Mamba is a popular mobile app for dating over 40 million users from over 50 countries around the world. The application is free and receives revenue from advertising.
How was the experiment
For two weeks, application developers tested different approaches to mobile mediation: the classic “ladder” or “waterfall” model and Mobile mediation from Yandex with a threshold selection system.
The key goal of the experiment is to compare the effectiveness of monetization selection models and increase revenue from ad units in a mobile application. The experiment was carried out in the application on native banners. At the same time, half of the requests to monetizers went through the waterfall model, and the second through the threshold selection system. Using Mobile mediation, they set up monetization in the application from the Yandex Advertising Network and AdMob. To add advertisements from third-party monetizers, the identifiers of blocks of selected advertising systems were indicated in the Mobile Mediation interface. Next, we integrated the latest version of Yandex Mobile Ads into the application and launched mediation. The experiment took about two weeks.
As a result, we saw that in all cases, regardless of the location of the ad unit in the application, the revenue from impressions with Mobile Media was higher than the income from the classic “waterfall”. On average, in all blocks with Mobile mediation, advertising revenue increased by 30%, while the average fill rate was 75%. The largest revenue growth was shown by the block in the carousel with photos: it amounted to 55%.
How Mobile Mediation Works
The system of threshold selection based on machine learning helps to get the maximum income from advertising in the application. She predicts which monetizer will buy the impression and makes a request primarily to him. In fact, the system predicts which of the monetizers will offer a large bid for a particular user, taking into account the statistics of impressions for past periods, the type of device, its behavior, geolocation and hundreds of other factors. Thus, the number of requests to monetizers is reduced and the advertisement is displayed faster.
At the same time, the site receives the maximum income, and the advertiser, in turn, receives the highest quality traffic and relevant audience. The training period for the algorithm for its maximum efficiency in the Mamba application was about two weeks.
Graudin Kirill, Head of Mamba Advertising: “For users, both the website and the Mamba mobile application are free. Therefore, the issue of effective advertising monetization of the project is very relevant for us. We are actively trying new ways to increase the effectiveness of advertising on the site, including through Mobile Mediation algorithms. In addition to monetary indicators, advertising quality and a variety of formats are also important to us. It is also important for us that advertising does not affect the use of the application by users, there should be no delays in the content of the application due to the selection of a banner. Given that the application is used in more than 50 countries around the world, an additional bonus for us is the opportunity to monetize a foreign audience. It was possible to do this by setting up a threshold selection system for Mobile Mediation and its integration with international advertising systems.