Yandex Direct tests ML tool that sets personalized discounts in Search ads
A machine learning model inside Yandex Direct now decides which online shoppers see a discount and how deep it runs, through a new beta tool called 'Discount Management.' The model scores anonymized buyer behavior before a click happens, so the discounted price already appears in Search results. Retailers keep control over the mechanics: they choose the products in the promotion, set a total discount budget and a maximum discount size, and the algorithm distributes that budget across shoppers. The system blends two approaches agencies typically run separately, an open discount available to any buyer and a personalized offer usually reserved for repeat customers. Yandex Pay processes payment on discounted purchases with any bank card, and the tool is free while the closed beta runs, though access stays limited to a small group of companies for now.
Discount budgets that agencies used to allocate by hand, guessing which segments would respond, now run through an algorithm that reallocates the same money toward the shoppers most likely to convert. Campaigns tied to 'Discount Management' show variable prices to different buyers on the same product, which complicates the flat-rate promotions clients often expect from a sale period. Because the price cut shows up before the click, in the ad itself, media buyers running Search campaigns through Yandex Direct will need to watch conversion and margin data closely to see whether the algorithm's picks match what a human planner would have chosen.