We recently had a bit of a head-scratcher with a new e-commerce client and problems with geo-targeting the correct site to the correct region. Basically, the correct site wasn’t targeting the correct region and in most cases “.com” seemed to rank vaguely well everywhere; with the .co.uk and .com.au sites nowhere in their respective regions.
The client is a small business with fairly low average sale values that ships to just these three English speaking countries referenced. Therefore, SEO director Peter Handley who was in charge of the project, talked through a number of solutions with the client, such as one-site-select-currency-options at payment stages or keeping the multiple TLDs but really changing up the content.
In each case, however, the client had tried various solutions before and nothing seemed to work or pose a sensible solution for them when considered against the size of their site and the size of their operation. Peter then suggested to them that we try implementing rel="alternate" hreflang="country" in XML sitemaps.
This post assumes a certain level of knowledge of when and how to implement hreflang for geo-targeting multiple sites. If this is completely new to you or you need a primer then check out our original case study as I’m going to whizz through the basics before getting to the kinks you need to be aware of.
What is rel="alternate" hreflang="x"?
It’s a way of telling Google that you have pages targeted for particular regions. If you have regional pages in a variety of languages, or pages in the same language that are targeting different regions that use that language, then Google can sometimes get confused, and needs a bit of help. Using rel="alternate" hreflang="x" you tell Google what pages you want to appear to which searchers using x languages, and with regional variations in language.
To implement you have two choices: either on a page by page basis, or using an XML sitemap.
<head> code was launched first in December 2011, but didn’t appear to have widespread take up, possibly because it’s hard work to implement as you have to map the pages for each location, then determine the HTML to add to each page. In our case we judged this to be too time-consuming so as to be inefficient and opted for the XML sitemaps option instead.
XML sitemaps for hreflang were released in May 2012 as an option, and since then although there seems to be a lot of chatter and frustration in implementing this on SEO forums we couldn’t find a single successful test case. It took a lot of research, test, trial and error.
Please understand that, even though XML sitemaps is the easier way to implement this solution, there is still a lot of fiddling around. However, only one file needs to be implemented on each website with this method, so once you’ve gone to the trouble of creating the map and extrapolating that into the sitemap correctly, it actually is simple from then on.
Example code for XML Sitemap can be found here.
As detailed in our original case study, this proved successful and solved the client problem. We now have an Australian site targetting .com.au and so on and so forth for the UK and U.S. markets.
It took almost exactly one week post-implementation for the regional sites to rank in the regional search engines correctly.
Hreflang isn’t about improving rankings, but ensuring regional sites are pointed correctly. Therefore the .com.au site appeared where previously the .com version had appeared in Google.com.au. In essence the correct site usurped the incorrect sites’ general place in the rankings of that engine.
Immediately, the traffic began to flow to the correct sites, reducing the paperwork and profit limiting headaches to the client, caused when customers purchased from the site foreign to their region.
Onto the Kinks...
The main problem we’ve had with this so far is that the site retains the snippet of the original site, which we hadn’t initially suspected. It seems clear from further feedback from the community that this has been talked about, but doesn’t appear to be visible in the documentation about how to set it up.
In fact, it seems that there’s been a lot of confusion and mistakes in implementing this, leading Google to deliver new canonical implementation instructions to the International Search Summit (ISS Munich).
Having seen this ourselves in a live case, we’re not sure that this is the correct path for Google to take. If for example Google rank a prominent .com site in all regions, despite this being specific to one region, and the titles and descriptions are marked-up on that regional basis to target appropriately, then why show those when they are swapping out the URL already!
The only thing that changes is the URL!
What we wanted to see:
From conversations on one of the Google forums, a Googler suggested that because the pages were substantially similar from site to site (there was localized/localised spelling variations, but not a great deal more variation), that they didn’t warrant having their own local titles, and should instead use the same title on all these pages where they are just about the same.
This doesn’t strike me as best for the user here. Any of us who have visited the U.S., UK and Australia can attest that what might be a resonating title for an American audience might be a total crock for an Australian audience. But that is the way it is at the moment and something that needs to be heeded when implementing this solution or proposing to clients.
Ideally we would like to see this change in the future, so that targeted titles and descriptions for these specific regional SERPs match the user query, not sticking to the original.
We’ve also been told by others of them having implemented this and nothing happened, so it doesn’t sound like it works perfectly in all cases just yet. A word of warning has also recently been sounded with using this in conjunction with canonicals.
Using hreflang is a new solution. In our case it was the one and only (almost) perfect solution for this client case.
We can imagine so many cases where this will be just the solution international e-commerce businesses are looking for. Particularly in Europe where we can have so many regional language variations even in one country, so this is likely to be a mainstay of international SEO targeting now and in the future.
Don’t let the lack of successful test cases put you off. Get testing and implementing.
We got some feedback from Googlers via Twitter during our implementation headache and they are interested to hear feedback from the community. Share your experience and let us know about any issues and kinks that you come across.