How do you debug canonicalization when Google keeps selecting a different canonical URL?

Treat canonicalization as a cluster of signals. The HTML canonical tag is one input, and it can be outweighed by stronger contradictory signals.

Debug in this order:

1. Compare the candidate URLs. Check content similarity, status codes, redirect behavior, trailing slashes, protocol/host variants, parameters, pagination, print views, and localized versions.

2. Inspect the canonical tag in the final rendered HTML, not just the template. Confirm it is absolute, valid, points to an indexable 200 URL, and is not changed by JavaScript.

3. Check internal links. If most internal links point to a duplicate URL while the canonical tag points elsewhere, the site is sending mixed signals. Fix navigation, breadcrumbs, sitemaps, hreflang references, and content links.

4. Check XML sitemaps. Include the preferred canonical URLs rather than every duplicate variant.

5. Review redirects and HTTP headers. A URL that intermittently redirects or returns inconsistent headers can undermine the declared canonical.

6. Compare content. If the supposed duplicates are materially different, Google may not treat them as duplicates at all. Conversely, near-identical pages with weak differentiation may be collapsed even if each self-canonicalizes.

7. Inspect external signals when relevant. Strong external links to an alternate URL can reinforce the duplicate.

After fixes, request recrawling for representative URLs and monitor the canonical cluster over time. The solution is consistency across discovery, linking, sitemaps, redirects, and content - not repeatedly changing one tag.