S7 Airlines together with CleverDATA implemented a recommendatory system on the basis of machine learning

2019

On March 20, 2019 the LANIT company reported that S7 Airlines together with CleverDATA implemented a recommendatory system on the basis of machine learning.

According to the company, for system operation by specialists of CleverDATA and S7 Airlines consolidated acquisition of passengers from the large volume of internal data sources of airline was organized and the management system for tags and analytics is implemented. On the basis of the acquired information a series of models of machine learning for determination of recommendations about the aviadirections was developed and tested. For formation of models data on the previous flights of passengers and activity of users on the website were used. The received recommendations are applied in the personalized communications with passengers in the automatic mode.

By results And / V-testirovaniya for a number of segments of passengers the block of recommendations of the directions of flights on the website executed using machine learning showed significant growth in conversion from viewings in armorings, than the same block created by specialists manually. For email mailings the positive result of work of models of machine learning is also recorded. Improvement of indicators on all campaigns on the basis of automatic recommendations about comparison with accidentally created set of recommendations is mentioned.

The implemented scenario in S7 Airlines is the first step on the way of creation of the personalized communications with passengers.

Machine learning technologies, work with Big Data is not just style. We want to select automatically for our passengers those offers and content which are most interesting to them. So we will be able to increase efficiency of communications and in general passenger experience. Nikita Matveev, director of S7 Group data management

When developing a recommendatory system software products of CleverDATA, including DMPkit, platform for the organization of own solutions regarding collecting, storage and processing of user data were used.

Problem of a recommendatory system – not only to make the relevant proposal, but also to build the most effective communication with each consumer, considering the optimum frequency of communications, convenient time and the correct sequence of interactions in different channels. Such personalized approach helps to come into closer contact between business and the client, positively influencing, both business indicators, and on loyalty level to a brand from end consumers. For formation it is important to provide with the system of the correct recommendations correct process of work with data – to integrate as much as possible sources of information on the consumers, having involved data on activity on the website, about consumer behavior from the CRM system and about the previous communications in email mailings channels and online advertizing. All this is valuable "fuel" for training of analytical models of a recommendatory system. Denis Afanasyev, CEO of CleverDATA