slash and casing variants: 301, not canonical-only
☐ Cross-template overlap: two templates must never produce the same primary keyword target
☐ Gate: run a shingled similarity pass; any pair above 90 percent similarity is flagged

Guardrail: canonical tags are a hint, not a command. For true duplicates use 301. Reserve canonical for parameter noise you can't redirect.

The cross-template overlap row is the one teams forget. Two well-meaning page types fighting over one query cannibalize each other at scale.

Ship gate: don't publish until all boxes are checked.