On 5th December last year Google launched a new tag used to mark-up translated content. Since its introduction however, there has been a great deal of confusion around what it does exactly and how it's used. Many of the big multinational players still haven't implemented the new mark-up because a) not many know about it and b) they don't understand it.
In fact, the International Search Summit held just last week in London revealed that a huge number of international web owners or webmasters are not using the tag correctly or not implementing it correctly in tandem with the canonical tags. It was stated that Google believes approx. 25% of Webmaster Central geo-targeting settings are incorrect.
So, before we walk through it, refer to this diagram to help get your head around it all!
The new mark-up essentially involves two parts:
the canonical tag <link rel=”canonical” href=(URL)/> and
the href lang tag <link rel= “alternate” hreflang=”(language code)” href=(URL)/>.
The Canonical Tag
Firstly, the canonical tag is, as we already know, a means of saying to Google: “This is the authoritative version of a page I'd like to present.”
To illustrate its use for multinational cases, let's say the main language of our site is US, e.g. en.example.com. If this page were to be auto-translated into French for fr.example.com, Google would consider this to be a duplicate version of our main English page, hence the need for a canonical tag. Here's Matt Cutts explaining this stance in a Webmaster video.
Now, those of you with your SEO hats on will be saying, “But won't this canonical tag pass all the value of the French page onto the main homepage?” Well, the answer is no, not when used in tandem with hreflang tags. The new change means that instead of pointing all the link value back to your main English subdomain, the total link value is allocated to them both in their own country versions of Google.
In the diagram above we are assuming that the two English versions (US + UK) have been uniquely written; they are not copies and contain significantly different content. In this case, there is obviously no need for a canonical tag, as we want each sub-domain to stand alone with its own link value.
However, if they were not unique pages, and contained much the same content, Google has now also stated that it supports the use of the canonical tag for distinguishing language variations. So, if you have a English page intended for British users (e.g. en-gb.example.com) and an English page intended for US users (en.example.com) which are very similar, you can now implement a canonical tag on each of these pages and, again, the total link value is allocated to them both as long as you are also using hreflang tags.
The Hreflang Tag
The hreflang tag is used to improve the targeting of multilingual websites. It helps differentiate between intended languages and countries. So while the canonical tag tells Google which version of a URL is the authoritative one, the hreflang tag states: “This is an alternate version of this URL, in language X, intended for audience Y.”
To illustrate this tag, let's use our en.example.com and its two auto-translated versions fr.example.com and ru.example.com. In addition to the canonical tags pointing back to the lead en.example.com, we'd have the following href lang tags in the head of every URL.
<link rel=“alternate” hreflang="EN-GB" href="en-gb.example.com"/>
<link rel=“alternate” hreflang="FR-FR" href="fr.example.com"/>
<link rel=“alternate” hreflang="RU-RU" href="ru.example.com"/>
The Hreflang in Action
Google itself has implemented the new mark-up in its Google Play and Google Help Centres. This first German listing for the Google Authenticator app shows the new mark-up in action. As you can see, the process is a fiddly one, as it requires a lot of extra code to be written on all pages concerned. That said, it's not a major technical implementation and the benefits for multinational sites are significant.
Let us know your thoughts on the new multilingual mark-up.