Using Machine Learning Methods to Improve the Efficiency of Attracting New Customers at Tinkoff

Today, many companies want to create their own full-fledged ecosystem that will allow them to earn more from their clients. Tinkoff Bank is not an exception in this matter, there are a lot of processes going on inside that are aimed at increasing the number of products each customer of the company has. This study will examine the cross-selling approach, which implies that the company tries to sell new products to its current customer. The purpose of the study is to find an optimal approach to identify a segment of customers who did not respond in any way to a regular marketing mailing, but could take a targeted action with a more favorable unique selling proposition. Uplift modeling was proposed as a solution. By comparing the segment of customers that the model identified with a random one, statistically significant gains in the target metric were obtained.