Digital Growth Day: September 18, 2014
HOW TO GO GLOBAL WITH INTERNATIONAL SEO
How to implement href lang tags and avoid international duplication.
Michal Magdziarz, DeepCrawl
2. Agenda / Table of Contents
SERP Influencers / Users
• What influences SERPs?
Physical Geolocation / Language / Language settings / Google version
• “wikipedia” search experiment
Website Setup
How to set up your website?
• ccTLDs vs. TLD subdomain vs. TLD subfolder
• Geographic Targeting
• Language/Country specification (hreflang)
• Language Specification
Workshop
Example Implementation
4. SERPs > Combinations of SERP Influencers
Even considering just two languages and two different geolocations and google
versions we have 2*2*2*2 = 24 = 16 combinations
For 20 languages 160K!
Let’s test how it works in real world and how the factors affect the search results. We
tested the 3 following factors:
• Physical location of the searcher (UK vs PL)
• Country specific Google version (google.co.uk vs google.pl)
• Language Settings (hl=en vs hl=pl)
The three influencing factors gave us eight combinations (4 for each physical location):
• Google PL / Language PL
• Google PL / Language EN
• Google UK / Language PL
• Google UK / Language EN
5. SERPs > Google UK / Language EN – “wikipedia” search
http://www.google.co.uk/#hl=en
Search completed in Poland using google.co.uk with langiuage settings set to english. Despite being in Poland the results are english
(international + co.uk). The physical location of a searcher didn’t seem to have had any impact on the search results.
6. SERPs > Google UK / Language PL – “wikipedia” search
http://www.google.co.uk/#hl=pl
Regardless of the physical location of the searcher and despite using google.co.uk polish version of wikipedia get a higher ranking than the
international and english version.
7. SERPs > Google PL / Language EN – “wikipedia” search
http://www.google.pl/#hl=en
Triple listing combining international, english and polish versions of Wikipedia. Again searches performed in Poland and in United Kingdom
are identical. The physical location of the searcher didn’t have any impact on the results.
8. SERPs > Google PL / Language PL – “wikipedia” search
http://www.google.pl/#hl=pl
As expected a search in Poland using google.pl with the browser language settings set to polish returns polish version of wikipedia. Again
there is a very little difference between a search performed in Poland comparing to the one performed in United Kingdom.
9. SERPs > Conclusions
• Language setup is more important than Physical location
• Physical location doesn’t have much impact on SERPs
• Google country specific version is important but not as much as
Language setup
10. Website Setup > ccTLDs vs. subdomain vs. subfolder
Country Code Top Level Domains (ccTLDs)
Each country can adopt its own ccTLDs (country specific root domain) e.g. www.domain.de
Example implementations:
eBay - http://www.ebay.com, http://www.ebay.de, http://www.ebay.es
Amazon - http://www.amazon.com, http://www.amazon.co.uk, http://www.amazon.de
TLD Subdomains
Each country can be set up on a subdomain e.g. de. domain.com
Example implementation:
Wikipedia - http://www.wikipedia.org, http://de.wikipedia.org, http://es.wikipedia.org
TLD Subfolders
Each country can be set up on a subfolder e.g. www. domain.com/de.
Subfolders can also be used in conjunction with ccTLDs and subdomains
Example implementations:
Microsoft - www.microsoft.com/en-gb/, www.microsoft.com/de-de/ www.microsoft.com/fr-fr/
IBM - http://www.ibm.com/us/en/, http://www.ibm.com/de/de/, http://www.ibm.com/es/es/
11. Website Setup > Geographic Targeting
There are a two ways in which the Geographic Targeting can be assigned:
• Country specific TLDs (domain.co.uk, domain.fr, etc) have the geographic target automatically
assigned and cannot be changed.
• Generic TLDs such as domain.com, domain.net etc do not have a geographic target assigned
and this can be specified in Google Webmaster Tools > Configuration > Settings > Geographic
Target.
NOTE: Independent geographic target settings can be assigned for Subfolders and Subdomains
of one Generic TLD. See examples below:
• TLD Subdomain: de.domain.com can be set to target Germany once fr.domain.com to
France
• TLD Subfolder : www.domain.com/de can be set to target Germany once
www.domain.com/fr to France
12. Website Setup > Language/Country specification (hreflang)
Hreflang tags are a powerful way to tell Google which international version of your site is most
relevant to a user, based on their language and location.
As well as improving your user experience (because it helps users land on the most relevant
version of your site), hreflang also helps to aggregate authority signals around a single version of a
page and are a great way to manage duplicate content issues caused by international websites.
How do I know if I need hreflang?
Google recommends using hreflang if you have:
• Similar regional variations of the same site in the same language (eg. currency variations for
GBP and USD)
• The same content in different languages
Does this solve duplication issue?
Yes, two identical websites targeting two different audiences are not duplicates e.g. US vs UK
13. Website Setup > Writing hreflang: three essential elements
All hreflang tags contain three elements:
• The rel=”alternate” attribute
• A language code on its own, or a language code plus a country code
• One URL for the most appropriate alternative for the specified language and country
For example, this hreflang tag will notify Google about an alternative version of the page suitable
for English-speaking users in Great Britain (en-gb).
rel=”alternate” hreflang="en-gb" href=“http://www.example.co.uk/”
The “hreflang=” attribute is made from a 2-letter ISO 639 language code and the appropriate ISO
3166 geography code separated by a dash. For example:
de-at: German, Austria
de-de: German, Germany
en-us: English, United States
14. Website Setup > Writing hreflang: three essential elements
It can be a good idea to have generic language sites for each of your main languages, to account
for the vast majority of your users where this isn’t a specific language + country available. A
language code on its own will be used as the default where there isn’t a relevant language +
country code.
The following tags would account for all English and German-speaking users, no matter which
country they are in:
hreflang="en" href="http://www.example.com/"
hreflang="de" href=“http://www.example.de/”
You cannot define country only without specifying a language e.g.
hreflang=“gb" href="http://www.example.com/" >> INVALID
hreflang="de" href=“http://www.example.de/” >> this means German language not
country
The order has to be kept! Reversed order can lead to invalid specification e.g.
hreflang=“gb-en" href="http://www.example.com/" >> INVALID
hreflang=“fr-de" href="http://www.example.com/" >> otherwise ambigouous
15. Website Setup > Where to add your hreflang tags
In your sitemaps
The best place to add hreflang is in your sitemap as including them in the headers or on the page
adds weight to every single page request.
The following example will inform Google about the English version from the German version of
the website:
<url>
<loc>http://www.example.com/deutsch/</loc>
<xhtml:link
rel="alternate"
hreflang="en"
href="http://www.example.com/english/"
/>
<xhtml:link
rel="alternate"
hreflang="de"
href="http://www.example.com/deutsch/"
/>
</url>
This method would need to be repeated in full for every page on the site and for all the
international websites.
16. Website Setup > Where to add your hreflang tags
In your headers and HTML
Hreflang tags can also be added to the HTTP header:
Link: <http://www.example.com/english/>; rel="alternate"; hreflang="en"
Link: <http://www.example.com/deutsch/>; rel="alternate"; hreflang="de"
Or in the <head> tag in the HTML:
<link rel="alternate" hreflang="en" href="http://www.example.com/english/" />
<link rel="alternate" hreflang="de" href="http://www.example.com/deutsch/" />
17. Website Setup > Testing hreflang with DeepCrawl
1. Pages with hreflang
All pages found with at least one hreflang tags found in the sitemaps, header or page tags.
In your DeepCrawl report, go to: Validation > Pages with hreflang Tags
The report shows a detailed view of each page’s alternative URLs for every language/country
variation used on the site. This view will also show you if a tag has been implemented
inconsistently - there should be no gaps in this table:
18. Website Setup > Testing hreflang with DeepCrawl
2. Pages without hreflang
You can quickly identify all pages missing hreflang tags within the sitemaps, header, or page tags.
Go to: Validation > Pages without hreflang Tags
3. Pages with inconsistent hreflang
If there more than one alternative URL specified for a country + language combination, they will
be flagged as inconsistent in this report.
Go to: Validation > Inconsistent hreflang Tags
19. Example Implementation (workshop)
Group 1 -
https://docs.google.com/spreadsheet/ccc?key=0ApzW7F75d-
1SdHRiR19PWjFpZVVydzI2OTBoVXNmUkE&usp=sharing
Group 2 -
https://docs.google.com/spreadsheet/ccc?key=0ApzW7F75d-
1SdHRiR19PWjFpZVVydzI2OTBoVXNmUkE&usp=drive_web#gid=10
Group 3 -
https://docs.google.com/spreadsheet/ccc?key=0ApzW7F75d-
1SdHRiR19PWjFpZVVydzI2OTBoVXNmUkE&usp=drive_web#gid=9
Group 4 -
https://docs.google.com/spreadsheet/ccc?key=0ApzW7F75d-
1SdHRiR19PWjFpZVVydzI2OTBoVXNmUkE&usp=drive_web#gid=8