OREANDA-NEWS. September 19, 2011. Yandex starts using its proprietary machine learning technology MatrixNet to display ads in Yandex’s Advertising Network (YAN). It allows the system to deliver ads better related to web users’ interests. During the first week of MatrixNet utilization, the average clickthrough rate has grown more than 20%.

About 80% of all ads placed on YAN are contextual. These ads are automatically served on web pages based on user interests or the content of the page where they are displayed. Using its machine learning technology, Yandex developed a new formula to select such ads. This formula takes into account more factors, including user’s online behavior. With a more elaborate formula, the system can more accurately predict the probability of an ad to be clicked.

“We have just started using MatrixNet in our advertising network and the results are quite impressive,” says Dmitry Popov, head of Yandex’s Advertising Network. “Using our machine learning technology, we’ll keep optimizing the ad selection formula and be adding new factors. This will help us to better understand what ad a particular user wants to see. With a higher clickthrough rate, the owners of advertisement platforms can increase their revenue, while advertisers can attract more customers.”

The Yandex Advertising Network includes thousands of various websites. According to сomScore Media Metrix, as of July 2011, a monthly audience of the network in Russia is over 43.6 million. Every month, an average web user views 540 participating web pages showing ads served by Yandex.Direct or Yandex.Market. Currently, Yandex’s Advertising Network is the largest in the Russian internet in terms of audience reach.