For a website, the web is only world-wide when it has been structured in such a way that Google will present the site in the search results of more than just one country and language. Part of SEO is, therefore, the theory of structuring websites targeting several countries and languages. This is called: geo targeting and language targeting. A lot of information on this subject can be found on the Internet, but nothing solid (that I could find) that brings it all together.
Below you find a very condensed but extensive design model for websites on how to target several geographic locations and languages. I based it on several recent publications of Google. All described variants of targeting are IP-independent, which means that for all solutions the website can be hosted anywhere. Also, there are no duplicate content problems for these solutions.
After explaining the variables first, the design model is presented. This is followed by several notes and additional remarks. I’ve referred to my sources of information, as much as possible. In the last paragraph it is explained how to respond to this article (via my seo forum).
I want to thank everybody who responded to my first article on this subject, earlier this year, on both this blog and on my forum: Erik-Jan, Pythagoras, Sint, Edwin, Reinier, Bas3pix, Reino and Rik. This gave me several new ideas.
Explanation used variables in design model
|#||The number of the variant discussed|
|Description||Short description of the variant|
|Eg (Geo/Lang)||An example described by Geographic location / Language. In the examples, especially The Netherlands and Belgium are used, since in both countries Dutch is spoken. Belgium’s second language is French.|
|TLD||The Top Level Domain (eg. .com, .eu, .nl, etc.). Country-neutral TLD’s are .com, .eu. .net, etc. World-wide TLD’s are .com, .net, etc.|
|GT||The Geographic Target, which is an input field in Google Webmaster Tools. This field is disabled for country-specific TLD’s.|
|IP||Requirements on the IP-address, which means: the server location. This column in the design model describes why none of the variants have limitations on the server location, since Google is able to determine the specific country without needing the IP-address.|
More info: server-location-cross-linking
Design model for websites targeting multiple countries and/or languages
Several countries, one similar language
G=Belgium, Netherlands/ L=Dutch
G=Belgium/ L=Dutch, French
For every country apply #1 (when country has one language) or #2 (when country has multiple languages)
be.domain.com/nl + be.domain.com/fr
nl.domain.com/be + fr.domain.com/be
In the opinion of Matt Cutts, the order of preference of the solutions of variant 3 is: 3a, 3b, 3c. Not because they differ very much on their scoring potential in Google, but because of simplicity of maintainance.
More info: subdomains-and-subdirectories
For the same reason (simplicity of maintainance), the order of preference of the solutions 4b and 4c would be: 4b, 4c. This because of the fact that solution 4c results in (much) more sites in Webmaster Tools, than 4b. After all, subdirs inherit the Geographic setting from the parent (sub)domain, which saves sites in case the parent (sub)domain represents a country. In solution 4c the (sub)domain represents a language. Above 4b and 4c, I prefer 4a, since country-specific TLD’s seem to be more valued by people searching from that country. In case of a lot of countries, this can be a quite complicated solution, though.
Where country-neutral or world-wide domains are used, the covering domain (www.domain.eu or www.domain.com) can be used as the overall page for all subdirs or subdomains. On that covering page, users can be asked to select geographic location and language.
Based on IP, Cookie or browser-settings, it is allowed by Google to redirect the user to a specific part of the website as long as Googlebot is considered as a normal user.
More info: how-google-defines-ip-delivery
For Googlebot to make sense out of the websitestructure with all its different languages and geographic targets, only link between pages within the same language and geographic region: minimize cross-linking.
More info: server-location-cross-linking-and-web
Try to obtain inbound links to the pages of the website from sites with the same language and geographic location as the concerning page.
The language metatag is not used by Google. Language of a page is determined by Google directly by examining the content of the page.
More info: answering-more-popular-picks-meta-tags
The same content in different languages is not considered as duplicate content.
From the fact that Google offers Geo Targeting via Webmaster Central it can be concluded that similar content in one language is also not considered as duplicate as long as the targeted regions are different.
Everybody is invited to respond to this article. In my seo forum I started a thread where this can be done. Thanks!