Matrixnet in the Yandex Advertising Network
In August, Yandex began using the Matriksnet machine learning system to display ads on thematic YAN sites, where advertising is shown to the user in accordance with his interests and the content of the web page.
The new selection formula takes into account a greater number of factors, including the peculiarities of the user's behavior in the network, which made it possible to select for the users to display more interesting advertisements. As a result, the system more accurately predicts the likelihood of a click on an ad — during the first week after the introduction of MatrixNet, the click-through rate of ads has grown by an average of 22%.
With the introduction of the new formula on YAN sites, low-cost-per-click ads appear more often. This led to an overall decrease in the cost per click on average for the network by 12%, that is, placing advertising on the YAN became more profitable. Due to the increase in the number of clicks, the income of site owners on average over the network increased by 10%.
Machine learning will allow YAN to further improve the formula, adding new factors to it. Thus, it will be possible to more accurately determine what kind of advertising the user needs at the moment.
')
According to comScore Media Metrix, in July 2011, the RSL audience in Russia amounted to 43.6 million people. For a month, a user views an average of 540 pages of YAN sites with Yandex.Direct and Market ads. Today, the Yandex Advertising Network is the largest in runet in terms of audience reach.
The new selection formula takes into account a greater number of factors, including the peculiarities of the user's behavior in the network, which made it possible to select for the users to display more interesting advertisements. As a result, the system more accurately predicts the likelihood of a click on an ad — during the first week after the introduction of MatrixNet, the click-through rate of ads has grown by an average of 22%.
With the introduction of the new formula on YAN sites, low-cost-per-click ads appear more often. This led to an overall decrease in the cost per click on average for the network by 12%, that is, placing advertising on the YAN became more profitable. Due to the increase in the number of clicks, the income of site owners on average over the network increased by 10%.
Machine learning will allow YAN to further improve the formula, adding new factors to it. Thus, it will be possible to more accurately determine what kind of advertising the user needs at the moment.
')
According to comScore Media Metrix, in July 2011, the RSL audience in Russia amounted to 43.6 million people. For a month, a user views an average of 540 pages of YAN sites with Yandex.Direct and Market ads. Today, the Yandex Advertising Network is the largest in runet in terms of audience reach.
Source: https://habr.com/ru/post/128724/
All Articles