Why does Google sometimes ignore your rel=canonical tag?

Is the canonical tag just a simple suggestion?

Exactly. Google treats rel=canonical as a strong recommendation, but never as an imperative instruction. Unlike 301 redirects which enforce a server-side URL change, the canonical tag remains a client-side indication that the engine may choose to ignore.

The search engine aggregates several dozen canonicalization signals and calculates a "canonical URL probability." If your tag points to example.com/fr/ but the majority of external backlinks point to example.com/ru/, the sitemap lists the Russian version, and internal links reinforce this version, Google may conclude that the Russian version is objectively the main page.

What exactly are these multiple signals?

Google combines at least: 301/302 redirects, rel=canonical tags, hreflang, XML sitemaps, internal link structure, anchor texts and context of external backlinks, content consistency, indexing history. Each signal has a variable weight depending on the context.

For example, a sitemap declaring example.com/produit-a/ as the main URL weighs less than a hundred authoritative backlinks pointing to example.com/produit-a?ref=promo. External signals escape your direct control, unlike on-site elements.

When does Google switch to another URL?

Two typical scenarios: massive inconsistency between your directives (canonical to A, but 80% of internal links to B) or overwhelming weight of external signals (the .ru version heavily linked while you canonicalize to .fr).

Another common case: a parameterized URL generates more organic traffic and links than the "clean" version you want to canonicalize. Google may decide that the parameterized version is the true main page as it better meets user expectations.

• Rel=canonical is just one signal among others, never a guarantee of indexing.
• Google weighs dozens of factors to determine the actual canonical URL.
• External signals (backlinks, mentions) often carry more weight than your internal directives.
• An internal inconsistency (sitemap vs. canonical vs. links) drastically weakens your signal.
• Monitor Search Console to detect when Google chooses a different URL than the one you declared.

Is this statement consistent with field observations?

Absolutely. It is regularly observed that Google indexes a different URL than the one declared in canonical, especially on multilingual sites or those with parametric variants. The Search Console explicitly reports this in the "Coverage" tab with the status "Alternative URL with appropriate canonical tag."

What is less documented: the exact weight of each signal. Google remains very vague on weighting. Does a 301 systematically override a canonical? [To be verified] in practice, yes in 99% of cases, but exceptions are sometimes observed during complex migrations where indexing history temporarily counters the redirect.

What nuances should be added to this rule?

First point: not all signals are created equal. A 301 redirect server-side remains the strongest signal, far above a canonical. If you really want to impose a URL, physically redirect — don’t rely solely on the tag.

Second nuance: timing plays a major role. Google may take weeks to reassess canonicalization after a change. If you add a canonical today but hundreds of backlinks point to the old URL, the engine won’t switch instantly. It takes time for the new signal to become predominant.

In what cases does this approach pose problems?

On e-commerce sites with thousands of product variations (color, size, filters). If you canonicalize all variations to the main product page, but Google sees massive backlinks to example.com/chaussure-rouge-taille-42, it may decide to index that parameterized URL.

Result: dilution of the SEO signal, internal cannibalization, inconsistency in SERPs. You thought you were consolidating authority on a clean URL, but Google indexes three. This happens especially when your filters are crawlable and generate external links (forums, price comparison sites).

What should be done concretely to master canonicalization?

Systematically audit the consistency of your signals. Ensure that sitemaps, canonicals, hreflangs, and internal links point to the same URL. A single contradictory element can be enough to make Google doubt.

Use Search Console to identify the URLs that Google really indexes. Compare with your declared canonicals. If you detect a discrepancy, look for external signals: backlinks to the undesired URL, mentions on third-party sites, old cached versions.

What mistakes should be absolutely avoided?

Never canonicalize URL A to B if A receives 90% of your backlinks. You would lose that authority signal. In this case, reverse the logic: make A your main URL and redirect or canonicalize B to A.

Avoid canonical chains (A → B → C). Google may stop midway. Always point directly to the final URL. The same goes for loops (A → B → A) that completely nullify the signal.

How can I check that my strategy works?

Three clear indicators: the URL indexed in Search Console matches your canonical, the number of indexed pages remains stable (no hidden duplication), and your main URLs actually concentrate organic traffic.

Also monitor server logs to see if Googlebot is massively crawling URLs you thought were canonicalized. If so, it means the engine has not validated your directive. Investigate why: residual internal links, external backlinks, outdated sitemap.

• Ensure that sitemap, canonical, hreflang, and internal links converge towards the same URL
• Identify in Search Console the URLs actually indexed by Google
• Fix canonical chains and loops
• Prioritize 301 redirects for critical consolidations
• Analyze backlinks pointing to unwanted URLs
• Regularly audit server logs to detect abnormal crawling