Yandex as a search engine was created 16 years ago, back in 1997, and became a brand in 2000. A lot has changed during the years, both for users and for SEOs. Let’s have a look at the evolution of Yandex search algorithm.
There are no records or announcements about Yandex ranking formula dated earlier than 2007. 2007 was the year when Yandex started actively working with their search algorithm and communicate the changes to the community. Ever since, every new ranking formula was named after a city in Russia, and the logic behind naming them is taken from an old Russian game called “cities”: each newly introduced formula’s name starts from the last letter of the previous formula’s name:
Magadan -> Nakhodka -> Arzamas -> Snezhinsk -> Konakovo -> Obninsk -> Krasnodar -> Reykjavik -> Kaliningrad. On the map this journey would look like this:
Let’s look a what this journey looked like for Russian SEO.
Magadan (April 2008)
The search algorithm Magadan was a revolution. It had at least twice as many ranking factors as the previous one, where uniqueness of site content received quite an important role.
Yandex learned to understand transliteration (spelling Russian words with Latin characters – for instance, in URLs), commonly used abbreviations, and rank documents based on this knowledge.
Another interesting feature was that Yandex started to better understand text and to rank relevant documents even if they did not contain exact keyword combinations.
Magadan was also the algorithm that started indexing and ranking websites in English.
Nakhodka (September 2008)
Nakhodka came half a year later. At that point Yandex learned to understand the meaning of stop-words (prepositions, conjunctions etc.) and use them for ranking documents.
Thesaurus of the search engine was expanded, which lead to to better understanding and ranking of synonyms and spelling variations.
Arzamas (April 2009)
Arzamas was the algorithm that brought real local search to Yandex. It was based on classification of all search queries into two types: geo-dependant and geo-indepenant. For geo-depenant queries the local businesses received priority in local SERPs (e.g. a Novosibirsk taxi company site would rank better in Novosibirsk than a site of a taxi company from Moscow).
Additionally, Arzamas was the first ranking algorithm that went hard on pages with pop-up and pop-under advertising.
Snezhinsk (November 2009)
In Snezhinks Yandex introduced several thousands new ranking factors, which provoked somewhat of a panic in the Russian SEO community. The main changes I noted were:
– Trusted sites (brands) got generally higher positions in search results
– Unique content because a must, and sites with duplicate content got punished
– Old links became more valuable than newly acquired ones
– New type of penalty for low quality sites – AGS – was introduced
All these changes made it more difficult and time-consuming for new sites to get first page of Yandex search results.
From a technology perspective, MatrixNet (self-learning engine) was launched.
All these changes took Yandex SEO to a new level, where quality of SEO work became more important than quantity of links.
Konakovo (December 2009)
Local search algorithm, introduced initially for 19 regions in Russia with the launch of Arzamas, was extended to 1250 cities throughout the country.
Obninsk (September 2010)
The main difference of that Obninsk brought to Yandex SEO was devaluing a large amount of “SEO links” (paid links).
Other changes in the algorithm included changes to ranking formula for geo-independant search queries and better way of identifying the initial source of content on case of duplication.
Krasnodar (December 2010)
Krasnodar algorithm incorporated another intelligent technology – Spectrum – the goal of which was to understand user’s search intent and serve better search results for words with more than one meaning or search queries that are too generic.
Reykjavik (August 2011)
Reykjavik was the next step towards understanding the user’s search intent. This algorithm learned to identify language preferences of a user, and depending on that, display more or fewer search results in English.
The algorithm mainly concerned searches in foreign languages, hence named after a foreign city.
Kaliningrad (December 2012)
Kaliningrad brought personalized search to Yandex users.
According to the company’s announcement 75-80% of all search results became personalized for everyone, regardless of whether the user is logged in or not (unlike Google’s personalization).
Yandex claims that the audience responded better to the personalized results, while SEOs have to once again learn ow to adjust to the new environment.
Looking back, Yandex has come up with a new significant change in their search algorithm at least one per year, and sometimes twice. What will be the next one? All we can say right now is that probably the name will start with a D.