Can the canonical tag really be ignored by Google?

What does "strong signal but not absolute" really mean?

Google uses a weighting system to determine which URL should be considered canonical among similar or duplicate pages. The link rel="canonical" tag is part of this system, and Gary Illyes describes it as a "strong signal" because it represents an explicit directive from the site's author.

However, "strong" does not mean "absolute". Google reserves the right to choose another URL if other conflicting signals seem more relevant or coherent. This is where many practitioners struggle, thinking that simply placing a canonical is enough to control indexing.

What other signals can contradict my canonical tag?

Google assesses several dozen signals to identify the canonical URL. Among the most influential are: 301 redirects, internal linking (which version of the content is the most linked), XML sitemaps, the age and popularity of a URL in its index, or the consistency of hreflang.

If your site sends conflicting signals — for instance, a canonical pointing to URL-A but 90% of your internal links point to URL-B — Google may decide that your canonical is an error and prefer URL-B. Let's be honest: the algorithm is not wrong in this case.

Why doesn't Google make the canonical a strict directive?

The answer lies in one word: errors. Google handles billions of pages, many of which are misconfigured. A canonical can mistakenly point to a 404 page, create loops, or be injected by a poorly configured plugin.

By treating the canonical as a signal among others, Google protects itself against aberrant configurations and can "fix" certain errors on the fly. This is convenient for poorly maintained sites, frustrating for rigorous SEOs who desire total control.

• The canonical tag is a strong signal but not an absolute directive like robots.txt or noindex
• Google weighs the canonical with other signals: redirects, internal linking, sitemaps, hreflang, historical popularity of the URL
• The algorithm can ignore your canonical if it seems inconsistent with the rest of your technical signals
• This approach allows Google to correct certain configuration errors but reduces the control of webmasters
• Regularly checking the URL chosen by Google in Search Console is essential

Is this statement consistent with real-world observations?

Yes, absolutely. Any SEO who has managed complex sites has encountered situations where Google completely ignores a well-implemented canonical tag. This is particularly common on e-commerce sites with thousands of product variants or poorly configured multilingual sites.

The problem is that Google never clearly communicates the exact weight it assigns to each signal. "Strong signal" remains a vague formulation. Is it 70% of the decision weight? 40%? [To be verified] — no public data allows us to precisely quantify this weighting, which makes optimization partially empirical.

What nuances should be added to this claim?

Gary Illyes refers here to the HTML tag link rel="canonical", but there is also the HTTP canonical header and canonicals in sitemaps. Do these three methods hold the same weight? According to real-world tests, the HTTP header seems slightly more reliable for certain types of content (PDFs, non-HTML files), but again, [to be verified] without official confirmation.

Another nuance: self-referencing canonical (pointing to itself) is often recommended but likely has no real weight if all other signals are coherent. It becomes relevant mainly when there is a risk of ghost URL parameters or tracking sessions.

In what concrete cases does Google systematically ignore the canonical?

First classic case: e-commerce facets. If you set a canonical towards the main product page from a page filtered by color/size, but that filtered page receives quality backlinks and generates organic traffic, Google may decide to index it while ignoring your directive.

Second case: translated content with poorly configured hreflang. If your FR canonical points to FR but your hreflang are inconsistent or absent, Google may index the EN version instead for French queries. And that's when it becomes painful to debug.

How to check which URL Google has actually chosen as canonical?

The Search Console remains the reference tool. In the "Coverage" or "Pages" tab, search for your URL and check the field "Canonical URL selected by Google". If it differs from the one you specified, it means that other signals have taken precedence.

You can also use the URL inspection tool to get real-time detail. Warning: what Google displays may differ from what it actually indexes if the page has not been crawled recently. Check the date of the last crawl before drawing conclusions.

What mistakes to avoid to maximize respect for the canonical?

First mistake: setting cross canonicals (A canonizes B, B canonizes A). It seems obvious, but this is still seen regularly on sites with multiple CMSs or poorly finalized migrations. Google will choose arbitrarily, and it probably won’t be what you want.

Second mistake: pointing a canonical to a redirecting URL. It is technically possible but significantly weakens the signal. If your canonical points to a 301, Google will have to "guess" your intention and may make a different choice. Always point to the final URL.

What concrete actions can be taken to align all canonical signals?

Audit your site with a full crawl (Screaming Frog, Oncrawl, Botify) and export all your canonicals. Cross-reference them with your internal linking: do you massively link to non-canonical URLs? If so, correct your templates and menus.

Next, check your XML sitemaps. They should contain ONLY canonical URLs. If you submit to Google URLs with a canonical pointing to another page, you send conflicting signals. Clean your sitemaps and resubmit them.

• Check in Search Console that the canonical URL chosen by Google matches your directive
• Audit canonicals to detect loops, chains, or pointers to 404/301
• Ensure that internal linking predominantly points to canonical URLs and not to variants
• Clean XML sitemaps to submit only canonical URLs
• Check consistency between canonicals and hreflang on multilingual sites
• Test implementation with the real-time URL inspection tool before global deployment