Official statement
Other statements from this video 19 ▾
- 2:17 Comment empêcher les URLs de login de polluer vos sitelinks dans Google ?
- 6:49 Pourquoi Google ignore-t-il parfois vos balises canonical ?
- 8:46 Les liens vers vos pages AMP sont-ils vraiment comptabilisés vers votre version canonique ?
- 9:43 Pourquoi les URLs avec session ID mettent-elles jusqu'à un an à disparaître de l'index ?
- 11:59 Hreflang et ciblage géographique : confondez-vous encore langue et région ?
- 14:52 Désactiver le géociblage dans Search Console : erreur tactique ou stratégie gagnante ?
- 17:38 La personnalisation du contenu selon les données démographiques nuit-elle au crawl Google ?
- 22:14 Pourquoi Google met-il jusqu'à un an à traiter toutes les redirections après une migration de domaine ?
- 26:31 Faut-il vraiment s'inquiéter des erreurs 'not-followed' dans Search Console ?
- 29:30 La balise meta NOODP doit-elle encore être respectée par Google ?
- 31:57 Pourquoi Google ignore-t-il des URLs présentes dans votre sitemap XML ?
- 43:38 Le support If-Modified-Since est-il vraiment universel sur tous les serveurs ?
- 46:53 Faut-il vraiment supprimer le JSON-LD des pages en NOINDEX ?
- 55:41 Pourquoi l'indexation des images SVG prend-elle plus de temps que celle des pages Web ?
- 62:36 Faut-il vraiment indexer vos pages de recherche interne et de tags ?
- 62:57 Rel 'next' et 'prev' : pourquoi Google les ignore-t-il vraiment aujourd'hui ?
- 71:08 L'outil de soumission d'URL accélère-t-il vraiment le classement de vos pages ?
- 78:26 Faut-il vraiment fusionner vos microsites locaux pour éviter la cannibalisation SEO ?
- 83:59 Comment Google traite-t-il vraiment les sites piratés dans ses résultats de recherche ?
Google recommends using rel=canonical pointing to the desktop version for sites with distinct mobile pages (m.site.com or example.com/mobile/). This setup allows for focused crawl budget and indexing on a single version while serving the mobile version to users through user-agent detection. Essentially, it prevents duplicate content and clarifies which version should rank, but it assumes your desktop version remains the canonical reference even with mobile-first indexing.
What you need to understand
Why does Google recommend this canonical setup?
Google aims to simplify its crawling and indexing. When a site maintains two distinct versions (desktop at www.site.com and mobile at m.site.com), the risk is twofold: crawl budget dilution across two different URLs and duplicate content since the content is generally the same.
The rel=canonical directive from the mobile page to the desktop page clearly indicates that the desktop version is the referenced version for indexing. Google can then concentrate its crawl resources on a single URL while understanding that the mobile version exists for user display. It’s a signal of consolidation, not exclusion.
How does this directive align with mobile-first indexing?
This is where it gets tricky: Google indexes from the mobile version of your content since the shift to mobile-first indexing. But if you have separate mobile URLs with a canonical pointing to the desktop, Google will still index the desktop version (the one pointed to by the canonical) while evaluating content from the mobile version.
In practice, Googlebot crawls your m.site.com, detects the canonical to www.site.com, and indexes www.site.com. However, it uses the content and technical signals from m.site.com to assess that page. It seems counterintuitive but aligns with their logic of consolidation.
What is the alternative to this setup?
The modern alternative is responsive design or dynamic serving on a single URL. A single URL that adapts its display based on the device completely eliminates the canonicalization issue between versions.
If you absolutely maintain two separate URLs, the reverse setup (canonical from desktop to mobile) is technically possible but not recommended as it complicates crawl budget management and conflicts with historically expected behaviors from Google. Mueller confirms here the expected canonical direction.
- Use rel=canonical from mobile to desktop to concentrate indexing on a single version
- Serve the mobile version to mobile user-agents via server detection or 302 redirection
- Ensure that both versions have equivalent content to avoid indexing inconsistencies
- Opt for responsive design if you are revamping your site to eliminate this complexity
- Check that Google is correctly following your canonical via Search Console (Coverage tab, User-defined canonical section)
SEO Expert opinion
Is this recommendation still relevant today?
Let’s be honest: maintaining two distinct URLs for mobile and desktop is a legacy architecture that dates back before 2015. Most modern sites have migrated to responsive design. This guidance from Mueller remains valid for sites that have not yet made this transition, but it should not encourage you to create separate mobile URLs.
The issue with this setup is that it adds a layer of technical complexity: you have to maintain two templates, manage content consistency between versions, and monitor that canonicals are correctly in place across all pages. It’s a common source of errors, especially on large e-commerce sites generating thousands of pages.
What inconsistencies are observed in practice?
In practice, we regularly see Google not respecting the canonical defined by the site. If the mobile version has stronger relevance signals (direct backlinks, higher user engagement, better Core Web Vitals), Google may decide to index the mobile version despite the canonical pointing to the desktop. [To be verified] Officially, Google claims to follow canonicals, but crawl logs often show different behaviors.
Another field observation: sites with mobile versions that have poor content (truncated text, missing images) while canonicalizing to the desktop often see their positions decline. Google evaluates from the mobile but indexes the desktop, creating an inconsistency between the evaluated content and the served content. It’s a classic pitfall.
In what cases is this configuration still justified?
It makes sense for sites with heavy technical constraints that prevent a responsive overhaul in the short term: old proprietary CMS, complex architectures with dozens of templates, small technical teams. In this context, the mobile→desktop canonical setup allows for time savings before migration.
Another case: sites that deliberately provide radically different experiences based on the device (progressive web app on mobile, feature-rich desktop version). But beware, this approach requires constant monitoring of indexing to prevent Google from favoring the wrong version.
Practical impact and recommendations
What should you prioritize checking on your site?
First instinct: open Search Console and filter by device type. Look at which URLs are indexed. If you see a mix of desktop and mobile URLs in the index, your canonical configuration is not working correctly. Google should only be indexing the desktop versions if your canonicals are appropriately followed.
Second check: crawl your site using Screaming Frog in mobile and desktop mode separately. Export the detected canonicals. Each mobile page should systematically point to its desktop equivalent. A single mobile page that points to itself or lacks a canonical creates a leak that dilutes your crawl budget.
How can you fix a faulty canonical configuration?
If your canonicals are absent or inconsistent, the correction depends on your technical setup. For a site using WordPress, check your mobile detection plugin (WPtouch, etc.) and make sure it correctly injects the canonical pointing to the desktop version. For custom development, add the tag in the
of your mobile template.Beware of the self-referential canonical trap: some CMSs automatically generate canonicals that point to the current URL. On a page m.site.com/page/, this would give you a canonical pointing to m.site.com/page/ instead of www.site.com/page/. It’s a common mistake that needs correction at the template level or through server rewriting.
Should you migrate to a responsive architecture?
Yes, if your technical roadmap allows for it. Responsive design eliminates all this complexity: a single URL, one content to maintain, no risk of misconfigured canonical. It’s the approach recommended by Google for years and the one that requires the least SEO maintenance.
Migration requires rigorous planning: auditing content differences between versions, preparing 301 redirects from mobile URLs, testing Core Web Vitals on the new responsive version. It’s a significant task, but it permanently removes the risk of indexing dilution between URLs. For complex sites with thousands of pages and critical business stakes, enlisting a specialized SEO agency can help secure this transition and avoid traffic losses during the migration.
- Check in Search Console that only desktop URLs are indexed
- Crawl the site in mobile mode and ensure that each mobile page contains rel=canonical pointing to desktop
- Test 10-15 random pages with the URL inspector in Search Console to validate that Google correctly detects the canonical
- Compare the visible content between mobile and desktop versions: they must be equivalent (text, images, links)
- Check that user-agent detection or 302 redirection to mobile functions correctly for visitors
- Plan a gradual migration to responsive design if resources allow
❓ Frequently Asked Questions
Dois-je utiliser rel=canonical ou rel=alternate entre mes versions mobile et desktop ?
Que se passe-t-il si j'oublie le canonical sur certaines pages mobiles ?
Le canonical mobile→desktop fonctionne-t-il si je redirige les utilisateurs desktop vers www et mobile vers m. ?
Google peut-il ignorer mon canonical et indexer quand même la version mobile ?
Dois-je dupliquer tous mes backlinks entre versions mobile et desktop ?
🎥 From the same video 19
Other SEO insights extracted from this same Google Search Central video · duration 1h06 · published on 24/03/2016
🎥 Watch the full video on YouTube →
💬 Comments (0)
Be the first to comment.