Is the canonical tag really just a suggestion for Google?

Why does Google refer to it as a "signal" and not a "directive"?

The distinction is significant. A directive would be a firm instruction that the search engine must follow to the letter, like the noindex (although even that is not always respected in some cases). A signal is something else: it is an indication that Google integrates into a set of factors to make its decision.

In short, placing a rel="canonical" on a page is expressing your opinion on which URL should be considered the main version. But if other signals—nearly identical content, similarity of fingerprint, internal link structure—contradict that choice, Google can easily ignore it.

What other signals does Google use to identify duplicates?

Google does not merely read your tags. It calculates a content fingerprint to detect similar or identical pages. It analyzes the site structure, XML sitemaps, URL patterns, the consistency of 301 redirects, and especially the internal and external link graph.

If two URLs have strictly identical content but one receives all the backlinks while the other receives none, Google will favor the one with the stronger SEO weight. Your canonical may point in the opposite direction—that probably won’t change anything.

What happens if you place a canonical on truly different pages?

This is where many practitioners go wrong. Some try to canonize page A to page B, which is actually dissimilar, hoping to consolidate ranking or mask thin content. It doesn’t work.

Google compares the actual content of the two URLs. If they are not sufficiently close, the engine completely ignores the canonical and treats each page separately. Worse: you create a contradictory signal that can confuse the algorithm about your true intentions.

• The canonical tag is not a binding instruction, but a preference indicator
• Google cross-references multiple signals: content fingerprint, architecture, links, sitemaps
• Placing a canonical between non-identical pages will be ignored by the algorithm
• The engine always prioritizes factual content analysis over webmaster declarations
• A misplaced canonical does not "force" anything—it can merely create confusion in the signals sent

Does Google's stance reflect what we observe in the field?

Yes, and it’s consistent with 15 years of practice. We have all seen Google ignore canonicals that were well-placed, or conversely, canonize two pages we thought were distinct. The engine has never treated this tag as an absolute.

What’s even more surprising is the complete opacity regarding the similarity thresholds required for a canonical to be accepted. Google does not publish any numbers or percentages. We know that "very close content" is needed, but how many textual variations are allowed? [To be verified] — no official data exists on this point.

What misinterpretation errors should be avoided?

The most common one: believing that a canonical can serve as a "shortcut to consolidate SEO juice" between thematically similar but textually different pages. That’s not its role. If you have five similar product pages with color variations, you cannot canonize the four to the fifth to boost its authority—Google will treat them as separate pages.

Another classic mistake: the mal-implemented self-referencing canonical. Putting a canonical on each page pointing to itself is good practice… provided the URL is strictly identical (no missing trailing slashes, no lingering UTM parameters, no http vs https issues). Otherwise, you create loops or contradictory signals.

In which cases does the canonical remain truly useful?

It remains essential for managing technical duplicates: pagination, sorting, filtering, session parameters, tracking URLs. All these variations generate strictly identical or nearly identical content—this is where the canonical does its job.

It also serves to declare your intent in cases of content syndication, separate mobile versions (even though this is becoming less relevant with mobile-first), or regional mirror pages. But in all these cases, the similarity of content must be substantial. If you have revised the text, added local sections, Google may not follow your canonical.

What should be prioritized when checking your site?

First step: audit your existing canonicals to identify those that point to URLs that are too different. Extract your canonicals via a crawl (Screaming Frog, Oncrawl, Botify), then compare the content of the source and target pages. If you see substantial discrepancies in text, structure, or tags, it’s a signal that Google is probably ignoring your directive.

Second critical point: verify that your self-referencing canonicals do not create contradictions. A URL in https with a trailing slash pointing to its version without a slash is a common error. Use Search Console to spot cases where Google has chosen a different canonical URL than the one you declare—that is a direct symptom of inconsistency.

What errors should be eliminated immediately from your strategy?

Never use the canonical as a ranking consolidation tool between distinct pages. If you have two product pages with different descriptions, keep them indexed separately—or actually merge them. Attempting to canonize one to the other sends a fraudulent signal that Google will detect.

Avoid canonical chains as well. If A points to B which points to C, Google will probably not go up the entire chain—it may choose B or ignore the signal. Each canonical should point directly to the final URL, with no intermediaries.

How to build a robust canonical strategy?

Start from the principle that Google will only believe you if all signals align. Your canonical must be consistent with your XML sitemap (which should only list canonical URLs), with your internal linking (always favor links to canonical versions), and with your 301 redirects (which must point to the same URLs).

Test the similarity perceived by Google by comparing the featured snippets or meta descriptions that the engine automatically generates for the pages in question. If Google produces very different snippets, it considers the pages to be distinct—your canonical will not hold.

• Crawl the site to extract all canonical tags and compare source/target contents
• Check in Search Console for discrepancies between declared canonical and canonical chosen by Google
• Eliminate canonical chains and always point to the final URL
• Ensure that XML sitemap, internal linking, and 301 redirects point to the same canonical URLs
• Remove canonicals between truly different pages—let Google index both or merge the content
• Audit self-referencing canonicals for URL inconsistencies (protocol, trailing slash, parameters)