Moscow, 1 November 2017 - X5 Retail Group N.V. ("X5" or the "Company"), a leading Russian food retailer (LSE ticker: "FIVE"), has introduced machine learning technology to improve targeted marketing at its Perekrestok supermarket chain and to develop personalised offers for all members of the Perekrestok Club loyalty programme. Machine learning is now used to create more than 70% of targeted promotions, and has sped up the process sevenfold. This was made possible thanks to the introduction of self-learning analytical CRM functionality.
Machine learning is enabled in the analytical module (aCRM) of the network's loyalty programme to group customers into segments based on common features and create offers tailored to them. The system analyses a given customer segment and models personalised marketing campaigns that account for several hundred behavioural and demographic factors. It processes data on the frequency and ticket size of purchases, customer preferences, lifestyle factors, acceptable price levels, favourite product categories and individual visiting hours. Customer sensitivity to offers is predicted using various analytical techniques that include regression, decision trees and neural networks. The aCRM module also identifies the most efficient communication channels and automatically generates marketing messages by customer segment. All of Perekrestok's communication channels are now integrated with the aCRM. In 2018, the system will be connected with the SmartWifi module to determine the best timing of messages to customers.
Fabricio Granja, X5 Retail Group's Chief Information Officer said: "Big Data analytics using automated intelligent solutions open up new development opportunities in terms of retail networks' value propositions. Intelligent systems are set to become increasingly important for businesses, which is why are paying close attention to data processing. The Perekrestok project is a successful case of using data analytics produced by a self-learning system to foster customer loyalty. Marketing offers that are created based on hundreds of factors increase the effectiveness of targeted marketing by 5% and reduce communication costs by 40%."
In addition to questionnaire data, the system also collects and uses information based on analysis of behavioural and demographic factors. The processes of data analysis and customer segmentation are fully automated. The analytical system's interface makes it possible to schedule the date and time of customer communications, keep records of marketing promotions, and fine-tune campaigns based on test runs. Marketing managers only need to set targets at the outset of a project and monitor the campaign results. On average, an effective model takes about three months to complete.
X5 had been calibrating this system since March 2016, and in autumn 2017 it launched advanced algorithms for automated customer segmentation and campaign targeting to include all members of the Perekrestok Club loyalty programme.
Download (PDF. 0.2 MB)