Are translated pages really treated as unique content by Google?

Why doesn’t Google consider translations as duplicates?

The answer from Mueller resolves an ongoing debate: the same article translated into French, English, and Spanish does not constitute duplicate content. The engine analyzes different words and distinct lexical fields, indexing each version autonomously.

This approach acknowledges that each language targets a differentiated audience, with potentially divergent search intents even if the topic is identical. Therefore, Google does not penalize sites that deploy translated content — as long as the implementation is clean.

Is hreflang essential for multilingual sites?

According to Mueller, hreflang is not necessary if traffic is already correctly reaching each language version. In other words: if Google sends French speakers to the French version and English speakers to the English version without technical intervention, hreflang annotations become optional.

That said, hreflang remains the most reliable signal to explicitly indicate relationships between language versions. Without it, Google can theoretically interpret the structure correctly — but there’s no guarantee it won’t make mistakes, especially in markets where language and geography do not coincide perfectly (multilingual Switzerland, French/English Canada, etc.).

Can each translation rank on its own queries?

This is exactly what Mueller confirms: each translated version is indexed independently and can position itself on keywords in its target language. A well-optimized translation can therefore rank for local queries even if the original version does not perform.

The risk? That two language versions cannibalize each other if they target the same market and the same language — for example, a .com site in US English and a .co.uk site in UK English on British queries. In this specific case, hreflang becomes critical to prevent Google from displaying the wrong version.

• Translations = distinct pages in Google’s eyes, independent indexing guaranteed
• Hreflang optional if traffic is already arriving at the correct language without technical help
• No duplicate content across language versions of the same content
• Each version can rank on its own keywords in its target language
• Be cautious in multilingual markets where hreflang remains essential to avoid confusion

Does this statement align with on-the-ground observations?

Yes, and it’s even one of the few points where the official doctrine perfectly matches practical reality. Well-structured multilingual sites indeed see each version indexed and ranked independently, without a duplicate penalty — provided the technical structure is clean (no mixed versions on the same URL, no partially translated content, etc.).

Where it sometimes falters: sites that mix languages and geographies without consistency or that serve the wrong language based on IP without respecting user-agent. In these cases, Google can indeed sort it out without hreflang — but it's playing with fire.

Should we really do without hreflang as suggested by Mueller?

Let’s be honest: saying hreflang is "not necessary" is technically correct, but pragmatically risky. If your site targets several languages in geographically overlapping markets (Europe, North America, multilingual Asia), hreflang remains the only reliable way to control which version appears in which SERP.

Without hreflang, Google does its best — but it can make mistakes, especially on ambiguous queries or for bilingual users. Mueller’s advice particularly applies to sites with a clear geographical separation (a .fr for France, a .de for Germany, etc.) and contents that are clearly distinct by language. Once you have a global .com with /en/ and /fr/, hreflang becomes almost indispensable.

What on-the-ground errors does this statement not cover?

Mueller does not mention poor-quality machine translations — and this is where many sites go wrong. Google may consider each translation as distinct, but if your French translated by a crude AI is riddled with errors, the French version will rank poorly or not at all, regardless of the original’s performance.

Another blind spot: cross-language canonicals. Some sites canonicalize all their translated versions to English, thinking they can "consolidate" their authority. This is a fatal mistake: Google will then index only the English version and ignore the other languages. Mueller's statement implicitly confirms this — but it would have been helpful had he stated it explicitly. [To verify]: Does Google actively penalize cross-language canonicals, or does it merely not index the canonicalized versions?

What concrete steps should be taken to structure a multilingual site?

The first rule: a distinct URL per language. Subdomain (en.site.com), subdirectory (site.com/en/) or ccTLD (site.co.uk) — all three work, but the subdirectory remains the easiest to manage technically and allows for consolidating domain authority.

The second rule: implement hreflang even if Mueller says it’s not required. The cost of implementation is low compared to the risk of Google displaying the wrong version in the SERPs. Place hreflang annotations in the XML sitemap rather than in the HTML — it's easier to maintain at scale.

What mistakes should absolutely be avoided on a translated site?

Never canonicalize a translated version towards another language. Each version should point to itself as canonical, or not have a canonical at all. If you canonicalize /fr/ to /en/, Google will only index the English version — and the French version will disappear from Francophone SERPs.

Also avoid mixed content: pages that are half translated, with blocks in language A and others in language B. Google detects the dominant language and can misclassify the page, or even consider it spam if the mix seems artificial. Translate fully or not at all.

How can I check that my multilingual site is correctly indexed?

Use the Search Console with a property for each language version (or by domain if you use ccTLDs). Check that each version generates impressions in its target language, and that translated pages are not cannibalizing each other for the same queries.

Also test manually: perform searches from different countries (VPN or Incognito) and check that Google is displaying the expected language version. If not, your hreflang is probably misconfigured or missing.

• Distinct URL per language: subdirectory, subdomain, or ccTLD
• Implement hreflang in the XML sitemap to link versions together
• Self-referential canonical: each version points to itself, never to another language
• Complete and quality translation: no mixed content, no raw Google Translate
• Search Console per version: monitor indexing and performance for each language independently
• Cross-country tests: ensure the correct version appears in target SERPs