Official statement
Other statements from this video 42 ▾
- 42:49 Can hreflang really be used across multiple distinct domains?
- 48:45 Can hreflang really be used across multiple distinct domains?
- 58:47 Should you really avoid duplicating your content across two distinct sites?
- 58:47 Should you really avoid creating multiple sites for the same content?
- 91:16 Is it really necessary to index the internal search pages on your site?
- 91:16 Should you block internal search pages to prevent indexing of infinite space?
- 125:44 Do Core Web Vitals Really Influence Google's Crawl Budget?
- 125:44 Can reducing page size really enhance your crawl budget?
- 152:31 Does the internal links report in Search Console truly reflect the state of your link structure?
- 152:31 Why does the Search Console's internal links report show only a sample?
- 172:13 Should you really be concerned about redirect chains for Google's crawl?
- 172:13 How many redirects does Google really follow before it splits the crawl?
- 201:37 How does Google actually segment your Core Web Vitals by groups of pages?
- 201:37 How does Google actually segment your Core Web Vitals by page groups?
- 257:21 Does the Chrome UX Report really count your cached AMP pages?
- 272:10 Is it necessary to redirect your AMP URLs during a change?
- 272:10 Should you really redirect your old AMP URLs to the new ones?
- 294:42 Is AMP really neutral for Google rankings, or does it hide an invisible visibility lever?
- 296:42 Is AMP really a Google ranking factor or just a ticket to access certain features?
- 342:21 Why does copied content sometimes outrank the original despite the DMCA?
- 342:21 Is the DMCA really effective in protecting your duplicated content on Google?
- 359:44 Why does copied content outrank your original material on Google?
- 409:35 Why do your featured snippets disappear seemingly without a technical reason?
- 409:35 Do featured snippets and rich results really fluctuate randomly?
- 455:08 Is it true that mobile hidden content is really indexed by Google?
- 455:08 Is it true that Google really indexes hidden content in responsive CSS?
- 563:51 Can structured data really force the display of a knowledge panel?
- 563:51 Is there any structured markup that guarantees the appearance of a Knowledge Panel?
- 583:50 Why do most websites never get sitelinks in Google?
- 583:50 Can you really force sitelinks to appear in Google?
- 649:39 Do 301 redirects really transfer 100% of SEO juice without any loss?
- 649:39 Do 301 redirects really transfer 100% of PageRank and SEO signals?
- 722:53 Should you really delete or redirect expired content instead of keeping it indexable?
- 722:53 Should you really remove expired pages or can you leave them labeled 'expired'?
- 859:32 Are keywords in the URL a ranking factor or just a temporary crutch?
- 859:32 Do words in the URL really influence Google rankings?
- 908:40 Should you really add structured data to embedded YouTube videos?
- 909:01 Should you really add video structured data when you're already embedding YouTube?
- 932:46 Does Page Experience really only matter for mobile SEO?
- 932:46 Why is Google ignoring desktop Core Web Vitals in its ranking algorithm?
- 952:49 Do the API and Search Console interface really display the same data?
- 963:49 Can you use different templates for each language version without harming international SEO?
Google now transfers all signals collected via AMP pages to the declared canonical URL. For Core Web Vitals, the search engine directly monitors the canonical version and measures its own performance, ignoring the metrics of the AMP version. In practical terms: your AMP strategy no longer boosts your CWV scores — only the performance of your standard page matters for ranking.
What you need to understand
What does the transfer of AMP signals to the canonical actually mean? <\/h3>
For years, the question of signal transfer between AMP and canonical <\/strong> has fueled SEO debates. Google finally clarifies: all signals collected via an AMP page — links, engagement, visibility — are now redirected to the canonical URL you declared. No division, no dilution.<\/p> In practice, if your article published in AMP version <\/strong> generates backlinks or traffic, those signals benefit your standard page. The system recognizes the canonical relationship and consolidates the data. This is a major evolution compared to past uncertainties where some practitioners feared signal cannibalization <\/strong> between the two versions.<\/p> The critical nuance lies in Core Web Vitals <\/strong>. Google no longer measures CWV metrics for your AMP page to evaluate your site. The engine tracks the canonical and collects real user experience data (CrUX) exclusively on that version.<\/p> This breaks a persistent myth: having an ultra-fast AMP version does not compensate for a slow canonical page. If your standard page shows a catastrophic LCP <\/strong> or an unstable CLS, it’s this score that weights in the ranking. AMP then becomes a delivery format, not a direct SEO performance lever.<\/p> Google does not specify any restrictions by content type. The rule seems universal: whether you publish blog articles <\/strong>, product sheets, or editorial pages, the signal transfer operates identically.<\/p> There remains a gray area in cases where the canonical itself is an AMP page (full-AMP site). In this scenario, the distinction mechanically disappears. But for the majority of sites that maintain a dual architecture <\/strong> (AMP + standard), the rule is clear: invest in the performance of your canonical.<\/p>Why does Google measure Core Web Vitals only on the canonical? <\/h3>
Does this logic apply to all types of content? <\/h3>
SEO Expert opinion
Does this statement align with recent field observations? <\/h3>
Yes, and it confirms a trend observable for several months. The audits we conduct show that sites with a performant AMP version <\/strong> but a mediocre canonical do not benefit from any CWV boost. CrUX data always reflects the performance of the standard page.<\/p> On the other hand, the transfer of conventional signals (backlinks, traffic) was already partially active before this clarification. What Google is formalizing is primarily the strict separation between popularity signals and experience metrics <\/strong>. The former merge, while the latter remain compartmentalized on the canonical.<\/p> Google remains silent on the timing of the transfer <\/strong>. Are AMP signals consolidated in real time or via spaced crawl cycles? For content with high news value, this delay could impact immediate visibility. [To be verified] <\/strong> in real conditions with traced URLs.<\/p> Another ambiguity: what happens if the canonical relationship is misdeclared or ambiguous <\/strong>? Does Google arbitrarily choose a URL or ignore the transfer? Experience shows that the engine can impose its own canonical, but nothing is stated about the management of signals in these conflicting cases.<\/p> Let’s be honest: the pure SEO interest of AMP has collapsed <\/strong>. If you are not targeting the Top Stories carousel or if you have no specific distribution constraints, the effort to maintain a dual architecture becomes hard to justify.<\/p> It is better to focus your resources on the direct optimization of the canonical: reducing blocking JavaScript, optimizing images, smart lazy loading <\/strong>, server caching. These gains immediately benefit Core Web Vitals without the complexity of maintaining two synchronized versions.<\/p>What uncertainties remain in this announcement? <\/h3>
Should we still invest in AMP outside of Google News carousels? <\/h3>
Practical impact and recommendations
What should you prioritize checking in your current architecture? <\/h3>
Start by auditing the canonical declaration <\/strong> on all your AMP pages. Use the URL inspector in Search Console to confirm that Google correctly recognizes the relationship. A disparity between your declaration and the engine's interpretation can block the transfer of signals.<\/p> Next, compare the Core Web Vitals of your canonicals <\/strong> via PageSpeed Insights (CrUX data) with those of your AMP pages. If the gap is significant, you know where to focus your efforts: it’s the scores of the canonical that count for ranking, not those of AMP.<\/p> If you were maintaining AMP primarily for performance reasons, it's time to transfer those optimizations to the canonical <\/strong>. Adopt AMP best practices (optimized images, critical inline CSS, deferring non-essential JavaScript) directly on your standard page.<\/p> Test frameworks such as Nuxt or Next.js <\/strong> that allow server-side rendering with progressive hydration. You achieve loading times comparable to AMP without the constraint of a dual architecture. And above all, you are working on the only version that matters for CWV.<\/p> AMP remains relevant if you depend on the Google News carousel <\/strong> or if distribution partners (aggregators, third-party applications) specifically consume this format. In this case, maintain the AMP version but invest heavily in the canonical.<\/p> For all other sites — e-commerce, corporate blogs, showcase sites — the cost/benefit ratio of AMP no longer justifies its maintenance. It’s better to gradually migrate to a high-performing standard architecture and monitor the evolution of Core Web Vitals post-migration.<\/p>How to optimize your canonical to capture all the benefits? <\/h3>
In what cases should you still keep AMP? <\/h3>
❓ Frequently Asked Questions
Si je supprime mes pages AMP, est-ce que je perds les signaux accumulés ?
Les backlinks pointant vers une URL AMP profitent-ils vraiment à la canonique ?
Faut-il rediriger les URLs AMP vers la canonique après désactivation ?
Comment vérifier que Google mesure bien les CWV sur ma canonique et non sur AMP ?
Un site full-AMP est-il impacté différemment par cette règle ?
🎥 From the same video 42
Other SEO insights extracted from this same Google Search Central video · duration 996h50 · published on 12/03/2021
🎥 Watch the full video on YouTube →Related statements
Get real-time analysis of the latest Google SEO declarations
Be the first to know every time a new official Google statement drops — with full expert analysis.
💬 Comments (0)
Be the first to comment.