How can you manage canonical URLs between mobile and desktop without breaking anything?

Does this rule still apply with mobile-first indexing?

Yes, and this is precisely what confuses many practitioners. One might think that with widespread mobile-first indexing, the mobile version should be the canonical one. Wrong.

Google continues to enforce that the desktop version is the canonical reference — even though it crawls and indexes mobile first. This logic dates back to the pre-mobile-first era and has never been reversed. For sites that maintain separate URLs (m.site.com vs www.site.com or desktop.site.com/page vs site.com/page), this bidirectional structure remains mandatory.

Why this two-way architecture?

Google needs to map the two versions to understand that they represent the same content. The rel=alternate on desktop says, "here is the equivalent mobile version." The rel=canonical on mobile says, "my reference version is the desktop one."

Without this dual annotation, Google risks treating both URLs as competing duplicate content. Worse: it may index both versions separately, diluting ranking signals, and fragmenting your link equity. The crawl budget is also impacted—Googlebot will attempt to understand the relationship between the URLs, which consumes unnecessary resources.

Who is still affected by this configuration?

Let’s be honest: most modern sites use responsive design, so there is a single URL for all versions. This statement does not concern them—no rel=alternate or canonical between versions to manage.

But some sectors or legacy platforms maintain separate URLs: news sites with inherited architecture, complex e-commerce platforms, some multilingual sites. For them, this rule remains critical and non-negotiable. Poorly implemented, it creates stubborn indexing issues that can take weeks to diagnose.

• Desktop must include rel=alternate pointing to the mobile version
• Mobile must include rel=canonical pointing to the desktop version
• This structure is mandatory only for separate URLs (m.site.com, different domains, dedicated subdirectories)
• Responsive design with a single URL is not affected—no annotations needed
• Errors in this configuration create duplicate content and fragment ranking signals

Is this statement consistent with observed practices in the field?

Yes, but with an important nuance: Google never clearly communicates on edge cases. The rule works perfectly when both versions are strictly equivalent in content. But what happens when the mobile offers reduced content, a simplified structure, or stripped-down features? [To verify]—Google remains vague on the acceptable threshold of divergence.

Field note: I've observed sites where the mobile version contained 40% less text than the desktop. Google continued to index the desktop as canonical, but the ranking was affected—likely because the crawled version (mobile) did not match the declared canonical. Consistency between the two versions is not just a matter of URLs: it’s also about content parity.

What risks come with a shaky implementation?

The first classic scenario: the rel=alternate is in place on desktop, but the mobile forgets the rel=canonical. Result? Google may arbitrarily choose which version to index. I’ve seen sites lose 30% visibility because Google decided to heavily index the mobile version—which was not optimized to be the reference.

Another trap: canonicalization loops. Desktop points to mobile in alternate, mobile points to... another mobile URL in canonical. Or worse: each version declares itself as canonical. Google hates ambiguity. In this case, it often ignores all signals and makes its own choice—which is rarely the one you want.

When does this rule absolutely not apply?

If you are in pure responsive design—a unique URL, identical HTML served to all devices—this statement simply does not concern you. You have no rel=alternate or inter-version rel=canonical to manage. This is the case for 80% of sites created in the last five years.

Another exception: Progressive Web Apps (PWAs) that serve dynamic content via JavaScript, with no distinct URLs. Again, not having separate URLs = no issue with mobile/desktop canonical. The topic only arises if you are intentionally or by technical legacy maintaining different URLs based on the device.

What should I do if my site has separate URLs?

Step one: audit the existing setup. Crawl both versions (desktop and mobile) with Screaming Frog or Oncrawl in specific user-agent mode. Extract all rel=canonical and rel=alternate. Cross-reference the data: each desktop must have an alternate pointing to its mobile equivalent, and each mobile must have a canonical pointing to the corresponding desktop.

If you find inconsistencies—alternate present but no canonical in return, or canonical pointing to a third-party URL—correct immediately. These errors propagate quickly and contaminate indexing. Use Search Console to check which URLs Google has actually indexed: if you see both versions competing, it means the annotation is failing.

What errors should be absolutely avoided in this configuration?

Never declare the mobile version as self-canonical if it has a desktop equivalent. This is the most common mistake: the mobile points to itself instead of the desktop. Google will interpret this as two distinct entities.

Another critical error: redirect chains between annotations. For example, desktop points alternate to mobile-temporaire.com, which then 301 redirects to mobile-final.com. Google poorly tracks these chains—it may abandon the mapping midway. Annotations must point directly to final URLs, without intermediaries.

How can I verify that my implementation is working correctly?

Use the URL Inspection Tool in Search Console on a few critical pages (desktop and mobile). Look at the "User-declared canonical" vs "Canonical chosen by Google" section. If both match your expectations, you're good to go. If Google chooses a different canonical from the one declared, dig deeper: this is a warning signal.

Complete with a live test: force Googlebot to crawl a desktop page, check that it correctly detects the rel=alternate. Then crawl the mobile version, check the rel=canonical. Both must be consistent and reciprocal. If not, Google will ignore your directives and do its own sorting—rarely in your favor.

• Crawl the desktop and mobile versions separately with dedicated user agents
• Extract and cross-reference rel=alternate and rel=canonical annotations
• Check for the absence of redirect chains between annotated URLs
• Verify in Search Console that Google is properly indexing the expected canonicals
• Test live with the URL Inspection Tool on strategic pages
• Audit content parity between mobile and desktop—too large divergences = ranking risk