Can Google arbitrarily choose which language version to index when the content is identical?

Why does Google canonicalize seemingly distinct language versions?

The challenge for Google is to avoid massive duplication in its index. If you publish 20 language versions of a product where only the currency symbol changes (€, $, £), technically you create 20 nearly identical pages.

Google then applies automatic canonicalization: it selects a reference URL and ignores the others in its main index. This is not a penalty — it’s a mechanism for saving crawl budget and consolidating signals.

Is hreflang still useful in this case?

Yes. Hreflang remains functional for user routing: a user in France will see the French URL, while another in Canada will see the Canadian URL. The difference is that these URLs will not all be indexed separately.

In practice, you will have a single URL in Search Console with impressions and clicks, but the actual traffic will be distributed via hreflang to the correct geographical variants.

What triggers this automatic canonicalization?

Google speaks of identical content. Practically, this means that the engine detects near-total textual similarity between the versions. If only the currency differs, the algorithm considers it to be the same duplicated content.

The nuance is important: actually translating content (even partially) generally avoids this forced consolidation. But a simple copy-paste with a currency symbol change won't fool anyone.

• Google chooses a canonical version among your language variants if the content is strictly identical
• Hreflang continues to function for user display, but not for indexing
• Search Console reporting occurs on the canonical version chosen by Google, not on each variant
• This mechanism aims to save crawl budget and avoid index pollution
• A true translation (even minimal) generally suffices to sufficiently differentiate the versions

Does this statement align with what is observed in the field?

Absolutely. We regularly see multilingual e-commerce sites where only one version dominates in the index, even though hreflang tags are technically correct. The classic trap: creating 15 European versions with just a currency change and believing that Google will index everything.

The engine applies a logic of signal consolidation. If you have 10 identical versions, it will prefer to index only one and concentrate ranking signals on it before redistributing via hreflang. This makes sense from its point of view.

What are the gray areas of this announcement?

Google does not specify how it chooses the canonical version. Is it based on the server's geolocation? On the version historically indexed first? On the volume of inbound links? [To verify] — no official data on this.

Another vague point: at what threshold of similarity does this canonicalization activate? If you change 5% of the text in addition to the currency, is that sufficient? Google remains evasive. In practice, it’s observed that minimal variations are not always enough.

In what cases does this rule not apply?

If your language versions have substantial content differences, even subtle ones, Google will treat them as distinct pages. An example: adapting product descriptions to cultural specifics (US vs EU sizes, local legal mentions, different arguments).

Additionally, if you have different ccTLDs (.fr, .de, .co.uk) rather than a single .com/fr, .com/de, Google may be less aggressive about canonicalization. But this is not a guarantee — content remains the determining factor.

What should be done practically to avoid this forced canonicalization?

The most robust solution: truly translate the content, even partially. Adapt product descriptions, selling points, customer testimonials — in short, create a real textual differentiation between versions.

If full translation is not feasible (too vast a catalog, limited budget), focus on strategic high-traffic pages. Better to have 10 well-differentiated versions than 50 cloned versions with just a € symbol replaced by $.

How to check if Google has canonicalized my language versions?

Use the URL Inspection tool in Search Console. For each language variant, check which URL Google considers canonical. If multiple of your URLs point to the same canonical, you are in this situation.

Another indicator: compare the indexing data in Search Console between your different language versions. If some show zero impressions despite correct hreflang, they are probably not indexed separately.

What mistakes should absolutely be avoided?

• Do not only change the currency between language versions
• Do not ignore canonicalization signals in Search Console
• Do not multiply language versions if you lack the resources to differentiate them
• Do not forget that hreflang does not guarantee separate indexing of variants
• Do not neglect cultural adaptation in favor of a simple automatic translation