What does Google say about multiple hreflang values on one URL?

Google's hreflang documentation leaves a lot of room for interpretation. One of the open questions for a long time was whether a single URL may carry several hreflang values. This article documents what Google has said about it and what the variants look like in practice.

The background

At SMX Munich in March 2016 there was a session on hreflang. Eoghan Henn showed how to implement hreflang correctly, how hreflang and canonical work together, which domain strategy suits international sites and when x-default makes sense. In the process, he expressed doubts about whether it makes sense to give a URL two hreflang values.

In the Q&A, one attendee said that John Mueller from Google had once declared such a double assignment permissible. John Mueller was in the audience and confirmed that Google accepts this kind of implementation.

x-default and language code on the same URL

In the example, the English page /en/about/ is also the default version for all users without a matching language. It therefore gets two values, “x-default” and “en”. There are also separate versions for the UK and Ireland.

As with any hreflang set, this block is identical on all four URLs involved. The hreflang guide explains what x-default means and when it fits.

One version for several countries

The confirmation immediately raised the next question. Does this also apply to other cases? In the second example, a website has an English version for the USA and an English version for several European countries. The US version gets “en-us”. The EU version gets its own value for every country it targets.

Before that conversation, Eoghan would not have recommended this implementation. When asked, John Mueller confirmed that Google accepts multiple values in this case too. On March 2, 2017 he confirmed it again on Twitter. In essence, he added that a pure language code or x-default is also an option and that everything can be combined.

Eoghan then used the country variant on several websites, and in his experience it worked well every time. There are no measurements on this.

Which variant fits when?

If there is only one version for all countries of a language, the pure language code is enough. The country variant is worthwhile if, besides the group, there are other versions in the same language that Google should tell apart.

Codes that match the language of the page

In one real case, a shop had tagged its English EU pages with codes such as “en-fr”. English-speaking searchers in France found the pages, French-speaking ones did not. Only a French version with “fr-fr” solved this. The language part of the code should therefore match the language of the page. There is no official way for pages with two languages on one URL.

What has changed since?

In 2016, the recommendation was to leave country targeting in Search Console empty when using several country codes on one URL. Google no longer supports this setting. Google's current documentation mentions multiple values for one URL neither as allowed nor as prohibited. It describes one link per page for each variant, including the page itself. The statements from 2016 and 2017 therefore remain the only ones that refer directly to this case. Newly documented is that reserved codes such as “EU” or “UK” have no effect. The path /en-eu/ is unproblematic, whereas an hreflang value “en-eu” would have no effect.

Common mistakes

• The country code “uk” instead of “gb” for the United Kingdom. Google ignores it.
• A value such as “en-eu” for a group of countries. Regions without an ISO country code have no effect.
• The set only appears on the EU version. Every URL involved needs the same block, otherwise return links are missing.
• Country codes for countries whose language the page does not speak.
• x-default on every language version instead of on a single URL.

Frequently asked questions