Should you really translate URLs for a multilingual website, or can you keep them in one language?

Why doesn't Google favor translated URLs?

The search engine treats URLs as technical identifiers, not as primary linguistic signals. Whether your structure displays /fr/chaussures-running/ or /fr/running-shoes/, the algorithm relies first on the page content, the lang tags, and especially on hreflang to determine the geographical and linguistic target.

This neutrality offers valuable operational flexibility. A site can maintain URLs in English across all its versions without SEO penalties — a common practice among tech pure players that maintain consistency in their backend systems. Conversely, translating slugs can improve the click-through rate in SERPs if your audience scans URLs before clicking, but this gain relates to user behavior, not ranking.

What makes a URL truly unique according to Google?

Uniqueness relies on the absolute distinction between each language version. /fr/ and /fr-ca/ must point to differentiated content, even if 80% of the text remains the same. Google indexes these URLs separately and expects each hreflang to correctly reference its alternatives.

The classic pitfall? URL parameters like ?lang=fr instead of a clear structure. These variants create indexing ambiguities that even a perfect hreflang struggles to compensate for. Google can crawl and index multiple versions of the same page if the URL changes dynamically, thus diluting the relevance signal.

Does hreflang replace all other linguistic signals?

No, it complements them without negating them. The hreflang tag acts as a conductor that coordinates the signals: it tells Google, "this French page has a German sister here." But if the content, HTML lang tag, or metadata contradict this indication, the algorithm may ignore the hreflang.

A misconfigured hreflang generates silent errors — Google does not block indexing, it simply chooses not to apply your directives. The result? A German user receives the French version in their results, and vice versa. The reciprocity of annotations (if A points to B, B must point to A) remains a non-negotiable rule.

• The URL can remain in English across all versions without direct ranking impact
• Each language/region requires a distinct and stable URL (avoid dynamic parameters)
• Hreflang functions in a network: all alternative pages must reference each other
• The consistency of signals (HTML lang, content, hreflang) takes precedence over the chosen URL structure
• Hreflang errors are invisible to the user but gradually degrade the geographical distribution of traffic

Does this statement align with field observations?

Yes, and it's confirmed by audits on thousands of multilingual sites. .com domains with uniform English URLs but impeccable hreflang consistently outperform competitors with translated slugs but shaky annotations. The case of Booking.com illustrates this: English URLs, excellent local performance everywhere.

However — and Mueller does not say this — translating URLs boosts organic CTR by 8 to 15% in certain sectors (fashion e-commerce, travel). This is not a ranking factor; it's a psychological trigger. A French user scans the URL displayed in SERPs: seeing /fr/chaussures-cuir/ is more reassuring than /fr/leather-shoes/. Google does not reward this algorithmically, but the post-click behavior (time on page, navigation) can indirectly influence UX signals.

What are the gray areas that Google doesn't address?

Mueller remains vague on hybrid sites: what to do when content targets fr-FR and fr-BE with 90% common text? Fully duplicate or share with hreflang x-default? Google officially recommends distinct pages, but in practice, many high-performing sites use a generic FR page + hreflang x-default without suffering. [To verify]: no published Google study quantifies the impact of this approach on local ranking.

Another silence: subdomains vs subdirectories. fr.site.com vs site.com/fr/ — Mueller does not rule. Both work if the hreflang is set up correctly, but the subdirectory inherits trust from the main domain (backlinks, history), while the subdomain starts from scratch in terms of authority. This technical nuance escapes the official statement.

In what cases does this flexibility become a trap?

When it justifies laziness in implementation. "Google says that the URL doesn't matter" becomes the excuse not to translate, not to structure, to leave ?lang= hanging around. The result: a site technically indexable but humanly off-putting. A Spanish user seeing /en/products/leather-bag/ understands intellectually, but their brain prefers /es/productos/bolso-cuero/.

Another trap: neglecting internal consistency. URLs in English, content in French, but internal links mixing /fr/ and /en/ in the anchors. Google indexes everything, but the interlinking becomes unreadable for the algorithm of thematic relevance. Internal PageRank dilutes, linguistic silos fissure.

How to choose the URL structure for a multilingual website?

Start by balancing SEO and UX. If your audience scans URLs (B2C sectors, public sectors), translate slugs to maximize CTR. If you're targeting professionals or a tech audience (SaaS, technical documentation), keeping English simplifies maintenance without SEO cost.

Always prefer subdirectories (/fr/, /de/) over a single domain rather than separate subdomains or ccTLDs, unless legal constraints like GDPR with local hosting apply. The subdirectory shares domain authority and simplifies hreflang. Absolutely avoid ?lang= parameters: Google indexes them poorly and Search Console reports cascading canonical/hreflang errors.

What silent errors are killing your international traffic?

The broken reciprocity of hreflang: your FR page points to DE and ES, but DE does not mention FR. Google detects the inconsistency and ignores the entire cluster. The result: each version competes against each other for the same keywords, cannibalizing your own traffic.

Another plague: the misconfigured hreflang x-default. It serves as a fallback page when no language matches — often the homepage or a geographic selector. Many point x-default to the default EN version, sending untargeted users to English content even when a closer version exists. The x-default must be neutral or direct to a true language choice.

How to audit and correct an existing implementation?

Use the Search Console, "International Targeting" section: it lists all hreflang errors (invalid language codes, unreachable URLs, missing reciprocity). Cross-reference with a Screaming Frog crawl in "Render JavaScript" mode to verify that annotations appear in the final DOM if your site is in JS.

Test the actual rendering with the URL inspection tool: force Google to crawl a page and check that hreflang appears in the returned HTML. If your annotations are in client-side JavaScript, they may arrive too late for Googlebot. In this case, migrate them to the HTTP header Link or the XML sitemap.

• Choose a consistent structure (/language/ or /language-region/) and apply it uniformly
• Implement hreflang in the <head>, in the HTTP header or via sitemap — never a mix of all three
• Check reciprocity: each page must list all its siblings, including itself
• Test with different IPs/language browser settings to confirm that Google serves the right version
• Monitor the Search Console "International Targeting" monthly for regressions
• Document your hreflang logic in a Google Sheet: source URL, all its alternatives, language codes used