OK has implemented a new AI-based user return and retention system

2025: Optimizing User Return and Retention

Classmates have introduced an updated system for returning and retaining users of the social network, which operates on the basis of ML technologies. The company announced this on March 21, 2025. Internal algorithms analyze the behavior of social network users, segment the audience, personalize the content, optimize the time for sending push notifications and their number per specific user, taking into account his interests. The introduction of the model allowed OK to return part of the previously inactive or less active audience and increase its size in 2024.

The system divides push notifications into two types - mandatory and advisory. The first are related to notifying the user of important updates that are sent to him at the time of the event. These are, for example, new requests for friends, messages or gifts. The second type of notification is regeneration pushi. They relate to proposals to study fresh content appearing on the OK site, taking into account the interests of the user. The model takes into account dozens of parameters when interacting with content and when sending a notification takes into account that the user, for example, began to watch Clips more often or read posts of bloggers of a certain topic.

This technology contributes to more accurate segmentation and personalization, increasing user engagement and increasing the time spent in the application and allowing the site to communicate more precisely with its users and choose the most relevant reason and timing for this.

The updated renewal system appeared in Odnoklassniki in the third quarter of 2024 and, along with other product updates, influenced the growth of the MAU (monthly active audience of users) of the social network for the first time in five years. At the end of 2024, this figure increased to 36.2 million. Over the year, the audience of the social network increased by 1.2 million users.