Dec 132011

Advanced International SEO with rel="alternate" hreflang="x"

Have you ever wanted a site to appear in country-specific organic search results, without going through the trouble of creating full-blown country-specific sites? Well, you're in luck, but be ready to act like a guinea pig!

One of the most rewarding aspects of search engine optimization, is continually working in unique, and sometimes, strange situations. We all have war stories to share. Technical SEO can be especially hairy, made even more so when verging into international scenarios.

Here's an example of one such off-the-wall problem, with an uncommon solution. We'd love to know from others who have handled these types of issues. Not surprisingly, rel="alternate" hreflang="x" is not widely adopted. This is fringe SEO work.

The Case

A client came to us with an interesting predicament. They wanted to expand with an international presence (they're a well known brand), but didn't want to create separate international sites. They wanted the primary domain to rank appropriately in each region. After some discussion, they agreed to consider using sub-domains with region-specific abbreviations. The catch was, the sub-domains would have exactly the same content (in the same language, English) as the primary site. This would create a whole mess of duplication.

Now, put aside for a moment asking the question why a company would want to do this. Companies do lots of things, for lots of reasons, and all of them are valid (even if they're not valid SEO decisions). That's why they have us to figure the rest out.

But how can we do this? How can we show region-specific subdomains to the right users, while avoiding mass duplication problems?

Doing the Impossible? Wait, Google Has a Tool for That.

The situation seems impossible: get a site to rank in a specific region, even though it's a duplicate of a canonical site. Luckily for us, Google had already figured this one out.

An exmpale of using rel=alternate and hreflang=x in source codeAn example of the highly technical, slightly arcane, rel="alternate" hreflang="x" in the wild. Note: example only. is not the client referenced in this article.

The problem is that this solution is technical and unlikely to to be implemented. In fact, when I asked about 4,000 followers on Twitter if anyone had experience using it, no one responded with a "yes."

Bueller? Not too many folks out there using this one.Bueller? Not much action with this one.

One important qualification: this technique is for language-specific templates, where the primary content is not translated. This is quite common in situations where a company has regional sites, but hasn't invested resources into localizing all the content (which can be a huge project).

SEO for International Sites with rel="alternate" hreflang="x"

First, a giant caveat: we haven't actually done this, so it's all theoretical. It's not a particularly easy thing to test, and there aren't many situations where clients are ready to implement something so unproven. That said, here are the steps that we will surely be testing and recommending when we get the chance. As this is still on the fringe of what anyone's really done, I'd love to get your experiences in the comments.

Problem: wants to appear in regional search engines. They use international sub-domains and language-specific templates, with the vast majority of content not translated and appearing as a complete duplicate of the default English site. Only navigation links have been localized.

Proposed Solution:

  1. creates sub-domains for each of its language sites, e.g.,,, etc. Please note that folders could also work (e.g.
    Requirement: create localized versions of site templates, so navigation links are in the appropriate language. Primary content can remain in English.
  2. Each international sub-domain is separately authenticated in GWT and geo-targeted to the appropriate country. (Note: this may be unnecessary, see below.)
  3. Each international sub-domain is given a rel canonical tag to the canonical USA site. (Note: this may make the GWT step unnecessary. Would need to test this prior to implementation.)
  4. Each international sub-domain is then given rel=alternate hreflang="x" tags in the HTML source code. For each "x" give the appropriate language URL. For example, would have these tags (and possibly more, depending on how many language templates existed):
    <link rel=”alternate” hreflang="es" href="" />
    <link rel=”alternate” hreflang="fr" href="" />

What this is supposed to accomplish: link metrics are consolidated to the canonical (www) version. However, region-specific versions of the sites will appear in search results for that region. This way we accomplish the goal has of targeting specific regions in search, while avoiding potential filters or penalties associated with duplicate content. Plus, they get the benefits of regional targeting without doing much heavy lifting in terms of development costs.

Obviously, Bing doesn't support this approach, so we'd need another strategy there (possibly just rolling up the language versions to with their Web Tools console).

Thoughts out there? Are we crazy?


Google has announced support for hreflang within XML sitemaps, which is a superior method to implement this annotation. Hurray! Go here for the skinny:


41 Responses to "Advanced International SEO with rel="alternate" hreflang="x""
Hubert says:
Hello Adam, In fact, I replied to your tweet yesterday evening with an example : We use it for this website to focus on various English version of the website : - Default for the US with USD currency - Another one for UK in GBP - A default one in Europe for EUR This will help the website to avoid duplicate content and help indexation of other languages. Do you think it's a good example of this new tag?
Hi Adam, When you said this was fringe SEO you weren't kidding... Would you then continue the form for different english speaking regions with EN-GB, EN-AU, EN-CA etc? It'd be interesting to see how google treats the subdomains in the SERPs. If it see's them as canonical versions of the USA domain, would it not just show the main site? Just my thoughts... Andrew
Hubert, a great example. I missed your tweet - just went back and saw it. Thanks for the great example. How has the tag been working for you? Are results as expected?
Andrew, yes, the hreflang attribute can define the region as well as the language, for example: • en-GB: English content for GB users • en-AU: English content for AU users Pretty cool stuff. To answer your second question, the sub-domains will only (should only) appear in the appropriate regional SERP. Use of rel=canonical alongside hreflang (as recommended at the bottom of this page: ) consolidates scoring signals.
Hey Adam, Great discussion. I have been pondering much the same and will be testing soon. It's been a problem for a long-time. I have a client who already have just that sub domain (.uk, .us) set-up in place (with English language) content targeting various regions UK, US etc. We went the route of making sure the content was not duplicated, but unique (to varying degrees depending on page importance). This seemed to work well, although I still feel there is something lost a little (in terms authority/trust, whatever) in the sub domain route rather than sub directories. (It will be interesting to see if the new tag will help a little here a little seeing as they recommend the sub domain approach in the examples.) The part I find quite interesting is the use of the canonical alongside the rel=”alternate” hreflang=”x” tag. Showing the canonical in the SERP but displaying the regional content. To what degree will the signals be consolidated for scoring and ranking regional content? By that I mean, you could be very well known and popular in one country (with great links etc), but unknown and less relevant in another :-) Cheers. Dan
Really good points, Dan, especially on regional quality signals. It will be interesting to see how this works in the field. I do agree with you on sub-domains vs directories -- seems in this day and age keeping things on the primary domain buckets the equity together more efficiently. That strategy does flip-flop a lot though, depending on the latest at Google! Thanks again for stopping by.
Cameron says:
Adam, re english speaking countries: if the content is exactly the same, as you put it, I don't understand why you'd setup an extra subdomain/folder for them. I fully understand the reasoning when/if the content changes or they're non-english speaking countries. Are you expecting a boost in the regional SERPs by doing this? We've faced some similar problems in our group. But our issue was that mirrored sites were setup for each country. A bit of a dup content problem :p. We looked at a lot of data and ultimately decided that all english speaking countries would be best off consolidated into the main .com version. We setup subfolders for each of these countries with regional specific hub pages. Any content that would be considered global would live off the root domain. If the content was regional, it would live in that's region's specific folder. But any of this content can be linked from anywhere in the site. I.e. we can link users to global content from their regional hubs. This was all before G announced the markup for multi-lingual content. Happy to discuss in more detail if you want to shoot me an email.
Hubert says:
Currently, I do not see any changes for the different versions of the website. Maybe, it will be taken into account later and it will impact positioning but evidence yet.
Cameron, we've not yet needed to recommend using this for regional languages (en-GB, etc), but it's certainly a valid option. Not significantly different, really, than different languages. Remember, this is for sites that have substantially the same content but localized navigations, for example, or language-specific profile pages, for another example. Hubert, thanks. Hopefully you start to see some pickup.
Simon says:
Hi Adam, Do you think it's possibly to apply [" hreflang="de-DE" href=] for folders as well? For instance: and ?
Simon, yes, folders can also work for this (as opposed to sub-domains). Cheers!
Simon says:
Thanks Adam! Will get to work on this now. Simon
Erik says:
We are trying to do the exact same thing. our website is targeted to the Australian market. However we also sell to New Zealand and the UK. so we have the domains and as we don't own (and the owner does not want to sell) this poses issues for us. I'm trying to get above solution to work however I'm struggling with automating the link rel in Joomla (1.6). does anyone know of an extension that works?
Oliver says:
Will this work if the text are on differing domains? i.e, and so on.. Oliver.
Oliver: yes, this works cross-domain. Just be really careful to list all the URLs that belong to the set on every page, so no one can "hijack" your set of pages and try to add theirs to it.
Elin says:
Really interesting Adam. Trying to create websites for specific markets/countries that rank highly is a fascinating conundrum. I'd never come across a suggestion like this before. Certainly something to consider when we start to try and develop fuller presences in new countries.
Serbay says:
Still you havent test it yet right? It seems really strange but i think if you use canonical to sub-domains like will not be crawled and indexed again. Maybe if href lang part is above the rel canonical tag it can be indexed but if google sees rel canonical first i dont think so. But it is really strange example i need to test it out.
Something Google hasn't described well is how this affects regional link building. If I have: Using this method, I would assume it would consolidate all link signals to, so if you are entering new markets, this is a great way to transport the equity in your .com domain to your But is there any points in then building regional links in the UK to improve your UK rank? The PressKing is a good example of how this should be set up in the wild. I did a few searches for keywords relevant to them across regions and it seems to be working for them. Thanks Kieran
Kieran, hreflang is specifically for region targeting, not consolidating link equity. That's what rel canonical is for. Using hreflang w/out rel canonical means you still need to focus on regional links and nothing really changes w/an international SEO strategy. Using rel canonical does consolidate link equity (and can be used in tandem with the regional targeting benefits of hreflang), but should only be used if the sites or pages are truly duplicates (such as with subtle template localizations, currencies, etc). Rel canonical should never be used for independent, international sites, as they will disappear from the index entirely.
Hi Adam I am not sure you are following my point. If I set up: and implement the following tag (on the UK site): Then I now have my link equity for being consolidated with the US site. For international companies who are running global SEO strategies, they may have a UK agency working on link building to improve the performance of their UK site (with the US having their own strategy and own agency). The question I have is around that, is there any point in having a regional SEO strategy for a UK site that is a duplicate version of the .com site (it may have some content that differs, but the core part of the site in terms of product pages etc, will be duplicate versions). This was a requirement in the past for global sites who either separated their regions via sub folders or sub domains as building regional links was a key factor in having those regional sites rank in their local version of google e.g. Thanks Kieran
Ah, for some reason my example code was taken out :) you can guess what it was. basically the use of the canonical tag + rel=”alternate” hreflang=”x” on the UK site.
Kieran - if you're using rel canonical on the UK version of the site then you wouldn't want to build links into it. Rel canonical = soft 301 and that puppy is GONE from the index. But you don't want to use rel canonical on a truly international UK site, because those often are NOT dupes. International sites (that are localized, translated) are not duplicate content. Int'l sites that are exactly the same besides the currency (for example) - yes those are dupes. This is a pretty complicated topic - sometimes easier to talk through. If you're up at Advanced in June give me a shout and we can meet about it.
Hey Adam, Thanks for the reply. Yes, probably easier to talk through in person. My point was Google hasn't given clear guidelines around the use of this in my mind (in terms of affects on link building). There was a good conversation on Google+ with Pierre from Google on this topic, but their answers on international link building where lacking in details. One thing I am not clear on is your first sentence. If you use canonical tag and the alternate tag for the UK site, that site is not gone from the index. Results for should show up results from the UK site if implemented correctly. That is the whole point of that tag and that is how pressking have it implemented in UK and it's working for them (I checked against a number of keywords). I am at SMX advanced in London so maybe we can have a coffee : Thanks Kieran
Kieran, Actually when rel canonical and hreflang are both present, URLs w/ rel canonical ARE pulled from the index. hreflang simply allows the display URL to fire at search time. The title and snippet text are taken from the rel canonical target. You need to be very careful here and understand the functions of these annotations, both separately and in tandem. I've spoken in detail about this w/ Google's team. This is how it works and why it's so important to watch rel canonical carefully, because it can and will yank an international site from the correct regional index. No one wants their US title and snippet text showing for their UK site. The URL is only 1 element of a successful search presentation, and in the case of hreflang, the URL is simply a display element fired at search time. Good luck!
Thanks for all the input Adam. Appreciate your comments.
Jesse says:
We're actually going to be trying an implementation for International Beer Day this year -- we're producing a dozen or so translations of the primary site content, and want to use rel="alternate" to consolidate and properly distribute them. We'll be going the folder route rather than subdomains, but I'd be happy to report back if that'd be useful!
Seeden says:
We use hreflang/canonical at , because we have for each country a different content. In the SERP we have bad results. Do not understand why... :(
Hi Adam, A few questions on the following comment you made in response to Kieran: "Actually when rel canonical and hreflang are both present, URLs w/ rel canonical ARE pulled from the index. hreflang simply allows the display URL to fire at search time. The title and snippet text are taken from the rel canonical target." I understand that having both hreflang and rel canonical can cause problems but do you think the order in which they appear in the source code matters? Maybe in your tests the hreflang was appearing below the rel canonical or have you tested it either way? Whether hreflang passes any link equity or not is an interesting topic but there are different cases e.g. subdomains, subfolders and different domains. If no link equity is being passed, probably having subfolders for the regional sites would be more beneficial in terms of having a bigger impact on rankings.
@Modesto the order of hreflang and rel canonical should not matter. To be clear: hreflang and rel canonical can be used in tandem without problems, in cases where the content is truly duplicative and minor things like currency are all that's different. But in cases where there are international sites which stand on their own rel canonical should not be used.
Hi Adam, really helpful Post! But i still need some more advice. We will use the hreflang for the following Site Structure: Each country as listed above has a different set and combination of services In Addition to those country specific pages in english there´s another worlwide english-Site that contains a list of all services.The worldwide site should only rank, if a countryspecific site doesnt offer the service. Do we need to put reflang attributes to that non country specific page? I´m looking forward to your reply. Beste Regards, Chris
Vladi says:
In theory, for a site that adopts the same strategy but actually takes the time to translate each page would you say implementing this rel="alternate" hreflang="X" is still required?
Sam says:
Check out this tool developed by the agency I work for that will definitely help with this International Sitemaps:
@Adam ok I understand that if canonical and hrelang both are present than URLs w/ rel canonical ARE pulled from the index. hreflang simply allows the display URL to fire at search time. So if my website is targeting USA and targeting UK Will subdomain will Rank in SERP for my keywords in UK?or i have to built backlinks to from UK region.
Sorry Sachin I don't understand your question, but it sounds like you've got a correct understanding of how rel canonical and hreflang work together.
Sachin Bisaani says:
@Adams Yes I learned from your post, Sorry I'm weak I'm English, My question is ranking of keywords
Steven Jonker says:
Hi Adam, great post but I still don't quite understand. Is the purpose of this to rank better in local serp's? If for example, company ABC who has one dot com website but has branches in 50+ countries decided they want to rank better in local serp's would they need to set up a duplicate for each country (only changing contact details etc) and use the method described above to target users with the country specific website? Looking forward to your reply. Cheers, Steve


Check out what others are saying...
[...] Advanced International SEO with rel=”alternate” hreflang=”x”, [...]
[...] hreflang=”es” href=”” />Richard Hearne, spotted a comment by Adam Audette on an article which talks about the usage and implementation of the language [...]
[...] International SEO with Rel=”Alternate” Hreflang=”X” [...]
[...] International SEO with Rel=”Alternate” Hreflang=”X” [...]

Leave A Comment