Does Google really consider rel=canonical cross-domain useless?

What is the original purpose of the canonical tag?

The rel=canonical tag was designed to address a fundamental problem: unintended duplicate content. When the same page exists under multiple URLs (with or without www, tracking parameters, dedicated mobile versions), Google must choose which version to index and display in the results.

Unlike a 301 redirect that imposes a URL change on the user and browser side, the canonical acts simply as a signal for search engines. The user remains on the visited URL, but Google understands that it needs to consolidate signals toward the preferred version. This is particularly useful for e-commerce sites with filters, blogs with multiple categories displaying the same articles, or poorly configured multilingual sites.

Why does Google insist on restricting it to the same domain?

The official statement sets a strict limit: the canonical would only work for pages hosted on the same domain. Google does not want this tag to be used to transfer PageRank or authority between distinct sites, as a typical backlink would.

This position aims to prevent abuse: imagine networks of low-quality sites all pointing to a main site via canonical to artificially concentrate SEO juice. Google prefers that cross-domain relationships occur through 301 redirects (in cases of permanent migration) or through regular links without any enforced consolidation attempts.

Is the canonical tag merely a signal or an absolute directive?

This is the crucial point that many ignore: the canonical is a signal among others, not a strict directive like robots.txt or noindex. Google may decide to ignore it if other signals (internal links, sitemaps, contradictory 302 redirects) point to a different URL.

In practice, Google cross-references multiple indicators before choosing the actual canonical page. If your canonical tag points to URL-A but 90% of your internal links lead to URL-B, Google might favor URL-B. Therefore, the consistency of signals is more important than the tag itself.

• The canonical avoids user redirections: it only operates on the search engine side, without impacting UX.
• Usage limited to the same domain: Google claims not to account for cross-domain canonicals.
• Non-absolute signal: Google can ignore the tag if stronger contradictory signals exist.
• Useful for technical duplications: URL parameters, multiple versions, pagination, internal syndicated content.
• Not a PageRank transfer: unlike a backlink, it should not be used to manipulate authority between sites.

Is this cross-domain restriction truly applied in all cases?

Let’s be honest: Google’s official statement dramatically simplifies a reality that is much more nuanced. In practice, cross-domain canonicals are sometimes honored, particularly in contexts of content syndication or gradual migrations. Empirical tests show that Google can consolidate signals between two related domains if the relationship is legitimate and coherent.

The problem is that Google never documents these exceptions. As a result, the SEO community often operates in the dark, and practitioners must test on a case-by-case basis. [To verify]: Does Google explicitly mention cases of official syndication (like AMP cache, Google News) where cross-domain canonicals are accepted? The current statement remains vague on these scenarios.

What common errors does this rule reveal?

The main error is believing that a cross-domain canonical can replace a real migration. Some SEOs try to point an old domain to a new one via canonical instead of a 301, hoping to keep the old site online while transferring juice. Google generally ignores this signal, and the site ends up with two competing versions in the index.

Another classic pitfall: using relative canonicals instead of absolute ones on multi-domain sites or complex CDN configurations. If your tag points to /page.html instead of https://example.com/page.html, it becomes ambiguous as soon as you cross subdomains or protocol variations. Google might ignore it altogether.

In what cases should a redirect be absolutely prioritized over a canonical?

If you are permanently migrating a site to a new domain, the 301 redirect remains the only valid option. The canonical does not transfer users, does not update bookmarks, and does not send any clear change of address signal to the engines. A clean migration requires permanent 301s, period.

Likewise, if you are consolidating multiple sites into one (merger of companies, streamlining of portfolio), redirects must handle the transfer of authority. The canonical cannot orchestrate this type of structural movement. It remains a tool for managing intra-domain duplicates, not for redesigning multi-site architecture.

How can you quickly audit canonicals on an existing site?

First step: extract all canonical tags using Screaming Frog or an equivalent crawl. Filter the canonical URLs pointing to a different domain than the source page. If you find any, it's likely a configuration error or a remnant of a poorly managed migration.

Next, cross-reference with Search Console data: check the Coverage section and URL Inspection. Google indicates which URL it has retained as canonical for each indexed page. If it does not match your tag, you have a contradictory signal to correct (internal links, sitemap, lingering temporary 302 redirects).

What concrete actions should be taken to correct errors?

If you discover unintended cross-domain canonicals, eliminate them immediately. Replace them with self-referencing canonicals (the page points to itself) or with the real preferred URL if there is legitimate internal duplicate content. On a multilingual or multi-regional site, use hreflang tags instead to manage alternative versions.

For ongoing migrations, switch to 301 redirects as soon as possible. If you must temporarily maintain two versions (A/B testing, transition phase), clearly document the strategy and plan an end date. Hybrid situations involving canonical + 301 create conflicts that Google resolves unpredictably.

Should every page be systematically canonicalized to itself?

This is a debated practice. Some SEOs recommend adding a self-referencing canonical on all pages to avoid any ambiguity (especially if lingering GET parameters exist in the URL). Others believe it is unnecessary noise if the site is technically clean.

My opinion: if your CMS generates uncontrolled URL variants (pagination, filters, session IDs), the self-canonical is a useful safeguard. If your architecture is strict and your URLs unique, you can skip it. The essential thing is consistency: either you apply it everywhere, or nowhere. No half measures that would create inconsistent patterns for Google.

• Crawl the site and list all present rel=canonical tags.
• Identify canonicals pointing to a different domain and remove them.
• Verify consistency with Search Console (canonical URL chosen by Google vs tag).
• Correct relative canonicals to absolute ones (https://domain.com/page).
• Replace cross-domain canonicals with 301 redirects if there is a real migration.
• Test strategic pages using the URL Inspection tool for validation.