How does hreflang really determine which version of your site Google displays?

Official statement

Hreflang tags are crucial for indicating to Google which version of a site should be presented to which international audience, based on the target language and country.

22:11

🎥 Source video

Extracted from a Google Search Central video

⏱ 56:12 💬 EN 📅 30/11/2017 ✂ 13 statements

Watch on YouTube (22:11) →

✂ Other statements from this video 12 ▾

2:45 Le snippet Google doit-il toujours correspondre exactement à la page de destination ?
3:45 Google détecte-t-il vraiment tout seul la langue de votre site multilingue ?
10:01 Faut-il vraiment multiplier les domaines pour son SEO international ?
12:02 Google peut-il ignorer vos versions linguistiques si elles se ressemblent trop ?
12:41 Les iframes nuisent-elles vraiment au SEO de votre site ?
19:33 Pourquoi la Search Console affiche-t-elle des erreurs de données structurées introuvables ailleurs ?
22:25 Faut-il vraiment traiter vos pages AMP comme du contenu principal pour qu'elles soient indexées ?
34:12 Pourquoi Google abandonne-t-il progressivement les pages redirigées vers des erreurs 403 ?
38:24 Comment Google traite-t-il vraiment les liens internes dupliqués sur une même page ?
41:02 Pourquoi les URLs avec hashbangs (#!) sont-elles un boulet pour votre référencement ?
51:10 La vitesse de chargement est-elle vraiment un critère de pénalité Google ?
61:18 Pourquoi un double canonical AMP/desktop peut-il tuer l'affichage de vos pages ?

What you need to understand

What does a hreflang tag actually do?

The hreflang tag tells Google that a page has one or more alternative versions tailored to specific languages or markets. It functions like a signaling system: "This page exists in English for the UK, in English for the US, in French for France." Google uses this information to display the most relevant version in its search results for the user's language and location.

Without hreflang, the engine must guess which version to serve, with a high risk of showing the wrong language to the wrong audience. This issue becomes critical on multilingual e-commerce sites or international media, where each version targets a distinct audience with content that can sometimes be very structurally similar.

Why does Müller emphasize the 'crucial' aspect?

Because international targeting remains one of the most fragile areas in SEO. Implementation errors are common: poorly formed tags, missing reciprocity, incorrect language codes. When a hreflang tag malfunctions, Google arbitrarily chooses a version, often the wrong one, which degrades the user experience and conversion rates.

Müller implicitly reminds us that this tagging is not an optional suggestion, but a structuring signal. On multi-country sites, hreflang becomes the primary lever to avoid cannibalization between language versions and to focus each page's authority on the right target market.

What are the impacts if hreflang is absent or misconfigured?

The total absence of hreflang forces Google to rely on other clues: hosting IP address, domain extension (.fr, .de), meta language tags, page content. These secondary signals are less reliable and often contradictory. The result: a French user can land on the English version, and an English user on the Spanish version.

Implementation errors have a direct cost on ranking and traffic. Google Search Console frequently reports hreflang alerts: non-reciprocal tags, invalid language codes, contradictory canonical URLs. Each of these errors sends conflicting signals that confuse geographical and linguistic targeting, diluting the visibility of each version.

Hreflang tags allow for precise targeting of language and country for each version of a page.
Implementation must respect reciprocity: each page must point to all alternatives, including itself.
Hreflang errors generate conflicting targeting signals, leading Google to choose arbitrarily.
The Search Console detects most common errors, and its monitoring is essential.
Hreflang is not a direct ranking factor but strongly influences which version appears for which query.

SEO Expert opinion

Does this statement reflect the observed field reality?

Yes, without ambiguity. Audits of international sites show that over 60% of hreflang implementations contain critical errors. The most frequent include tags found only in the <head> with no equivalent in the XML sitemap, incomplete reciprocity, and mixing incompatible ISO language codes. Each time, Google struggles to understand the intended targeting, and the SERPs suffer.

Müller does not mention a crucial point: hreflang never corrects an unsubstantiated identical content. If two language versions share 90% of content without justification, hreflang will not be enough to avoid treatment as duplication. Tagging does not exempt from a differentiated content strategy by market.

What nuances should be added to this statement?

Google claims that hreflang helps to "present the right version," but it never guarantees a 100% compliance. In practice, the engine may ignore hreflang if other signals are too contradictory: a canonical pointing to another language, inconsistent geographic redirects, misconfigured Search Console geographic targeting. Divergence cases exist, and they are documented in professional forums.

Another nuance rarely mentioned: hreflang works better for well-defined language-country pairs (fr-FR, en-GB) than for generic languages (fr, en). The latter work, but Google interprets them with more leeway, which can create unexpected displays based on the user's actual geolocation.

In what cases does this rule fail or cause problems?

Hreflang becomes unmanageable on very large sites with dozens of languages and markets. An international store with 30 language versions generates hreflang clusters of 30 tags per page, leading to thousands of lines to maintain in case of a redesign or URL change. Each error propagates exponentially.

Sites that combine multiple structures (subdomains, subdirectories, distinct domains) also face challenges. Google recommends consistency, but in reality, many large groups mix approaches according to local constraints. The result: fragile hybrid hreflang implementations, where each additional market introduces new conflicts.

Warning: Google does not officially document how hreflang interacts with multilingual hreflang-alternative sitemaps and meta language tags. Field tests show that the XML sitemap often takes precedence over the <head> in case of divergence, but Google does not confirm this anywhere. [To be verified] through your own tests before large-scale deployment.

Practical impact and recommendations

What should be done specifically on a multilingual site?

First, thoroughly map each linguistic and geographic version of the site. List all URLs, their target languages, and their markets. This step seems trivial but often reveals inconsistencies: orphan pages with no equivalent, incomplete language versions, vague geographic targeting.

Next, choose your implementation method: HTML tags in the <head>, annotations in the XML sitemap, or HTTP headers. The XML sitemap is recommended for large volumes because it centralizes maintenance and avoids bloating the HTML code. Whatever method you choose, ensure absolute reciprocity: each page must point to all its alternatives, including itself.

What errors should be absolutely avoided?

Never mix canonical and hreflang in a contradictory way. If a FR page canonicalizes to EN, then declares an hreflang alternative EN, you are sending conflicting signals. Google will prioritize the canonical, effectively nullifying the hreflang. Keep canonical references self-referential except in specific documented cases.

Avoid invented or approximate language codes. Use ISO 639-1 codes for language (fr, en, de) and ISO 3166-1 Alpha-2 for country (FR, GB, DE). An incorrect code (en-UK instead of en-GB) may be silently ignored by Google, with no visible alert in Search Console.

How to check if the implementation works?

The Search Console remains the basic tool: section "International Targeting", tab "Language". Google reports detected errors here, but be careful, the absence of errors does not guarantee an optimal implementation. Some subtle inconsistencies do not trigger any alerts.

Complement with manual tests: change your browser's language and your simulated geolocation, then type target queries in Google. Check that the displayed version corresponds to the intended targeting. To automate, use tools like Screaming Frog or OnCrawl that validate reciprocity and syntax across the site.

Map all linguistic and geographical versions of the site before implementation.
Implement hreflang via XML sitemap for sites with more than 100 pages per language.
Check reciprocity: each page must point to all its alternatives, including itself.
Never create a conflict between canonical and hreflang: keep self-referential canonicals.
Use exclusively ISO 639-1 (language) and ISO 3166-1 Alpha-2 (country) codes.
Monitor the Search Console weekly for new hreflang errors.

Establishing a robust hreflang architecture requires sharp technical expertise and rigorous maintenance. For medium or large sites, configurations quickly become complex, with risks of costly visibility errors. If your organization lacks the internal resources to audit, deploy, and monitor hreflang in the long term, consulting an SEO agency specialized in international issues can prove wise to secure your positions in each market.

❓ Frequently Asked Questions

Hreflang est-il obligatoire pour un site multilingue ?

Non, techniquement Google peut deviner la langue cible via d'autres indices (contenu, extension de domaine, IP). Mais sans hreflang, le risque d'afficher la mauvaise version à la mauvaise audience est très élevé, surtout si les contenus sont proches ou si plusieurs versions ciblent la même langue pour des pays différents.

Peut-on utiliser hreflang uniquement pour la langue, sans spécifier de pays ?

Oui, les balises hreflang acceptent les codes langue seuls (ex: hreflang="fr"). Google les interprète comme un ciblage linguistique large, sans restriction géographique. Cependant, si plusieurs versions existent pour la même langue dans des pays différents, il faut absolument ajouter le code pays pour éviter toute ambiguïté.

Que se passe-t-il si deux pages pointent l'une vers l'autre avec des hreflang non réciproques ?

Google ignore généralement ces balises incohérentes et choisit arbitrairement quelle version afficher. La Search Console remonte souvent une erreur « balises de retour manquantes », signalant que la réciprocité n'est pas respectée. La correction de cette erreur est prioritaire pour restaurer un ciblage fiable.

Hreflang influence-t-il directement le classement d'une page dans les résultats ?

Non, hreflang n'est pas un facteur de ranking. Il détermine quelle version d'une page Google choisit d'afficher pour un utilisateur donné, mais ne booste ni ne pénalise le positionnement. En revanche, afficher la bonne version améliore engagement et taux de conversion, signaux indirects qui peuvent influencer le SEO.

Faut-il déclarer hreflang dans le HTML, le sitemap XML, ou les deux ?

Google accepte les trois méthodes (HTML, sitemap XML, en-têtes HTTP), mais il ne faut jamais les mélanger pour une même page sous peine de conflits. Le sitemap XML est recommandé pour les gros sites car il centralise la gestion. Une fois la méthode choisie, tenez-vous-y de manière cohérente sur tout le site.

🎥 From the same video 12

Other SEO insights extracted from this same Google Search Central video · duration 56 min · published on 30/11/2017

🎥 Watch the full video on YouTube →