The ML model analyzes data on purchasing behavior and distributes the discount budget more accurately than can be done manually.

The Yandex team has started closed testing of the “Discount Management” tool in “Yandex Direct”. It allows online stores to automatically distribute their discount budget to increase the number of orders without increasing promotional costs.

The solution is based on an ML model that analyzes anonymized data on customer behavior and determines to whom and what discount size might be most effective. The algorithm can take into account, for example, new customers or users who have not made purchases for a long time.

The company notes that previously, marketers set up such campaigns manually, either offering the same discount to everyone or limiting personal offers to regular customers. The new approach combines personalization with the ability to attract customers who have not previously purchased goods from a specific store.

Businesses independently select products for the promotion, set the total budget, and the maximum discount size. The calculated offer is displayed in search ads even before the user goes to the website.

In the future, the tool is planned to be used in other channels, including regular search results and “Alisa” recommendations. Orders with a discount can be paid with any bank card via “Yandex Pay”. During the closed testing phase, access to the solution is provided free of charge.