Should you really use hreflang for untranslated pages?

What does it really mean to "use hreflang with untranslated pages"?

The principle is simple: you have content in French, for example, and you want to offer it to other language markets (English-speaking, German-speaking, etc.) without translating it. Technically, you can create multiple URLs — /fr/, /en/, /de/ — all pointing to the same French content, using a canonical tag to a reference version and implementing hreflang to signal to Google that these URLs are language variants.

Google tolerates this practice. But tolerance is not the same as recommendation. Mueller emphasizes the risk of authority dilution: multiplying URLs without added value (here, no actual translation) tends to fragment your ranking signals instead of concentrating them on a single strong URL.

Why does Google mention "authority dilution" in this context?

Because in SEO, page authority (PageRank, incoming links, behavioral signals) is built URL by URL. When you create /en/ and /de/ that display the same content as /fr/, you potentially divide backlinks, social shares, and engagement signals among three URLs instead of one.

Sure, the canonical tag is supposed to consolidate these signals to the reference version. But in practice, it's rarely that clear-cut: Google doesn't always strictly follow canonicals (especially if hreflang sends conflicting signals), and third-party authority measurement tools (Ahrefs DR, Majestic TF) don't always recognize canonical consolidation. The result: you lose clarity and power.

In what cases might this approach still be justified?

There are situations where an international site temporarily lacks the resources to translate its entire catalog. Rather than completely blocking access to certain markets, some publishers create placeholder URLs with hreflang, while waiting for actual translation. This is a stopgap measure, not a sustainable strategy.

Another case: very visual content (portfolios, photo galleries) where the text is minimal. Here, the difference between /fr/ and /en/ may be limited to a few translated menus, and the rest of the content is identical. But even in this case, Mueller suggests that a single multilingual URL (with a language selector on the interface side) would be more effective.

• Hreflang with untranslated content: technically possible, but it dilutes authority.
• Canonical + hreflang: Google can follow them, but nothing is guaranteed — conflicting signals create confusion.
• Better to consolidate on a reduced number of strong URLs, with content genuinely differentiated by language.
• If translation is not immediately possible, prioritize a single multilingual URL rather than multiplying clone URLs.
• Third-party authority measurement tools do not always recognize canonical consolidation — you lose metric visibility.

Is this statement consistent with field observations?

Yes, and it serves as a wake-up call. We regularly see international sites creating /en/, /es/, /de/ without translating anything, just to "be present" in those markets. The result: Google indexes 5 URLs with the same content, users land on a French page while searching in English, and the bounce rate explodes.

In practice, Google manages contradictory canonicals with hreflang quite well — but "quite well" doesn’t mean "perfectly". We often observe that Google chooses its own canonical version, ignoring the one you've declared. When hreflang says "this page is the English version" and canonical says "the French version is the reference", Google has to arbitrate — and it doesn’t always choose what you expect. [To be checked] on your own site with regular monitoring of indexing.

What nuances should we add to Mueller's recommendation?

The idea of "authority dilution" is true, but it doesn’t apply uniformly to all sites. A large e-commerce site with an aged domain and strong overall authority can afford a few redundant URLs without dramatic impact. A younger or less authoritative site, on the other hand, must absolutely focus its strength.

Moreover, Mueller does not specify how measurable this dilution is. It's a matter of common sense rather than a documented algorithmic penalty. You won't lose 50% of rankings just because you created an untranslated /en/ URL. But you lose editorial clarity, user experience, and potentially the ability to attract quality backlinks — because an English-speaking site will never link to a French page it believes to be in English.

When does this rule not really apply?

If your site has a complex technical architecture (multi-zone CDN, regional subdomains imposed by your tech stack), you may be technically constrained to create multiple URLs even if the content is not translated. In this case, the canonical + hreflang approach becomes a lesser evil.

Another exception: ultra-technical niche sites where English vocabulary is universal (e.g., developer documentation, technical specs). Here, even a French-speaking audience searches in English. Creating /fr/ and /en/ with the same English content can make sense if search queries differ by market. But this is a rare case, and you should still localize at least the title/meta tags.

What should you do concretely if you already have untranslated URLs with hreflang?

First reflex: audit indexing. Check in Google Search Console which versions are indexed for which markets. If Google indexes the /en/ version while the content is in French, it’s a clear signal that your hreflang + canonical implementation is not functioning as expected.

Next, measure traffic impact: do these untranslated URLs generate qualified traffic, or just bounce? If the bounce rate exceeds 80% and the time on page is under 10 seconds, you have your answer — these pages harm your user experience and probably your Core Web Vitals (CLS, INP) through fast exits.

What mistakes should you absolutely avoid in this configuration?

Never create an /en/ URL with hreflang="en" if the content displayed remains in French. This is the worst mistake: you promise Google (and users) an English page, and you serve French. Google may see this as reverse cloaking or misleading content.

Also avoid canonical chains: /de/ → canonical to /fr/, and /fr/ → canonical to /en/. Google follows one or two hops, but beyond that, it abandons. If you absolutely must canonicalize, do it directly to the reference version, without intermediaries.

How to verify that your site complies with this recommendation from Mueller?

Use a crawler (Screaming Frog, Oncrawl) to extract all your URLs with hreflang attributes. Compare the displayed content (via MD5 hash of the HTML body, or lexical analysis) to identify content duplicates between language versions. If two URLs have the same MD5 hash but different hreflang, you are at risk.

Next, check Google Search Console reports "Coverage" and "hreflang" (available in the old interface or via API). Google often flags hreflang/canonical conflicts. If you see warnings "Canonical tag with hreflang pointing to another language", that's a red flag.

• Audit actual indexing in GSC for each language market
• Measure bounce rate and time on page of untranslated URLs — if >80% bounce, remove or consolidate
• Crawl the site to detect content duplicates with different hreflang
• Eliminate canonical chains: always point directly to the reference version
• Never promise a language (hreflang) that the content does not respect
• Prioritize translation or consolidation on a single multilingual URL