Official statement
Other statements from this video 32 ▾
- 0:36 Comment vérifier si un domaine a des problèmes SEO invisibles depuis Google Search Console ?
- 1:48 Peut-on vraiment détecter les pénalités algorithmiques cachées d'un domaine expiré ?
- 3:50 Comment gérer le contenu dupliqué quand on gère plusieurs entités distinctes ?
- 4:25 Faut-il dupliquer son contenu pour chaque établissement local ou tout regrouper sur une page ?
- 6:18 Pourquoi les suppressions DMCA massives peuvent-elles détruire le classement d'un site entier ?
- 6:18 Les retraits DMCA massifs peuvent-ils vraiment dégrader le classement d'un site ?
- 7:18 Faut-il privilégier un sous-domaine ou un sous-répertoire pour héberger vos pages AMP ?
- 7:22 Où héberger vos pages AMP : sous-domaine, sous-répertoire ou paramètre ?
- 8:25 La balise canonical fonctionne-t-elle vraiment si les pages sont différentes ?
- 8:35 Faut-il vraiment bannir le rel=canonical de vos pages paginées ?
- 10:04 Le scraping peut-il vraiment détruire le référencement d'un site à faible autorité ?
- 11:23 L'adresse IP du serveur influence-t-elle encore le référencement local ?
- 11:45 L'adresse IP de votre serveur impacte-t-elle encore votre SEO local ?
- 13:39 Les images cliquables sans balise <a> sont-elles vraiment invisibles pour Google ?
- 13:39 Un lien sans balise <a> peut-il transmettre du PageRank ?
- 15:11 Comment Google indexe-t-il vraiment vos pages AMP en présence d'un noindex ?
- 18:21 Combien de temps faut-il pour récupérer après une action manuelle complète ?
- 18:25 Combien de temps faut-il pour récupérer d'une action manuelle Google ?
- 21:59 Faut-il intégrer des mots-clés dans son nom de domaine pour mieux ranker ?
- 22:43 Faut-il vraiment indexer son fichier robots.txt dans Google ?
- 24:08 Pourquoi le cache Google affiche-t-il votre page différemment du rendu réel ?
- 25:29 DMCA et disavow : pourquoi Google privilégie-t-il l'une sur l'autre pour gérer contenu dupliqué et backlinks toxiques ?
- 28:19 Le taux de crawl influence-t-il vraiment le classement dans Google ?
- 28:19 Votre serveur limite-t-il le crawl de Google plus que vous ne le pensez ?
- 31:00 Les signaux sociaux sont-ils vraiment inutiles pour le référencement Google ?
- 31:25 Les profils sociaux améliorent-ils le classement Google ?
- 32:03 Les profils sociaux multiples boostent-ils vraiment votre SEO ?
- 33:00 Les répertoires de liens sont-ils vraiment ignorés par Google ?
- 33:25 Les liens d'annuaires sont-ils vraiment tous ignorés par Google ?
- 36:14 Faut-il activer HSTS immédiatement lors d'une migration de domaine vers HTTPS ?
- 42:35 Pourquoi les étoiles d'avis mettent-elles autant de temps à apparaître dans Google ?
- 52:00 Le niveau de stock influence-t-il vraiment le classement de vos fiches produits ?
Google does not follow the rel=amphtml link if the canonical HTML page is marked noindex. The AMP version remains invisible even if technically valid. Notable exception: a standalone AMP page, with no HTML equivalent, can be indexed independently if it is correctly linked through other channels like the XML sitemap. This distinction is rarely understood by technical teams.
What you need to understand
What exactly is the mechanism between HTML noindex and AMP discovery?
When Google crawls a classic HTML page, it looks for the rel="amphtml" tag in the <head>. This is the primary signal to discover the corresponding AMP version. If this HTML page carries a noindex directive (meta robots or X-Robots-Tag), Google halts processing: it does not index the page AND does not follow the link to the AMP.
The reason is simple. Noindex is a voluntary exclusion instruction. Google assumes that if you block the main page, you do not wish to expose its technical variants either. The crawler respects this intent by bypassing the exploration of linked resources, including the AMP versions declared via rel="amphtml".
How can a standalone AMP still be indexed?
A standalone AMP (without an HTML equivalent) is not discovered via rel="amphtml" as there is no source HTML page. Google can find it through other channels: XML sitemap, direct internal linking, external backlinks pointing to the AMP URL. In this scenario, the AMP functions like a regular page.
If this standalone AMP has no noindex directive of its own and is correctly linked (the mention "linked correctly" in Mueller's statement), it can enter the index. The mention "linked correctly" remains vague: it likely implies a self-referential rel="canonical" and presence in the sitemap, but Google does not detail the exhaustive criteria.
Why is this rule problematic in practice?
Many sites use noindex on intermediary pages (facets, filters, funnel steps) while hoping to index an alternative AMP version for mobile. This is a technical contradiction. If the HTML page is excluded, the associated AMP version also disappears from Google's radar.
Another common case: developers put a temporary noindex on a pre-prod page, forgetting that it also blocks AMP discovery. As a result, even after lifting the noindex, the AMP remains invisible until the next full crawl of the HTML page, which can take weeks on large sites.
- The noindex on the HTML page prevents Google from following the rel="amphtml" link
- A standalone AMP without an HTML equivalent can be indexed if it meets linking criteria (self-referential canonical, sitemap)
- The phrasing "linked correctly" remains imprecise and requires testing to validate exact criteria
- The blocking of the AMP via HTML noindex is not explicitly reported in Search Console, complicating diagnostics
- To index an AMP without HTML, ensure it is discoverable through channels other than rel="amphtml" (sitemap, direct internal links)
SEO Expert opinion
Is this statement consistent with field observations?
Yes, this rule is confirmed by empirical tests. When an HTML page is set to noindex, its AMP version systematically disappears from the index if it was only discovered via rel="amphtml". I have observed this behavior on dozens of e-commerce sites where filtered category pages were blocked with noindex to avoid duplication, leading to the silent de-indexation of the associated mobile AMPs.
On the other hand, the notion of "linked correctly" for standalone AMPs deserves [To verify]. Google does not specify whether a self-referential canonical is sufficient or if other signals are required (presence in a separate AMP sitemap, AMP validation without error, active crawling). Tests show that a standalone AMP in the XML sitemap without validation errors is generally indexed, but the speed of discovery varies greatly depending on domain authority.
What nuances should be added to this rule?
First point: timing. If you remove the noindex from an HTML page, Google does not instantly recrawl the rel="amphtml" link. On sites with a limited crawl budget, this can take several weeks. During this time, the AMP version remains invisible even if technically eligible. Forcing a recrawl via the URL Inspection tool in Search Console accelerates the process.
Second nuance: hybrid AMPs (serving both as mobile versions and standalone pages depending on the context) create ambiguous situations. If they are discovered both via rel="amphtml" AND via direct links, the noindex status of the HTML page might not completely block them, but they risk losing their canonical association and appearing as duplicates in the index.
In what cases does this rule not fully apply?
If an AMP page receives direct quality external backlinks, Google may discover and index it even if the source HTML page is set to noindex. I have observed this case with AMP blog articles shared on social media: the AMP URL enters the index via social links, regardless of the status of the HTML version.
Another exception: dedicated AMP sitemaps. If you submit a separate sitemap listing only AMP URLs with a self-referential canonical, Google may treat them as standalone pages even if HTML equivalents exist in noindex elsewhere. This remains a gray area that Google does not explicitly document, but crawl logs confirm this behavior.
Practical impact and recommendations
What concrete steps should be taken to avoid AMP de-indexation?
First step: audit all HTML pages carrying a rel="amphtml" tag to ensure none carry an inadvertent noindex directive. Use Screaming Frog or an equivalent crawler with a filter on meta robots and X-Robots-Tag. Export the cross-check list of pages with rel="amphtml" AND noindex: these are your blind spots.
Second action: if you must block an HTML page (duplication, low content, funnel steps), ask yourself if the AMP version provides a distinct mobile value. If so, transform it into a standalone AMP: remove the rel="amphtml" from the HTML page, add a self-referential canonical on the AMP, and include the AMP URL in your main XML sitemap or a dedicated AMP sitemap.
What mistakes should be avoided in managing indexing directives?
Never put a noindex "just in case" on an HTML page without checking if it has a rel="amphtml". This is the most common mistake in pre-production: you temporarily block a page in staging, forget to lift the noindex in production, and the AMP remains invisible for months without alert.
Avoid mixing signals as well. An AMP page with a canonical pointing to a noindex HTML URL creates a contradiction: Google must choose between respecting the canonical (thus not indexing the AMP) or treating the AMP as standalone. Generally, it opts for complete exclusion. Ensure that canonicals always point to indexable URLs.
How can I check if my site complies with this rule?
Use Search Console to cross-reference two reports: "Coverage" (pages excluded by noindex) and "AMP" (validation errors or non-indexed AMP pages). If valid AMP URLs do not appear in the index while their HTML page is excluded, you are likely in the scenario described by Mueller.
To validate a standalone AMP, test its URL directly in the URL Inspection tool. Check that Google can fetch it, that it has a self-referential canonical, and that it appears in the sitemap. If all systems are green but it is still not indexed after several weeks, [To verify] the crawl budget and overall domain authority may be an issue.
- Crawl all pages with rel="amphtml" and cross-check with noindex directives (meta robots + X-Robots-Tag)
- Transform critical AMPs into standalone versions if the HTML page must remain noindex
- Add standalone AMP URLs into a dedicated XML sitemap with a self-referential canonical
- Check in Search Console that valid AMPs properly appear in the mobile index
- Test the URL Inspection tool on standalone AMPs to validate fetching and the canonical
- Monitor crawl logs to detect AMPs discovered via sitemap vs rel="amphtml"
❓ Frequently Asked Questions
Si je retire le noindex d'une page HTML, combien de temps faut-il pour que Google réindexe l'AMP associée ?
Une AMP autonome peut-elle être indexée si elle n'apparaît dans aucun sitemap ?
Que se passe-t-il si une page AMP porte un canonical vers une URL HTML en noindex ?
Comment détecter rapidement les AMP bloquées par un noindex HTML dans Search Console ?
Faut-il créer un sitemap AMP séparé ou inclure les URLs AMP dans le sitemap principal ?
🎥 From the same video 32
Other SEO insights extracted from this same Google Search Central video · duration 1h00 · published on 27/07/2018
🎥 Watch the full video on YouTube →
💬 Comments (0)
Be the first to comment.