Official statement
Other statements from this video 25 ▾
- 1:36 Comment tester efficacement le rendu JavaScript avant de mettre un site en production ?
- 1:36 Pourquoi tester le rendu JavaScript avant le lancement est-il devenu incontournable pour l'indexation Google ?
- 1:38 Pourquoi une refonte de site fait-elle chuter le ranking même sans modifier le contenu ?
- 1:38 Migrer vers JavaScript impacte-t-il vraiment le classement SEO ?
- 3:40 Hreflang : pourquoi Google insiste-t-il encore sur cette balise pour le contenu multilingue ?
- 3:40 Googlebot crawle-t-il vraiment toutes les versions localisées de vos pages ?
- 4:11 Comment rendre découvrables vos URLs de contenu hyper-local sans perdre de trafic ?
- 4:11 Comment structurer vos URLs pour maximiser la découvrabilité du contenu hyper-local ?
- 5:14 La personnalisation utilisateur peut-elle déclencher une pénalité pour cloaking ?
- 5:14 Est-ce que personnaliser du contenu pour vos utilisateurs peut vous valoir une pénalité pour cloaking ?
- 6:15 Les Core Web Vitals sont-ils réellement mesurés sur les utilisateurs ou sur les bots ?
- 6:15 Les Core Web Vitals sont-ils vraiment mesurés depuis les bots Google ou depuis vos utilisateurs réels ?
- 7:18 Pourquoi le schema markup ne suffit-il pas à garantir l'affichage des rich snippets ?
- 7:18 Pourquoi les rich snippets n'apparaissent-ils pas malgré un markup Schema.org valide ?
- 9:14 Le dynamic rendering est-il vraiment mort pour le SEO ?
- 9:29 Faut-il abandonner le dynamic rendering pour du SSR avec hydration ?
- 11:40 Pourquoi le main thread JavaScript bloque-t-il l'interactivité de vos pages aux yeux de Google ?
- 11:40 Pourquoi le thread principal JavaScript bloque-t-il l'indexation de vos pages ?
- 12:33 HTML initial vs HTML rendu : pourquoi Google peut-il ignorer vos balises critiques ?
- 13:12 Que se passe-t-il quand votre HTML initial diffère du HTML rendu par JavaScript ?
- 15:50 Googlebot clique-t-il sur les boutons de votre site ?
- 15:50 Faut-il vraiment s'inquiéter si Googlebot ne clique pas sur vos boutons ?
- 26:58 La performance JavaScript pour vos utilisateurs réels doit-elle primer sur l'optimisation pour Googlebot ?
- 28:20 Les web workers sont-ils vraiment compatibles avec le rendu JavaScript de Google ?
- 28:20 Faut-il vraiment se méfier des Web Workers pour le SEO ?
Google claims that hreflang is used to group linguistic variations of the same content. In practice, this allows the search engine to understand that a FR page and its EN version are linked, and to serve the appropriate version based on the user's language/geolocation. However, this statement remains vague on one point: does this grouping actually influence ranking, or does it only serve to display in the SERPs?
What you need to understand
What does "grouping as a single piece of content" really mean?
Martin Splitt talks about grouping, not merging. Google does not merge your multilingual pages into a single entity in its index — each URL remains distinct, crawlable, and indexable. Hreflang indicates a semantic relationship between these URLs: they cover the same topic, in different languages.
What happens behind the scenes: Google uses these signals to select which version to display in the results based on browser language, IP geolocation, or user preferences. Without hreflang, Google may index all your versions, but it risks serving the wrong one — or worse, considering some as duplicate content.
Why does Google emphasize this setup?
Because hreflang solves a recurring problem: how to distinguish a legitimate translation from an opportunistic copy? Without a clear signal, the algorithm may hesitate, dilute the ranking signal across multiple URLs, or completely ignore some versions.
Google needs structural clarity. Hreflang prevents it from having to guess — and when Google guesses, it often gets it wrong. It’s also a way to channel link equity: if a backlink points to your ES version, Google knows it should value the entire cluster, not just that isolated page.
Is hreflang a ranking factor?
Google has never explicitly confirmed that hreflang improves ranking. What Splitt says here is that hreflang helps with understanding and display. But in practice, SEOs find that misconfiguring hreflang can fragment the authority of an international site.
In practice, if your multilingual pages cannibalize each other or if Google serves the wrong version, your CTR collapses — and a poor CTR impacts ranking. So indirectly, yes, hreflang does affect your positions. But it’s not a direct boost, rather a correction for interferences.
- Hreflang does not merge pages — each URL remains independent in the index
- It serves to indicate a semantic relationship between language variants
- Google uses these signals to select the version to display according to language/localization
- Without hreflang, risks of duplicate content or wrong version being served
- Indirect impact on ranking through CTR and authority consolidation
SEO Expert opinion
Is this statement consistent with real-world observations?
Yes and no. The principle of grouping is observable: when hreflang is properly configured, Google displays the right version in the right countries. However, the idea of “a single piece of content with variations” remains conceptual — in reality, each URL has its own crawl budget, its own PageRank, its own indexing.
What’s missing in this statement: Google says nothing about how it handles signal conflicts. For example, if your hreflang FR points to /fr/ but your XML sitemap indicates /en/ as canonical for this URL, what happens? [To be verified] — Which signal does Google prioritize? The official documentation remains vague.
What nuances should be made to this assertion?
Splitt speaks of "localized content," but hreflang also covers regional variants (en-US vs en-GB, fr-FR vs fr-CA). The term "languages" is restrictive — hreflang works on language-region pairs, and some subtleties (dialects, currencies, units) are not automatically detected by Google.
Another point: hreflang solves nothing if the content is genuinely identical across versions. Google wants to see real translations or adaptations, not just changing the URL while keeping the EN text everywhere. If your /fr/ is Lorem Ipsum or an unrefined machine translation, hreflang won't prevent Google from penalizing.
When is hreflang not enough?
Hreflang does not replace a solid site architecture. If you have distinct domains (.fr, .de, .es), subdomains (fr.site.com), or subdirectories (/fr/, /de/), hreflang must be consistent with this structure — but it does not correct it if it’s shaky.
A common problematic case: sites with mixed content (products in EN, blog in FR). Google might ignore hreflang if the on-page signals (lang tag, internal anchors, content language) contradict the annotations. Result: you think hreflang is working, but Google still serves the wrong version.
Practical impact and recommendations
What should you do to configure hreflang concretely?
Three implementation methods: HTML tags in the
, HTTP headers (for PDFs, non-HTML files), or via XML sitemap. Each method has its advantages — HTML is simple for small sites, XML sitemap scales better for thousands of pages, HTTP for non-web content.Golden rule: every page must point to all its variants and to itself. If /fr/ points to /en/ and /de/, but /en/ does not point back to /fr/, Google ignores the signal. It’s a two-way confirmation system — if one page of the cluster is inconsistent, the whole group can be invalidated.
What mistakes should absolutely be avoided?
First mistake: using fanciful language codes. ISO 639-1 for the language (fr, en, de) and ISO 3166-1 Alpha 2 for the region (FR, US, GB). A hreflang="francais" or hreflang="en-UK" (UK instead of GB) will be ignored — Google is strict about these standards.
Second mistake: forgetting the x-default tag. It serves as a fallback for users whose language/region does not match any variant. Without x-default, Google chooses randomly — often the US version, which frustrates European visitors. Another pitfall: duplicating hreflang between HTML and sitemap — this creates conflicts, and Google no longer knows which signal to listen to.
How can I check that my implementation works?
Three tools: Google Search Console (International Targeting report, hreflang tab), Screaming Frog (crawl + hreflang validation), and online validators like hreflang.ninja. Search Console reports errors (orphan URLs, invalid codes, redirect chains), but with a 2-3 week latency.
Manual test: search for a page by changing browser language settings or using a VPN. If Google consistently serves the wrong version despite an active hreflang for several weeks, it indicates that a stronger on-page signal (canonical, HTML language, IP redirections) contradicts your annotations. In this case, diagnosing the hierarchy of signals becomes complex — this is where assistance from an SEO agency specialized in international SEO can make a difference to avoid months of trial and error.
- Implement hreflang via HTML, HTTP, or XML sitemap according to site size
- Ensure every page points to all its variants AND to itself
- Use ISO 639-1 codes (language) and ISO 3166-1 Alpha 2 codes (region) — never use fanciful codes
- Consistently add an x-default tag for fallback
- Avoid duplicates between HTML and sitemap, choose a single method
- Regularly audit via Search Console, Screaming Frog, and manual multilingual tests
❓ Frequently Asked Questions
Hreflang est-il obligatoire pour un site multilingue ?
Peut-on utiliser hreflang uniquement pour certaines pages du site ?
Que se passe-t-il si deux pages ont des annotations hreflang contradictoires ?
La balise x-default doit-elle pointer vers quelle version ?
Hreflang fonctionne-t-il entre domaines différents (.fr, .de, .com) ?
🎥 From the same video 25
Other SEO insights extracted from this same Google Search Central video · duration 30 min · published on 11/11/2020
🎥 Watch the full video on YouTube →
💬 Comments (0)
Be the first to comment.