A business that sells in more than one country, or publishes in more than one language, has to decide how to lay the site out. The decision shapes the URLs, the hosting, the maintenance and how search engines understand who each page is for. It is also hard to undo once pages are indexed. Google documents the options in Managing multi-regional and multilingual sites, and this article summarises it, with our view of what follows.
Multilingual and multi-regional are different problems
Google separates the two. A multilingual site offers content in more than one language, for example a Canadian business with English and French versions, and Google tries to match the searcher’s language. A multi-regional site explicitly targets users in different countries, for example a manufacturer that ships to both Canada and the United States, and Google tries to find the right locale page for the searcher. Some sites are both: US and Canadian versions, with French and English versions of the Canadian content.
The distinction matters because the fixes differ. Language is mostly about labelling and navigation; country targeting is mostly about URL structure and signals.
Declare each language version on its own URL
For identical content in several languages, Google recommends a different URL for each language version, rather than using cookies or browser settings to change the language on the same page. It advises hreflang annotations to link the versions. It warns that if you reroute or change content dynamically by language, Google might not find and crawl every variation, because Googlebot usually originates from the USA and sends requests without an Accept-Language header.
Two further points are easy to miss. Google says it uses the visible content of a page to decide its language, not lang attributes or the URL, so each page should use one language for content and navigation. And it advises against automatically redirecting people between language versions, suggesting links that let them choose. Our article on hreflang return tags covers the annotation side.
Boilerplate, translation and URLs in other scripts
Google makes two further points that affect how pages are built. Translating only the boilerplate of a page, while the bulk of its content stays in one language, can create a poor experience if the same content appears several times in search results under different boilerplate languages. Google gives user-generated content pages as the usual cause. On URLs, it says it is fine to use localised words in the path, or an Internationalized Domain Name, provided the encoding is UTF-8 and the URLs are escaped properly when linked to.
Neither point changes the structure decision, but both are cheap to get right at the start and awkward to repair later.
The four URL structures Google compares
- Country-specific domain such as example.de. Google lists clear geotargeting, server location being irrelevant and easy separation of sites as advantages, and expense, limited availability, more infrastructure and sometimes strict registration requirements as drawbacks. It can target only a single country.
- Subdomain on a generic domain such as de.example.com. Easy to set up, allows different server locations and separates sites easily; users may not recognise the geotargeting from the URL alone.
- Subdirectory on a generic domain such as example.com/de/. Easy to set up and low maintenance on one host; users may not recognise the targeting, there is a single server location and separation is harder.
- URL parameters such as site.com?loc=de. Google marks these as not recommended, because segmentation is difficult and the targeting is not obvious.
Google does not say which of the first three to choose. Its table is a statement of trade-offs, and the right answer depends on how separate the markets are, who runs them and how much infrastructure the business wants to carry.
How Google decides who a page is for
Google says it relies on several signals to determine the target audience for a page. Country-code top-level domains are a strong signal, though some countries restrict who can register them. Hreflang statements, in tags, headers or sitemaps, are another. Server location, taken from the IP address, can indicate the audience, but Google notes that CDNs and overseas hosting make it a non-definitive signal. Other cues include local addresses and phone numbers, local language and currency, links from other local sites and, where available, Business Profile signals.
Generic domains behave differently. Google treats most top-level domains as generic, including .com and .org, and treats .eu and .asia as generic too even though they refer to regions. It also treats some country-code domains, such as .tv and .me, as generic because owners and users often see them that way. If a site on a generic domain wants to target a country, Google says to set that target explicitly using one of the documented methods.
What Google says it does not do
Several lines in the documentation are warnings against assumptions. Google crawls from different places but does not vary the crawler’s source for a single site to discover variations. It ignores locational meta tags such as geo.position and distribution, and geotargeting HTML attributes. It advises against adapting content by IP location, calling that analysis difficult and generally unreliable. The consequence is simple: if a variation is not declared through hreflang, alternate URLs, a country-specific domain or explicit links, you cannot assume Google will find it.
It also notes that geotargeting is not an exact science, so people will land on the wrong version. Visible links that let them choose a region or language are the documented answer.
Duplicate content across regional versions
Where similar or identical content in the same language sits on different URLs, for example example.de/ and example.com/de/ showing similar German content, Google says to pick a preferred version and use rel=”canonical” together with hreflang so the right URL is served. Our piece on how Google picks a canonical URL explains why that choice is a hint rather than an instruction.
Where this fits at Cultured Digital: international work
Cultured Digital is based in Sunderland and works UK-wide and internationally. Our international SEO guide lists what the work covers: hreflang and market structure, domain and folder strategy, localisation without duplication and multi-market migrations. Because a change of structure is a migration, the notes on redirect chains and site moves are the companion to this article. The service is part of our technical SEO work.