Langues

Guide Technical International SEO

The technical side of running a multi-region or multi-language site — hreflang, regional Structure d’URLs (ccTLD vs subdirectory vs subdomain), and geotargeting. A technical-SEO view of the International SEO pillar.

Première publication : 27 juin 2026 · Dernière mise à jour : 3 août 2026 · Advanced
Langues
AIRédigé parAI v2

International SEO has a strategy half (qui markets, qui languages, who translates) and a technical half (telling moteur de recherches qui URL sert qui language/region, and pas creating duplicate-content or explorer problems pendant que doing it). Ce hub gathers que technical half — hreflang and x-default, the ccTLD vs subdirectory vs subdomain decision, canonical plus hreflang pour same-language regional duplicates, and geotargeting — pour SEO techniques. None of it guarantees indexation, rankings, or trafic; it supports discovery and locale matching. The complet strategy lives in the International SEO pillar; ces pages are cross-listed ici parce que they're où international fonctionner breaks at the implementation level.

TL;DR — Technical international SEO has silent échec modes, pas loud ones. hreflang is a bidirectional confirmation system: every alternate URL doit point back at every autre; manquant retourner liens peut causer Google to ignore affected annotations. Near-identical same-language regional pages besoin canonical plus hreflang ensemble, pas hreflang alone. Ci-dessous is a correct hreflang définir, the return-tag rule, the geotargeting and operational tradeoffs among ccTLDs, subdirectories, and subdomains, and pourquoi none of ce guarantees indexation, rankings, trafic, or a spécifique affiché locale.

Evidence for this claim Google recommends separate locale-specific URLs, crawlable navigation, and explicit locale signals for multi-regional or multilingual sites rather than automatic IP-based adaptation alone. Scope: Current Google international site guidance. Confidence: high · Verified: Google Search Central: Managing multi-regional sites Evidence for this claim hreflang annotations identify localized alternates and require reciprocal, valid URL relationships; they are not a ranking guarantee. Scope: Current Google hreflang implementation. Confidence: high · Verified: Google Search Central: Localized versions

How hreflang fonctionne, and pourquoi it fails silently

hreflang annotates qui URL sert qui language and region, so Google peut swap the appropriate indexé alternate into the SERP pour a searcher. It fait pas consolidate duplicates: canonicalization is the separate system que clusters duplicate URLs and consolidates signals. hreflang jamais throws a visible implementation error. A broken cluster peut simply arrêter influencing locale selection, leaving vous to trouver the problem in explorer, index, or performances evidence plutôt que a dedicated warning.

Two rules govern it. Language uses ISO 639-1 codes and region uses ISO 3166-1 alpha-2 codes — qui is pourquoi the United Kingdom is gb, pas uk (a courant, silent mistake). And every page in a définir doit référence every autre page in que définir, notamment itself.

A worked hreflang définir (with x-default)

Pour une page publié in US English, UK English, and Spain Spanish, every un of the three URLs carries ce même block in its <head>:

<link rel="alternate" hreflang="en-us" href="https://example.com/us/page/" />
<link rel="alternate" hreflang="en-gb" href="https://example.com/uk/page/" />
<link rel="alternate" hreflang="es-es" href="https://example.com/es/page/" />
<link rel="alternate" hreflang="x-default" href="https://example.com/page/" />

The x-default valeur identifies a fallback URL pour languages or regions vous did pas explicitly liste. It is a search annotation, pas a redirection or a promise que every unmatched searcher lands on que URL. Omitting it n’est pas fatal. Pour tout three delivery méthodes — HTML <link> elements, HTTP Link headers, and XML sitemap annotations — and every edge cas tel as one-way sets, subsets, and redirections à l’intérieur a définir, voir how to implement hreflang and x-default hreflang; ce hub garde seulement the stable core.

Return-tag reciprocity: the mismatch que voids a pair

The la plupart courant façon an hreflang définir fails is a manquant retourner tag. Si /uk/page/ listes /us/page/ as an alternate but /us/page/ ne fait pas liste /uk/page/ back, Google treats the annotation as unconfirmed and ignores que pair. The confirmation has to be mutual:

<!-- WRONG: /us/page/ lists UK below but never gets pointed back to from /uk/page/ -->
<link rel="alternate" hreflang="en-us" href="https://example.com/us/page/" />
<link rel="alternate" hreflang="en-gb" href="https://example.com/uk/page/" />

Conflicting canonical and hreflang signals échouer the même quiet façon: si a regional page canonicalizes to a different-language URL, you’ve told Google the two are the même page pendant que hreflang dit they’re alternates, and hreflang loses. To audit a whole cluster — pas simplement un pair — at scale, voir how to audit hreflang.

Que canonical-versus-hreflang conflict matters la plupart pour a narrower cas: near-identical pages in the même language à travers countries (US English and UK English product pages que differ seulement in currency and a few spellings, Par exemple). Google’s guidance treats que as a duplicate-content problem, pas a translation problem — pick un preferred version and combine rel="canonical" with hreflang plutôt que trying to garder every regional variant entièrement independent. Entièrement translated, substantively different-language pages are a separate cas and don’t besoin ce treatment. Voir cross-language contenu dupliqué and canonicalization pour the mechanics.

ccTLD vs subdirectory vs subdomain: où signals accrue

L’URL-structure decision is really a question of où geotargeting and popularité des liens land. A ccTLD (example.de) is the strongest geotargeting signal and nécessite aucun Search Console setting, but it starts chaque market’s authority from zero and is expensive to run. Subdirectories (example.com/de/) pool operations and signals sous un domain and are the pragmatic par défaut pour nombreux sites. Là is ne … plus a Search Console country-targeting override pour chaque folder; Google infers country relevance from the ccTLD où applicable, hreflang relationships, localized page content, local liens, and autre site and page signals. Subdomains (de.example.com) sit in entre and, comme ccTLDs, tend to accrue authority plus independently of the root. Google documents ces as tradeoffs to weigh, pas a ranking hierarchy — it doesn’t publish a preferred structure or nom a universal winner among the three. The deeper decision guide, plus the pillar’s autre market, localization, and performances guidance, is native to the International SEO pillar and cross-listed into ce technical cluster: voir ccTLD vs. subdomain vs. subdirectory.

Whichever structure vous pick, garder the locale versions on stable, separate, crawlable URLs plutôt que switching content by IP adresse or an Accept-Language header alone. Googlebot commonly crawls from US-based infrastructure and doesn’t send an Accept-Language header, so une page que seulement adapts its content pour a detected locale — sans a distinct, linkable URL pour chaque version — risks leaving some of ceux versions uncrawled, unindexed, or unranked.

None of ce is a guarantee. Correct hreflang, Structure d’URL, and canonicalization prise en charge discovery and locale matching; ils don’t guarantee indexation, a spécifique ranking, trafic, qui locale Google en réalité displays to a donné searcher, conversions, or citation by AI systems. Geotargeting signals are inputs Google weighs, pas switches vous flip.

Add an expert note

Pin an expert quote

New person? Create their unclaimed profile at /admin/experts/ → Pin a quote first.