SEO multilingual website without geo-referencing under Google

Introduction

In this article, we will discuss important elements when promoting a website translated into a large number of languages and not having a specific geographic location, that is, we are interested in organic traffic in any language from anywhere in the world.

My experience is based on promoting a site for expats and travelers from all over the world, whose content pages are formed from a database. A site with millions of pages generated from a content database.

Multilingual and URL Formation

In my example, the site needed to be translated into 15 main languages, respectively, it was necessary to choose a URL construction scheme for each language.

There are four options for how this can be done:
')
1- Each language on a separate subdomain
Example: de.example.com, ru.example.com

Benefits - not identified.
Disadvantages - inconvenience of administration.

2- Each language on a separate domain (Example: example.de, example.ru)

Advantages - a higher CTR from the search due to the “native” user of the domain zone (Example: A German would rather click on a domain in the .de zone rather than on a .com).
Disadvantages - the difficulty of working with external links, extremely inconvenient administration.

3- For each language its own parameters in the URL (Example: example.com?leng=de, example.com?leng=ru)

Benefits - not identified.
Disadvantages - URLs with parameters are worse indexed and participate in the search.

4- Each language in a separate subdirectory (Example example.com/de/, example.com/ru/)

Benefits - not identified.
Disadvantages - not identified.

Even though the second option, unlike the others, has an advantage, I chose the fourth because of the simplicity of implementation and the absence of flaws.

Attribute hreflang - show Google in what language to show the page in the issue

In the work on the site I ran into a problem. Google and Yandex now use the hreflang attribute to determine the language and country for which to display this version of the page.

Example:

<link rel="alternate" href="http://example.com/ru-au" hreflang="ru-au" />

users who request in Russian from Australia will be shown a version of example.com/ru-au

 <link rel="alternate" href="http://example.com/ru-gb" hreflang="ru-gb" />

users who request in Russian from the UK will be shown a version of example.com/ru-gb

 <link rel="alternate" href="http://example.com/de-au" hreflang="de-au" />

Version.com/de-au will be shown to users requesting German from the UK

 <link rel="alternate" href="http://example.com/de-gb" hreflang="de-gb" />

Version.com/de-gb will be shown to users requesting German from the UK

The problem with my site was that any language version of the page could be requested from any geographic location. For example, a Russian-speaking person may request a page from Russia, Australia, Germany and any other place in the world.

Consequently, I would have to write attribute sheets that take into account all combinations of languages and countries for each page.

The solution is to set the attribute parameters hreflang without specifying geography, that is, we show Google only the language versions of the page, without reference to the region.

Example:

 <link rel="alternate" href="http://example.com/ru/" hreflang="ru" />

For users who request in Russian from anywhere in the world, version.com will be shown.

 <link rel="alternate" href="http://example.com/de/" hreflang="de" />

For users who request German from anywhere in the world, version.com will be shown.

GeoShape - microdata geographic location of objects

As it turned out, hreflang only partially solved the problem. Since the site generates pages from the database (and its content is not translated into different languages), the situation is that I have millions of duplicate pages on the main content, differing only in the name of the requested country in the title, and the rest of the content is often absolutely identical, or slightly different. As an example, the analogues of Russian medicines in France, Italy and in general in the European Union will be practically the same. It turns out that the pages of Italy and France that are different for users will be doubles in the eyes of Google.

The challenge now was to show Google that pages with the same or similar content were intended for different countries. As we remember, hreflang cannot be used for this task, since it will be either hundreds of attribute lines, or pages with different languages will not be available from all countries.

I came to the aid of micromarking. Faced with the problem, I began to analyze sites like mine, and on booking.com I came across a geoShape structured data tag. They use it to indicate the location of hotels.

After checking the validity of the code when using GeoShape in the markup of drugs (the site is about in the article about analogs of drugs), I installed everything on the site.

 <div class="items-preparation full_match" itemprop="relatedDrug" itemscope itemtype = "http://schema.org/Drug"> <span itemprop="subjectOf" itemscope itemtype="http://schema.org/GeoShape"> <meta itemprop="addressCountry" content="AU"> </span>...

Now Google sees the difference in geography between pages destined for different countries, but with similar content, and users can get their language version of the page from anywhere in the world.

Sitemap.xml - Improving indexing

Everything is pretty simple here, but there are also nuances.

Simple

Mark up all the URLs of the site with the language versions in sitemap.xml using the hreflang attribute in accordance with the recommendations of Google .

 <url> <loc>http://www.example.com/english/</loc> <xhtml:link rel="alternate" hreflang="de" href="http://www.example.com/deutsch/" /> <xhtml:link rel="alternate" hreflang="de-ch" href="http://www.example.com/schweiz-deutsch/" /> <xhtml:link rel="alternate" hreflang="en" href="http://www.example.com/english/" /> </url>

It turns out that hreflang is duplicated in both the pages and sitemap.xml.

Nuances

When translated into many languages, the number of pages is directly proportional, so it’s worth remembering that the maximum number of links in one sitemap.xml file is 50,000, and the file itself should weigh no more than 10 MB.
If there are a large number of pages, it is recommended to split the sitemap files by language and site structure.

References:

Source: https://habr.com/ru/post/352082/

All Articles