Otkritie Bank, with the support of GlowByte, has introduced a system for personalizing categories and cashback rates

2023: Introduction of a system for personalizing categories and cashback rates into the loyalty program

Otkritie Bank has introduced a system for personalizing categories and cashback rates into its MAKS loyalty program. The IT partner of the project was the Advanced Analytics GlowByte team. Analysts have built a complex balanced system based on open-source technologies that takes into account both customer needs and business interests and risks. This was reported to TAdviser on November 24, 2023 by representatives of GlowByte.

When selecting relevant offers for the client (an individual set of MCC categories and cashback rates in them), the system determines the client's satisfaction - how much he expected to receive at a given rate and how much he will receive, taking into account the restrictions, as well as the difficulty of obtaining cashback, that is, how much the client is able to spend in the proposed category.

At the same time, the system takes into account the interests and risks of the business: restriction on the available amount of cashback; a strategy to retain customers who reduce activity; strategy for the development of customers who have potential for this (if the client does not have transactions in the category where most cardholders make them, then it is in this category that the client can be offered a high cashback rate in order to expand his category profile and thereby increase loyalty).

Also, within the framework of the project, a statistical monitoring system was implemented, which allows you to assess the economic effect of personalizing bonus offers in each individual micro-segment of the client base. Individual analysis of microsegments with especially high or especially low effect allows you to get new insights about client behavior.

There are no ready-made ("boxed") analogues of the system. Personal proposals are formed by solving a complex optimization problem in which the forecasts of several mathematical models act as dynamic constraints. At the same time, the problem is solved by minimal technical means: a significant part of the code is written in SQL (and MVP is implemented only by SQL) - such a solution does not require any additional costs from the customer for servers, software and their support.

Previously, we accrued fixed cashback to all customers without using a personalized approach. But gradually came to the hypothesis that not all clients equally need cashback: if you offer customers to determine cashback categories on their own every month, not everyone will take advantage of this opportunity. Such mechanics allow you to charge more cashback to those customers who are really interested in it. But to switch to this mechanics, it was necessary to abandon a single fixed cashback and introduce a personalized loyalty program. The personalization system was developed and implemented by colleagues from GlowByte. Thanks to her, we not only provided the client with the opportunity to "turn on" cashback every month on their own, but offered to choose one of the bonus packages specially formed for him, which will allow to receive the maximum benefit from purchases, said Nikolay Bubnov, business leader of the Loyalty and Partners tribe, Otkritie Bank.

In its complexity, this project allowed us to put together all the best developments from the field of game theory and imitation budget management. As a result, we managed to create a universal platform for the subsequent solution of similar problems in other industries, noted Dmitry Zamonin, solution architect, Advanced Analytics GlowByte.