Guide International SEO Audit

TL;DR — An international SEO audit checks si votre site is telling search engines the correct choses à propos de qui language and country chaque page is pour. It’s différent from a regular technical audit parce que it adds a whole couche of signals — hreflang, Structure d’URL, geo-redirects, localized content — que seulement exist une fois vous serve plus que un market. Vous usually run un avant expanding to a nouveau country, après trafic drops in a spécifique market, or the premier temps vous définir up hreflang.

Ce que an international SEO audit covers

A normal SEO technique audit demande “peut moteur de recherches explorer, render, and index ce site?” An international audit adds a second question on top: “are moteur de recherches serving the correct version of chaque page to the correct personnes?”

Que second question pulls in a définir of signals vous don’t have on a single-market site:

• URL / domain structure — are vous en utilisant a country domain (example.de), a subdomain (de.example.com), or a subfolder (example.com/de/)? And is que choice applied consistently?
• Hreflang — the annotations que tell Google “here’s the même page in French, in German, in Japanese.” Ce is où la plupart of the problems live.
• Geo-redirects — si vous automatically send visitors to a version fondé on leur emplacement, fait que quietly break how Google crawls the rest of votre site?
• Localized content — is le contenu en réalité written pour chaque market, or run via traduction automatique and left alone?
• Country-level performances — are some markets underperforming relative to how beaucoup search demand exists là?

Pourquoi it’s différent from a regular audit

Hreflang is the big raison. It feels comme a petit meta tag, but it’s un of the fiddliest choses in SEO to obtenir correct. John Mueller from Google put it bluntly:

“TBH hreflang is un of the la plupart complex aspects of SEO (si pas the la plupart complex un). Feels as facile as a meta-tag, but it obtient really hard quickly.”

He’s correct. Quand I studied 374 756 domains pour my Brighton SEO 2023 talk, 67% of the domains en utilisant hreflang had au moins un error. Two thirds. Ces are problems vous simply don’t have on a single-language site, qui is pourquoi international audits besoin leur propre traiter.

Quand vous besoin un

Run an international SEO audit quand:

• You’re à propos de to expand into a nouveau market. Audit avant vous launch so vous définir up hreflang and structure correctement the premier temps, au lieu de fixing it plus tard.
• Trafic dropped in a spécifique country. Si Germany cratered but the US is fine, that’s almost toujours an international-signal problem — broken hreflang, a geo-redirect, or content que stopped ranking locally.
• You’re implementing hreflang pour the premier temps. It’s so facile to obtenir incorrect que a pre-launch audit (and a re-audit après it goes live) pays pour itself.
• As routine maintenance. Hreflang breaks silently quand developers ship changements. Checking it periodically catches drift avant it costs vous trafic.

Un chose que trips personnes up: the Recherche Google Console “International Targeting” report is gone. Google supprimé it on September 22, 2022. There’s aucun dedicated dial in GSC anymore, so an audit today leans on explorer outils (Ahrefs Site Audit, Screaming Frog) plus ce que little surfaces in GSC’s Enhancements section.

Vouloir the complet six-area traiter — every hreflang error type, the geo-redirect trap, and the outil walkthrough? Switch to the Avancé tab.

SOP: recurring international SEO audit

Run ce quarterly and après quelconque locale-template, routing, canonical, or hreflang release.

1. Freeze the locale inventory. Export every pris en charge language/region combination, its URL pattern, and its attendu fallback. Fait quand: chaque live locale has an owner and representative URLs.
2. Explorer every locale sans redirections. Collect status, indexability, canonical, language signals, lien internes, and hreflang from HTML, headers, and sitemaps. Fait quand: tout declared locale URLs are represented in un explorer dataset.
3. Validate clusters as graphs. Vérifier self-references, retourner liens, valid codes, direct 200 targets, and agreement entre implementation méthodes. Fait quand: chaque error is assigned to a cluster and template.
4. Examiner routing and accès. Comparer par défaut requêtes with relevant Accept-Language requêtes and vérifier robots d’exploration peut reach every version. Fait quand: forced redirection and inaccessible-switcher findings are documented.
5. Sample localization quality. Examiner visible language, currency, formats, local terms, and market-specific claims with a qualified reviewer. Fait quand: chaque priority market has an approved sample and remediation owner.
6. Comparer market performances. Segment search données by country, page directory, and requête language; annotate launches and technical incidents. Fait quand: movement is separated into demand, targeting, and technical hypotheses.
7. Prioritize and retest. Fix broken targets and manquant retourner liens avant optional x-default cleanup. Fait quand: high-impact fixes réussir the même explorer checks and the suivant examiner date is scheduled.

Metrics que garder the audit operational

Hreflang cluster integrity

Metric: Percentage of attendu locale pairs with reciprocal, self-referential annotations to direct canonical 200 URLs. Ce que it indique vous: Si alternate relationships are technically usable. How to pull it: Join robot d’exploration exports by cluster and comparer réel edges with the locale inventory. Benchmark / realistic range: Establish the current baseline and drive critical-template échecs toward zero. Cadence: Après relevant releases and monthly pour grand modification sites.

Indexable locale coverage

Metric: Attendu locale URLs que are indexable and self-canonical. Ce que it indique vous: Si routing, robots, status, or canonical rules are removing intended pages. How to pull it: Join the locale inventory to explorer status, robots directives, and canonical targets. Benchmark / realistic range: Comparer with the approved inventory; investigate every unexplained gap. Cadence: Monthly and après migrations.

Non-default market search visibility

Metric: Search clicks and impressions pour chaque locale directory and target country. Ce que it indique vous: Si intended markets are receiving search exposure plutôt que default-market URLs absorbing it. How to pull it: Segment Search Console page and country données en utilisant le site’s locale patterns. Benchmark / realistic range: Utiliser chaque market’s propre seasonally comparable baseline plutôt que borrowing un autre market’s target. Cadence: Weekly monitoring with monthly examiner.

Redirigé entry exposure

Metric: Locale entry URLs que redirection fondé on requête language, emplacement, or cookie state. Ce que it indique vous: Où discovery and user-choice risks exist. How to pull it: Run une requête matrix contre representative entry URLs and comparer status and Location headers. Benchmark / realistic range: Aucun forced redirection on URLs que doit remain directement crawlable; document quelconque homepage-only exception. Cadence: Chaque routing release and quarterly.

Testez vos connaissances: International SEO audits

TL;DR — Audit international SEO à travers six areas: (1) URL/domain structure — commencer with a locale inventory (market, URL pattern, audience, canonical, indexability, alternate méthode), alors ccTLD vs subdomain vs subfolder, applied consistently; (2) hreflang — the core, où 67% of hreflang-using domains in my 374 756-domain study had errors; (3) geo-redirects vs Googlebot — tester une requête-state matrix (IP, Accept-Language, cookie, deep liens) since Googlebot mostly crawls from US IPs and sends aucun Accept-Language; (4) technical signals — html lang, Content-Language, JS rendering, pagination, server geolocation; (5) content quality — traduction automatique (judged on valeur and objectif, pas the méthode alone), gaps, local E-E-A-T; (6) country-level performances, keeping observed GSC/GA4 données separate from Ahrefs’ estimates. Run it at scale with a robot d’exploration, pas page-by-page. Prioritize manquant retourner tags and broken URLs over x-default and self-referential tags. The GSC International Targeting report is gone — hreflang errors now surface in Enhancements.

Run the whole chose with a robot d’exploration, pas by hand. Manual checks seulement reveal problèmes on the exact pages vous regarder at; a explorer (Ahrefs Site Audit or Screaming Frog) catches the rest. Here’s the six-area traiter I fonctionner via.

1. Structure d’URL audit

Avant vous score anything, construire a locale inventory — un row per market/language combination, listing l’URL pattern, intended audience, canonical target, indexability, and qui hreflang delivery méthode (HTML head tags, HTTP headers, or sitemap) is en réalité deployed. Ce turns the rest of the audit from a pile of individual URL findings into a dataset vous pouvez join votre explorer contre:

Replace ceux rows with votre propre locales — the point is un normalized table avant vous commencer scoring hreflang, geo-redirect, or content errors contre it.

Three structures, chaque with a différent trade-off, and Google treats les as legitimate:

• ccTLD (example.de) — strongest geotargeting signal, server emplacement irrelevant, but expensive and vous construire domain authority separately per domain.
• Subdomain (de.example.com) — facile setup, flexible hosting, but utilisateurs may pas lire it as country-targeted.
• Subfolder (example.com/de/) — simplest, consolidates authority on un domain, but harder to split out and tied to un server emplacement.

There’s aucun universally superior structure — chaque has had success stories. The audit question isn’t “which is best,” it’s “is the choisi structure implemented consistently?” The échec mode is mixed signals: a .de domain hosted on US servers serving English content, or a site que uses /de/ on some pages and de.example.com on others. Consistency à travers TLD + content language + server/CDN emplacement + lien internes + backlinks is ce que you’re verifying.

Two plus checks ici: jamais utiliser URL parameters pour language/country targeting (Google calls ce “not recommended” — it rend segmentation hard), and remember GSC ne … plus has a manual country-targeting setting. Pour ccTLDs Google infers geotargeting from the TLD automatically. Si vous target Bing markets, Bing Webmaster Outils encore has a Geo Targeting fonctionnalité (domain/subdomain/directory/page level) that’s worth setting.

2. Hreflang audit (the core)

Ce is où la plupart international audits live and die. Hreflang tags fonctionner in pairs to formulaire a cluster of equivalent pages — quand configuré correctement, the correct version swaps in pour the correct country during re-ranking; quand broken, the swap doesn’t se produire and utilisateurs voir the incorrect language.

Ce section covers ce que belongs in a broad audit — the error taxonomy, ce que en réalité breaks a cluster, and how I triage with a robot d’exploration. Pour the complet extraction-to-matrix mechanics (pulling every annotation from tout three implementation méthodes, building the reciprocal matrix, and reading cluster graphs at scale), voir How to Audit Hreflang at Scale.

The nine error types

From my Brighton SEO 2023 study (The La plupart Courant Hreflang Problèmes À travers 374 756 Domains, the largest hreflang study to date — nearly 10× plus grand que quelconque autre), ici are the errors to hunt pour, with how courant chaque is:

67% of domains en utilisant hreflang have au moins un problème. But pas tout errors are equal, and ce is the nuance la plupart checklists miss:

• Manquant retourner tags are the la plupart functionally damaging. Si page A références page B but B doesn’t lien back to A, Google may ignore the whole annotation pair. “Si page X liens to page Y, page Y doit lien back to page X.”
• Manquant x-default is the la plupart courant (56,3%) but pas strictly requis — it’s a recommended fallback, pas a hard requirement.
• Self-referencing tags are a meilleur pratique, pas a requirement soit. 18% miss les and nombreux encore fonctionner fine. Fix pour robustness, but it’s pas a critical blocker.
• Valeur errors are réel breakers: en utilisant jp au lieu de ja pour Japanese, typos comme js pour ja, three-letter codes (gbr au lieu de gb), la (Laos) misused pour Latin America. Utiliser ISO 639-1 pour language and ISO 3166-1 Alpha-2 pour region. (Google fait tolerate some choses — underscores comme en_gb, the specially reserved en-UK, relative URLs, UTF-8 encoded characters.)
• Inconsistent URL formats (trailing slash, http vs https) and orphaned locale URLs (hreflang pointing at 404s or redirections) quietly break clusters aussi. Alternate URLs doit be entièrement qualified with the transport méthode — https://example.com/page, pas //example.com/page or /page.

A clé distinction I drew at Pubcon 2019: hreflang is à propos de the indexé version, pas the canonical. It doesn’t matter ce que votre canonical is définir as — it matters ce que the indexé version is. Si votre hreflang points to une URL that’s canonical-chained away from the indexé URL, the signal peut be lost. Vérifier the indexé URL with GSC’s Inspection d’URL, pas simplement the balise canonical.

Ahrefs Site Audit hreflang checks

Ahrefs Site Audit runs a spécifique définir of hreflang checks I lean on:

1. Hreflang annotation invalid (incorrect ISO codes, e.g. en-en or en-uk)
2. Self-referencing hreflang annotation manquant
3. Page referenced pour plus que un language
4. Plus que un page pour the même language
5. Hreflang and HTML lang mismatch
6. Manquant reciprocal hreflang (aucun retourner tag)
7. Hreflang to non-canonical
8. Hreflang to broken page
9. HTML lang attribute manquant / hreflang défini but HTML lang manquant

Ahrefs hreflang cluster visualization (the differentiator)

Ce is the partie aucun competing guide calls out, and it genuinely modifié how I audit. In Site Audit, ouvrir the URL Details panel pour quelconque page, go to the Hreflangs tab, and vous voir lune page’s complet hreflang cluster rendered as a network graph — every page in the cluster as a node, every annotation as an edge. Errors are highlighted in red, so a manquant or mistakenly-added lien in the cluster jumps out visually.

It’s the premier outil to montrer hreflang clusters ce façon. Au lieu de reading a spreadsheet of “page X is missing a return tag to page Y,” vous voir the broken edge in the graph. It rend it far easier to spot qui cluster is broken En un coup d’œil — and far easier to expliquer the problem to stakeholders que a typical export.

Screaming Frog hreflang méthode

Si vous préférer Screaming Frog: enable Configuration > Spider > Explorer Hreflang avant exploration. Pour multi-domain implementations (ccTLDs que référence chaque autre), ajouter the external domains sous Config > CDNs — sans que, hreflang tags pointing to .de, .fr, etc. won’t be validated. Run post-crawl analysis to populate the hreflang filters, alors examiner les (Contient Hreflang, Non-200 Hreflang URLs, Unlinked Hreflang URLs, Manquant Retourner Liens, Inconsistent Language & Region Retourner Liens, Non-Canonical Retourner Liens, Noindex Retourner Liens, Incorrect Language & Region Codes, Multiple Entries, Manquant Self Référence, Pas En utilisant Canonical, Manquant X-Default, and tags placed outside the head). Export via Reports > Hreflang.

Pour a manual spot-check: view source on a sample page, search pour hreflang, and pour chaque alternate URL confirmer it renvoie 200 and has a retourner tag pointing back, with matching ISO codes.

3. Geo-redirect audit

Ce is the silent killer. Googlebot crawls primarily from US IP addresses. Si votre site auto-redirects every non-local visitor to a “local” version fondé on IP, Googlebot — coming from a US IP — obtient sent to the US version every temps. The consequence: Googlebot jamais reaches votre autre language versions, can’t voir leur retourner tags, and ceux versions peut effectively fall out of the index. A silent deindex.

Là is aucun chemin que follows tout international SEO meilleur practices pendant que aussi maintaining geo-IP redirections on individual pages. Google itself notes “IP emplacement analysis is difficult and généralement pas reliable” and that “la plupart, but pas tout, Google crawls originate from the US.” On locale-adaptive pages (ones que serve différent content by detected emplacement/language au lieu de separate URLs), Google is explicit que it “might not crawl, index, or rank all your content for different locales” — largely parce que Googlebot typically sends aucun Accept-Language header at tout. (Managing Multi-Regional and Multilingual Sites; How Google Crawls Locale-Adaptive Pages)

Tester à travers une requête-state matrix, pas un condition:

• Aucun IP signal, aucun Accept-Language, aucun cookie — Googlebot’s typical state. Confirmer l’URL encore resolves to a crawlable, indexable page au lieu de bouncing to a par défaut market.
• US IP (or VPN) with par défaut navigateur settings — voir qui version loads.
• Varied Accept-Language valeurs — vérifier the redirection doesn’t override an explicit hreflang/language choice.
• Existing locale cookie — vérifier it doesn’t permanently trap a utilisateur who tries to switch markets.
• Deep liens straight to a non-default locale URL (pas simplement the homepage) — confirmer ils charger la requêteed locale au lieu de bouncing to a par défaut.

From que matrix:

• Identifier quelconque IP- or cookie-based redirections.
• The seulement acceptable auto-redirect is from the x-default / root URL. Individual language/country pages devrait jamais auto-redirect.
• Confirmer utilisateurs peut manually override with a language selector, and que a cookie preference doesn’t permanently trap les.

4. Technical checks

A cluster of signals au-delà hreflang:

• The html lang attribute. Définir it correctement, but know its role: Google uses the visible content of lune page to detect language, pas the lang attribute. So html lang is mostly pour UX/accessibility (screen readers) and autre engines — Google ignores it pour language detection. Une page with lang="en" serving French content is encore a bug worth fixing. (Ahrefs flags hreflang/lang mismatches.)
• Content-Language HTTP header. The <meta http-equiv="content-language"> tag is deprecated by the HTML spec — don’t utiliser it. The HTTP réponse header is encore valid, and it matters parce que Bing weights Content-Language as a principal language signal pendant que Google treats it as a weak hint. Pour Bing markets, définir the header correctement.
• JavaScript rendering. Si hreflang is injected by client-side JS après charger, robots d’exploration que don’t entièrement render JS may miss it, or traiter it late. Comparer the raw page source (aucun JS) contre the rendered version in Inspection d’URL. Meilleur pratique: hreflang in server-rendered HTML head, HTTP Lien headers, or XML sitemaps. SPA sites (React, Vue, Suivant.js in CSR mode) are the highest risk.
• Pagination with hreflang. Chaque paginated URL nécessite its propre complet hreflang définir, and page 2 in English devrait point to page 2 in French — pas the French homepage. Si vous aussi canonicalize paginated pages, assurez-vous hreflang points to the canonical/indexé URL, since Google ignores hreflang on non-canonical pages.
• Server IP / CDN geolocation pour ccTLDs. Pour ccTLD properties, hosting in-country (or a CDN with PoPs là) is a mild reinforcing signal — TLD dominates, but consistency helps in competitive markets. Plus practically, regional PoPs cut TTFB and améliorer Core Web Vitals, qui is measurable.
• GSC après deprecation. The International Targeting report was deprecated/supprimé on September 22, 2022 (live since 2014), and nothing replaced it directement. Hreflang errors now surface in the Enhancements section, with limited detail — treat it as a secondary signal behind votre explorer outils.

5. Content quality audit

Implementation peut be perfect and le site peut encore échouer in-market parce que le contenu isn’t en réalité localized:

• Traduction automatique vérifier. Unedited MT is a red flag. It misses local search intent (the terms personnes in que country en réalité search), cultural context, and cohesive semantic structure. Google holds translated pages to the même quality bar as original content. Vérifier si titles, meta descriptions, and URL slugs were localized aussi — slugs devrait utiliser local keywords, pas transliteration. Un nuance worth auditing accurately: machine translation isn’t automatically a policy violation. Google’s scaled-content-abuse policy turns on objectif and valeur — it noms automated translation as un exemple of a technique que becomes abusive quand it’s utilisé primarily to manipulate rankings and adds little valeur to utilisateurs, and the même tester s’applique to human-written scaled content. Judge translated pages on usefulness, originality, accuracy, and market fit — pas on “was ce translated by a machine” alone.
• Thin/contenu dupliqué à travers versions. Watch pour near-identical pages à travers languages, and pour primary-market pages cannibalizing the localized versions pour the même requêtes.
• Market-specific content gaps. Utiliser Ahrefs Site Explorer > Organic Keywords filtered by target country to voir ce que le site ranks pour in chaque market, and comparer to the principal market. High-performing English content with aucun localized equivalent is votre gap liste.
• Local E-E-A-T. Authority doesn’t travel automatically — relevance is construit, pas assumed. US PR coverage doesn’t earn trust in Japan. Audit pour local author bios and credentials, local références, local certifications/compliance, reviews on local platforms (e.g. Rakuten in Japan), and liens from local sites. A central content team with aucun local input can’t hit the même threshold as a local subject-matter expert.

6. Performances audit by market

Finalement, où are vous en réalité winning and losing. Garder two données types separate as vous go: Search Console and GA4 give vous first-party observed rows (réel clicks, impressions, sessions); Ahrefs’ country traffic-share and keyword-volume numbers are third-party estimates. Les deux are utile pour building a hypothesis à propos de a market — neither un, by itself, proves pourquoi Google chose to rank or pas rank une page in que market. Google dit as beaucoup à propos de geotargeting généralement: “geotargeting isn’t an exact science.” Treat explorer and performances findings as leads to confirmer, pas final proof.

• Country-level trafic. Ahrefs Site Explorer > Metrics by Country montre organic

• Country-level trafic. Ahrefs Site Explorer > Metrics by Country montre organic trafic share and keyword counts per country. Comparer chaque market’s réel trafic share contre its potential (search demand là). Markets où vous have content but disproportionately low trafic signal a technical/hreflang problem; bon rankings with poor CTR signal a title/description localization problem.

• Benchmark local competitors. Site Explorer > Organic Competitors filtered by country surfaces who en réalité ranks in que market — souvent entirely différent, local-language entities que votre global competitors. Utiliser a lien/content gap contre les.

• GA4 country segmentation rounds it out: élevé trafic + poor engagement = content quality; low trafic to a market you’ve localized = discovery/hreflang. Vérifier Core Web Vitals by country aussi (CrUX in PageSpeed Insights), since regional performances varies widely sans a CDN.

AI summary

A condensed prendre on the Avancé version:

• An international SEO audit covers six areas: URL/domain structure, hreflang, geo-redirects, technical signals, content quality, and country-level performances. Commencer with a locale inventory (market, URL pattern, audience, canonical, indexability, alternate méthode) avant scoring errors. Run it at scale with a robot d’exploration, pas page-by-page.
• Hreflang is the core. My Brighton SEO 2023 study of 374 756 domains trouvé 67% en utilisant hreflang have au moins un error. The nine error types: manquant retourner tags, incorrect language/country codes, non-canonical targets, manquant self-referential tags, manquant x-default, hreflang on non-canonical pages, inconsistent URL formats, and orphaned/broken locale URLs.
• Priorities: manquant retourner tags and broken URLs matter la plupart; x-default (manquant on 56,3%) and self-referential tags are meilleur practices, pas blockers. Hreflang follows the indexé version, pas the canonical.
• Ahrefs Site Audit runs 9 hreflang checks and is the premier outil to montrer clusters as a network graph with red-highlighted errors (URL Details > Hreflangs tab). Screaming Frog nécessite Explorer Hreflang + CDN exploration enabled.
• Geo-redirects vs Googlebot: Googlebot mostly crawls from US IPs and sends aucun Accept-Language header, so site-wide IP redirections send it seulement to the US version — a silent deindex risk. Tester une requête-state matrix (aucun signal, US IP, varied Accept-Language, existing cookie, deep liens) plutôt que un condition. Seulement the x-default/root devrait auto-redirect.
• Technical signals: html lang is pour UX/Bing (Google ignores it pour language); Content-Language header is Bing’s principal signal; JS-injected hreflang risks being missed; paginated pages besoin per-page hreflang.
• GSC International Targeting report supprimé Sept 22, 2022 — hreflang errors now in Enhancements (limited detail).
• Content + performances: flag unedited traduction automatique, but judge translated pages on usefulness, originality, accuracy, and market fit — the méthode alone isn’t a policy violation. Trouver gaps and benchmark local competitors via Ahrefs country filters and Metrics by Country, keeping observed GSC/GA4 données separate from Ahrefs’ estimates, and earn local E-E-A-T — authority doesn’t travel.

Documentation officielle

Primary-source documentation pour international targeting.

Google

• Tell Google à propos de localized versions of votre page — the hreflang référence: return-tag requirement, fully-qualified URLs, x-default, and the three equivalent méthodes (HTML lien tags, HTTP Lien headers, XML sitemap).
• Managing Multi-Regional and Multilingual Sites — Structure d’URL trade-offs (ccTLD / subdomain / subfolder), geotargeting signals, and pourquoi IP-based adaptation and URL parameters are discouraged.
• How Google Crawls Locale-Adaptive Pages — pourquoi serving différent content by detected emplacement/language (au lieu de separate URLs) risks Google pas exploration, indexation, or ranking every locale, and Googlebot’s IP/Accept-Language behavior.
• x-default hreflang pour international landing pages (2013) — the original x-default référence.
• How x-default peut aider vous (2023) — mis à jour x-default utiliser cas.
• GSC Enhancements / structured-data and hreflang reporting — où hreflang problèmes surface now que the International Targeting report is gone.

Bing / Microsoft

• Bing Webmaster Outils — Geo Targeting (archived, 2013) — the old fonctionnalité que let vous définir targets at domain, subdomain, directory, or page level; supprimé from the redesigned Bing Webmaster Outils in 2020 and pas brought back. Bing à la place weights the Content-Language meta tag and HTTP header as its principal language/geo signal.
• Bing Webmaster Outils Removes Geo-Targeting Fonctionnalité (Moteur de recherche Roundtable, Sept 2020) — Fabrice Canel (Bing) confirms the removal and points to content-language meta/header as the replacement.

Quotes from the source

On-the-record statements relevant to an international audit.

Google — on hreflang complexity

• “TBH hreflang is one of the most complex aspects of SEO (if not the most complex one). Feels as easy as a meta-tag, but it gets really hard quickly.” — John Mueller, Google, Twitter/X, February 19, 2018. Coverage

Google — on the return-tag requirement

• “If page X links to page Y, page Y must link back to page X. If this is not the case for all pages that use hreflang annotations, those annotations may be ignored or not interpreted correctly.” — Recherche Google Central docs. Jump to quote

Google — on IP-based geolocation

• “IP location analysis is difficult and generally not reliable.” — Recherche Google Central docs (Managing Multi-Regional and Multilingual Sites). Lire the doc

Google — on locale-adaptive exploration

• “If your site has locale-adaptive pages (that is, your site returns different content based on the perceived country or preferred language of the visitor), Google might not crawl, index, or rank all your content for different locales.” — Recherche Google Central docs (How Google Crawls Locale-Adaptive Pages). Jump to quote
• “Don’t use IP analysis to adapt your content. IP location analysis is difficult and generally not reliable. Furthermore, Google may not be able to crawl variations of your site properly. Most, but not all, Google crawls originate from the US, and we don’t attempt to vary the location to detect site variations.” — Recherche Google Central docs (Managing Multi-Regional and Multilingual Sites). Jump to quote

Google — on the International Targeting report

• “The International Targeting report has been deprecated, and will be removed from Search Console soon.” Google ajouté que “we continue to support hreflang and our recommendations for managing multilingual and multiregional sites still stand.” (Supprimé September 22, 2022.) Coverage

The international SEO audit checklist

Organized by the six audit areas. Run it top to bottom.

1. URL / structure

• Locale inventory construit (market, URL pattern, audience, canonical, indexability, alternate méthode)
• Un structure choisi (ccTLD / subdomain / subfolder) and applied consistently
• Aucun mixed signals (TLD vs server/CDN emplacement vs content language vs lien internes)
• Aucun URL parameters utilisé pour language/country targeting
• GSC Enhancements vérifié pour hreflang errors (the old International Targeting report is gone)
• content-language meta tag or HTTP header définir pour Bing markets (Bing supprimé its Geo Targeting fonctionnalité in 2020; ce is now the preferred signal)

2. Hreflang

• Crawled at scale (Ahrefs Site Audit or Screaming Frog) — pas simplement spot-checked
• Manquant retourner tags fixed (la plupart critical)
• Broken / redirigé hreflang URLs fixed
• Non-canonical hreflang URLs fixed (point to the indexé URL)
• Language codes validated (ISO 639-1 + ISO 3166-1 Alpha-2)
• html lang matches hreflang language codes
• Self-referencing tags présent (meilleur pratique)
• x-default présent, pointing to an appropriate fallback (meilleur pratique)
• Ahrefs cluster graph reviewed pour chaque affected cluster
• Alternate URLs entièrement qualified, consistent format (aucun http/https or trailing-slash drift)

3. Geo-redirects

• Site testé à travers une requête-state matrix (aucun signal, US IP/VPN, varied Accept-Language, existing cookie, deep liens)
• Aucun site-wide geo-IP redirection sending Googlebot seulement to the US version
• Seulement the x-default / root URL auto-redirects; language pages jamais do
• Language selector disponible; cookie preference doesn’t trap utilisateurs

4. Technical

• html lang définir correctement on every page (UX/accessibility + non-Google engines)
• Content-Language HTTP header définir pour Bing markets (pas the deprecated meta tag)
• Hreflang in server-rendered HTML / headers / sitemap — pas client-side JS seulement
• Hreflang à l’intérieur the head, pas outside it
• Paginated pages have correct per-page hreflang sets (page 2 → page 2)
• Aucun international pages accidentally noindexed or robots.txt-blocked
• Server/CDN geolocation consistent with ccTLD targeting

5. Content

• Aucun unedited machine-translated pages live
• Titles, meta descriptions, and URL slugs localized (slugs = local keywords, pas transliteration)
• Content gaps identified per market (Ahrefs country filter)
• Local author bios / credentials on market-specific content
• Local trust signals présent (certifications, local contact details, local-platform reviews)

6. Performances

• Ahrefs Metrics by Country reviewed; underperforming markets flagged
• Trafic share comparé to keyword potential per market
• Local competitors benchmarked per market (Organic Competitors filtered by country)
• GA4 segmented by country; Core Web Vitals vérifié by region

How to prioritize ce que to fix

You’ll surface plus problèmes que vous pouvez fix at une fois. Sort every finding into a simple impact × effort matrix:

Mapping the hreflang error types onto it:

• Élevé impact + facile → fix now: manquant retourner tags on a petit définir of clusters, hreflang pointing at broken/redirigé URLs, incorrect language/country codes (jp→ja).
• Élevé impact + hard → plan now: site-wide geo-IP redirection blocking Googlebot (nécessite engineering + a language-selector pattern), client-side-JS hreflang on an SPA (nécessite SSR/sitemap rework), non-canonical hreflang from a deep canonical-chain problem.
• Low impact + facile → batch: manquant self-referential tags, manquant x-default, html lang mismatches — réel meilleur practices, but pas cluster-breaking, so do les in un réussir.
• Low impact + hard → deprioritize: minor server-geolocation tuning pour a ccTLD où TLD déjà dominates the signal.

The guiding principle: retourner tags and broken URLs break clusters; x-default and self-referential tags are robustness. Fix the breakers premier.

The 6-step audit sequence

Fonctionner the areas in ce order — chaque builds on the dernier:

1. URL / structure — confirmer the foundation is consistent avant auditing anything on top.
2. Hreflang — the core; explorer at scale, alors examiner the cluster graph and triage.
3. Geo-redirects — vérifier Googlebot peut en réalité reach every version.
4. Technical signals — html lang, Content-Language, JS rendering, pagination, server geo.
5. Content quality — localization, gaps, and local E-E-A-T.
6. Performances by market — mesurer où you’re winning/losing and benchmark locally.

Outils pour an international SEO audit

• Ahrefs Site Audit (principal). Runs the 9 hreflang checks (invalid annotation, manquant self-reference, multi-language/multi-page conflicts, hreflang↔HTML-lang mismatch, manquant retourner tag, non-canonical target, broken target, manquant HTML lang). Its hreflang cluster visualization — URL Details > Hreflangs tab — renders chaque cluster as a network graph with errors in red, the fastest façon to spot a broken cluster. Pair it with Site Explorer > Metrics by Country and Organic Competitors filtered by country pour the performances audit.
• Screaming Frog SEO Spider. Enable Configuration > Spider > Explorer Hreflang; ajouter referenced external domains sous Config > CDNs pour multi-domain (ccTLD) setups. Examiner the hreflang filters post-crawl and export via Reports > Hreflang. Aussi reads the Content-Language réponse header and flags hreflang placed outside the head and pointing to noindexed pages.
• Recherche Google Console. With the International Targeting report gone, vérifier the Enhancements section pour hreflang problèmes (limited detail), and utiliser Inspection d’URL to confirmer the indexé URL votre hreflang devrait point to.
• Bing Webmaster Outils. Aucun Geo Targeting fonctionnalité anymore — Bing supprimé it from the redesigned outil in 2020 and jamais brought it back. Audit Content-Language meta tag/header handling à la place, since that’s Bing’s réel signal and it differs from Google’s.
• technicalseo.com Hreflang Tags Outil — a manual validator pour generating and spot-checking hreflang annotations on individual pages.

Journal des modifications