SEO Guides

International SEO Audit: Going Global Without Losing Rankings

2026-06-13·8 min read

Key Takeaways

Most international SEO failures trace back to three auditable layers: hreflang, URL architecture, and content localization
A single broken hreflang return-link can cause Google to ignore every language annotation on your domain
Subdirectories (example.com/fr/) keep domain authority consolidated, while ccTLDs force each country to build authority from zero
Direct keyword translation often misses local search behavior -- the high-volume term in each market must be validated with local keyword data, not assumed from translation

A multinational retailer launches its German storefront. Within weeks, Google starts serving the German product pages to English-speaking users in the US, while German shoppers see the English originals. Organic traffic drops across both markets. The cause: a missing return-link in their hreflang implementation. Every page pointed from English to German, but the German pages never pointed back.

This kind of failure is common because international SEO breaks across three layers simultaneously -- hreflang configuration, URL architecture, and content localization. Skip any one layer and rankings degrade in every target market at once. An international SEO audit works through all three layers systematically, plus the local signals that determine whether your pages actually win in each geography.

Whether you're running an SEO audit London-focused, expanding across the EU, or entering APAC markets, the audit structure below covers what matters and what to check first.

Broken Hreflang Silently Kills Multi-Market Rankings

Hreflang tags tell search engines which language-region version of a page to serve. They are also the single most common point of failure in international SEO. Google's own documentation states that hreflang annotations are "hints, not directives" -- meaning Google can (and does) ignore them entirely when the implementation is inconsistent.

According to a 2023 Ahrefs study of 374,756 domains using hreflang tags, 67% had at least one hreflang implementation issue severe enough to reduce or eliminate the tags' effectiveness. The most damaging errors fall into a predictable pattern:

Bidirectional reference failures

If page A declares page B as its French alternate, page B must declare page A as its English alternate. Google's documentation is explicit: when two pages do not confirm each other's annotations, the annotations are ignored. This is the error that hit the retailer in our opening example -- and it is the most common one at scale because CMS migrations and new market launches often update only the source-language templates.

Conflicting signals

Hreflang pointing to redirected URLs instead of final destinations
Hreflang on pages with noindex directives (contradictory instructions to the crawler)
Hreflang declared in both the HTML <head> and the XML sitemap with different values
Missing x-default fallback, leaving unmatched languages to Google's guesswork

Validation checklist

Use ISO 639-1 language codes (en, fr, de, zh) and ISO 3166-1 Alpha-2 region codes (US, GB, FR)
Confirm every page includes a self-referencing hreflang tag
Crawl the full site to verify bidirectional links -- manual spot-checking misses orphaned pages
Verify x-default points to your primary language or a language-selector page

URL Architecture Determines How Authority Flows Between Markets

Your URL structure decides whether international pages share or compete for domain authority. This choice is hard to reverse after launch, so the audit should confirm the architecture matches the business's resources and goals.

Structure	Example	Authority Model	Trade-off
Subdirectories	example.com/fr/	All pages share one domain's authority	Weaker geo-targeting signal per country
Subdomains	fr.example.com	Google treats each as a separate site	Authority splits across subdomains
ccTLDs	example.fr	Each domain starts from zero authority	Strongest geo signal, highest cost and effort

Google's multi-regional site documentation outlines the trade-offs without declaring a winner: subdirectories are easy to set up and let pages inherit existing domain signals, while ccTLDs provide the strongest geographic signal but require building authority independently for each domain. There is no universally superior option. The right choice depends on your existing domain strength, number of target markets, and operational capacity. Brands entering a single high-priority market (e.g., a US company targeting only Japan) sometimes benefit from a local ccTLD because it sends a strong trust signal to users who associate .jp domains with local businesses.

For the audit, check three things beyond the structure itself:

CDN coverage -- page speed varies by geography. A subdirectory site without edge nodes in target markets loses the speed advantage of a locally hosted ccTLD.
Locale signals beyond hreflang -- use locale-specific URLs, correct hreflang annotations, and explicit in-page language/region links. Google's legacy International Targeting setting in Search Console has been retired; geo-targeting now relies entirely on URL structure, hreflang, and on-page signals.
Internal linking -- cross-language links should use hreflang-annotated canonicals, not raw URLs that bypass language routing.

Translated Keywords Often Target the Wrong Searches

Machine translation produces grammatically passable content that ranks poorly because it translates words rather than search intent. The keyword layer is where this matters most.

Consider "cheap flights" as a target keyword. A direct French translation gives "vols bon marché," but the high-volume term French users actually search may be "vol pas cher" -- a colloquial phrasing no literal translation would produce. The exact volume difference must be validated with local keyword data (Google Keyword Planner or Google Trends set to the target country), but this pattern of divergence between literal translation and actual search behavior exists in nearly every language pair, and it compounds across hundreds of product or service pages.

An international content localization audit should verify:

Per-market keyword research -- are target keywords based on actual local search volume data, or back-translated from the English keyword list?
Cultural adaptation -- currency, date formats, units of measurement, imagery, and examples should match local expectations. A US case study on a French landing page signals to both users and search engines that the content was not built for this market.
Schema markup localization -- LocalBusiness schema should carry the local address, phone number, and business hours for each market. Currency in Product or Offer schema must match what the page displays.
Mobile-first localization -- mobile traffic share varies sharply by market. Validate the device split for each target country (StatCounter or GA4 by region) before designing localized pages. Building desktop-first for a mobile-dominant market wastes your localization investment.

Local Signals Decide Who Wins in Each Geography

Hreflang, URL structure, and content localization get your pages indexed correctly. Local signals determine whether those pages actually rank above local competitors. This is especially visible in city-level targeting -- running an SEO audit London, for example, requires checks that go beyond what a standard international audit covers.

Google's documentation on local ranking explains that results are determined by three factors: relevance (how well the listing matches the query), distance (proximity to the searcher), and prominence (how well-known the business is, based on links, reviews, and broader web signals). For international sites targeting specific cities or regions, prominence in that local market matters more than raw global domain strength. Auditing local signals means checking:

Local backlink profile -- links from country-specific directories, regional industry associations, and local press matter more than high-DA international links for geo-targeted queries
Google Business Profile -- for any market where you want to appear in local pack results, a verified GBP with the correct local address, category, and hours is non-negotiable
Regulatory compliance -- UK sites need GDPR cookie consent and ICO registration; EU sites need GDPR; other markets have their own requirements. Non-compliance affects both user trust and search visibility
AI search variation -- AI-generated answers differ by language and region. The same query in English and French may surface entirely different recommended businesses. Track AI visibility per market to catch gaps that traditional rank tracking misses.

The framework scales to any city or region. Replace "London" with "Sydney," "Dubai," or "Toronto" -- the audit categories are identical, the specific signals change.

MendMySEO audits sites in any language with hreflang validation, localized schema checks, and AI search visibility tracking across regional markets. Join the waitlist.

Frequently Asked Questions

How many markets can I launch at once without breaking existing rankings?

There is no universal cap, but a staged approach reduces risk. Launch one new market at a time, validate hreflang bidirectional links and indexing status for the new pages, then wait 2-4 weeks for Google to process the annotations before adding the next market. Launching five markets simultaneously multiplies the chance of hreflang conflicts and makes it harder to isolate which market introduction caused a traffic drop.

How do I prioritize which hreflang errors to fix first?

Start with pages that have the most organic traffic at stake. A broken return-link on your top 10 landing pages will cause more damage than the same error on a low-traffic blog post. After high-traffic pages, fix noindex conflicts (pages with both hreflang and noindex), then address missing x-default tags across the site.

When does a ccTLD make more sense than a subdirectory?

When you are entering a single high-priority market with a dedicated local team and budget to build authority from scratch. Japan (.jp) and China (.cn) are common examples where local users have strong domain-extension preferences. If you are expanding to three or more markets simultaneously, subdirectories almost always win because you cannot build independent domain authority across multiple new ccTLDs at the same time.

What is the fastest way to check if my translated content targets the right keywords?

Pull the top 20 keywords driving traffic to your English pages, then check each translated equivalent against actual search volume in the target language using a tool like Ahrefs, SEMrush, or Google Keyword Planner set to the target country. If more than a third of your translated keywords show near-zero volume compared to alternative local phrases, the translation was likely done word-for-word rather than intent-for-intent.

Does Google treat AI-translated content differently from human-translated content?

Google's stated position is that it evaluates content quality regardless of how it was produced. In practice, AI-translated pages that pass a native-speaker quality review and target locally researched keywords perform comparably to human translations. The risk is not the translation method itself but skipping the local keyword research and cultural adaptation steps -- which happens more often with AI translation because the speed encourages a "translate and ship" workflow without local review.