A multilingual site can rank perfectly well while sending searchers to the wrong regional page. The UK URL appears in the United States, the global page replaces the German version, and a country team blames weak demand for what is really a routing problem. Hreflang exists to clarify those relationships, but only when the URL map and every technical signal agree.
This is not a rare issue. In a 2023 study of 18,786 websites with hreflang in their HTML, search analyst Dan Taylor found that 31.02% had conflicting directives. The same homepage-level study found missing self-references on 16.04% of sites and at least one invalid language code on 8.91%. Its methodology did not cover every deeper URL or implementations delivered only through XML sitemaps or HTTP headers, so treat those figures as a useful warning rather than a universal failure rate.
Broken hreflang does not automatically remove a page from the index. The more familiar symptom is subtler: Google shows the UK page to a US searcher, the global English URL earns impressions in Germany, or a localized page never becomes the preferred result in its intended market. Marketing reports then mix technical routing errors with genuine demand and ranking changes.
A useful audit follows the same order as Google’s decision path: define equivalent URLs, validate the annotations, repair return links, remove indexing conflicts, and then check what Google actually selected.
Mapping Language and Regional URL Versions
Start with a complete inventory of the international URLs. A crawler can tell you which tags exist, but it cannot decide whether a German service page is genuinely equivalent to the US page your template assigned to it. The URL map must settle that decision.
Start by exporting your current URL structure into a spreadsheet. Group the URLs into clusters. A cluster is simply the set of pages that contain the same core content but are translated or localized for different audiences. For example, your pricing page in US English, UK English, and German forms one cluster. Every URL in that cluster must eventually point to every other URL in the cluster, including itself.
Large catalogs become difficult when product availability, legal copy, and URL patterns vary by market. This is often the point where an in-house team asks an SEO agency to review the cluster map and reconcile CMS rules with the intended market structure before new tags are generated.
Once you have your clusters mapped, decide on your implementation method. Google supports hreflang via HTML head tags, HTTP response headers, and XML sitemaps. Google’s official documentation states that all three methods are treated equally. Choose one method and stick to it.
Google says there is no Search benefit to repeating the same hreflang mapping in several places, and parallel implementations create more records that can drift out of sync. For a content-heavy site, XML sitemaps can be easier to update without changing every page template. HTML head tags are equally valid when your CMS can generate and maintain them consistently. The right choice is the one your team can keep accurate after launches, migrations, and market closures.
Validating Hreflang Codes and Target URLs
A malformed value is easy to miss and enough for Google to ignore the affected annotation. Validate the language-region pair before investigating more complicated indexing problems.
Language codes must follow the ISO 639-1 format. Regional codes, which are optional, must follow the ISO 3166-1 Alpha 2 format. The language code must always come first. A frequent mistake is using familiar abbreviations instead of the official ISO codes. For example, the correct regional code for the United Kingdom is gb, not uk. If you write en-uk, search engines will not recognize it. Furthermore, you cannot use a regional code by itself; it must always be paired with a language code.
You must also validate the target URLs within your tags. Google requires every alternate to be fully qualified, including the protocol and hostname. Use https://example.com/de/preise/, not a relative path such as /de/preise/. The same rule applies whether the mapping lives in HTML, an HTTP header, or a sitemap.
Finally, check the status of the target URLs. An hreflang tag should point directly to a live, indexable page that returns a 200 OK status code. If your tag points to a URL that redirects, or worse, returns a 404 error, you are sending search engine bots into a dead end.
This is a common issue after a migration. Internal links may point to the new URL while an old hreflang export still points to the redirecting address. A redirect is not the final indexable alternate you want Google to associate with the cluster. Update the tag to the destination URL, then run the complete export through a status checker and confirm every target resolves to the intended 200-status page on the first request.
Finding Missing Reciprocal Hreflang Tags
Hreflang tags operate on a strict rule of reciprocity. If Page A claims that Page B is its alternate version, Page B must point back to Page A and confirm the relationship. If the return link is missing, the tags can be ignored. Search engines enforce this rule so that a random website cannot arbitrarily claim to be the alternate version of your content.
During your audit, you must check every cluster for broken loops. This is where crawling tools become essential. Run your site through a crawler and filter for hreflang errors. Look specifically for “missing return links” or “non-reciprocal hreflang” warnings.
You also need to check for missing self-referencing tags. Every page in a cluster must include an hreflang tag pointing to its own URL. If you have an English page and a French page, the English page must have a tag pointing to the French page, and a tag pointing to itself. The 2017 Semrush study found that when sites had hreflang conflicts in their source code, 96% of the time they were missing a self-referencing tag.
If you cannot maintain a complete set of bidirectional links for every language on a massive site, Google notes that you can omit some languages on some pages. However, you must ensure that new regional pages link bidirectionally to your dominant language versions.
For instance, a new German-language Swiss page marked de-ch should link back to the generic German version from which it was launched, and that de page must return the link. The new page does not have to be added to unrelated clusters, but every annotation that does appear must form a valid two-way relationship.
Resolving Canonical, Redirect, and Indexing Conflicts
Hreflang tags do not work in isolation. They interact heavily with your canonical tags and redirect rules. If these signals contradict one another, search engines have to guess what you want, and they often guess wrong.
Hreflang and canonical annotations must describe the same architecture. A French page intended for French users should normally be self-canonical, while the English alternate should be self-canonical on its own URL. Pointing the French page’s canonical to the global English page gives Google a strong signal that the English URL is the representative version, undermining the alternate mapping. Google advises choosing a canonical in the same language when hreflang is present, or the closest available substitute when no same-language canonical exists.
Automatic location redirects create another conflict. Google warns that locale-adaptive pages may not be fully crawled because Googlebot usually originates from the United States and does not send an Accept-Language header in the way a human browser might. Keep every regional URL crawlable, offer users a visible country or language switcher, and use x-default for a neutral selector or fallback page when no listed locale is a clear match.
A historical agency case shows what the wrong regional selection can look like at scale. In 2013, Seer Interactive reported that a client’s UK domain was outranking its US subdomain in American results. After hreflang was implemented, reported indexation for the US subdomain rose from about 230,000 URLs to 655,000.
Use that case carefully. It predates today’s indexing systems, came from the agency that performed the work, and was not a controlled experiment. The raw before-and-after counts do not prove that hreflang alone caused the change. They do illustrate why an audit should compare regional indexation and visibility rather than treating a technically present tag as evidence that Google accepted the mapping.
Verifying Correct Regional URLs in Search
Once you have mapped your clusters, validated your codes, ensured reciprocity, and aligned your canonicals, you need to verify that Google actually understands your setup.
Do not rely on manual Google searches to check your work. Search results are heavily personalized based on your location, device, and search history. Instead, use Google Search Console.
First, use the URL Inspection tool. Enter a localized URL and check the “Page indexing” section. You want to see that the “Google-selected canonical” matches your “User-declared canonical.” If Google has selected a different URL, you still have a conflict to resolve.
Next, open the Performance report. Filter the data by a specific country, such as Germany, and look at the Pages tab. You should see your /de/ URLs generating the impressions and clicks. If your global English URLs are still getting the bulk of the traffic in Germany, your hreflang implementation is not fully recognized yet.
Compare equal pre-audit and post-audit periods, then review both the Page and Country dimensions. Recrawling and canonical processing take time, so a same-day check tells you little. Over the following weeks, look for the intended regional URL gaining a larger share of that country’s impressions while the wrong version loses share.
The movement will not always be one-for-one. Search demand, rankings, seasonality, and unrelated site changes can move at the same time. Keep an annotation of the deployment date, inspect a representative sample of clusters, and separate correct URL selection from traffic growth. A technically successful fix can route existing impressions to the right page before total clicks begin to rise.
