The Russian IT company from Moscow was established in 2018. At the moment, the company is developing a cloud service for the formation of individual recommendations to customers based on a retrospective analysis of sales data. Typically, customers are identified by the retailer’s loyalty program.
The product provides the formation of unique personalized offers for each identified client of the trading network. Based on the retailer’s data and the external data that the company collects itself, the service predicts the future behavior of each customer and tells the retailer what personalized promotional offers and at what moment of time should be sent to each buyer to motivate them to come and purchase suggested goods. The recommendations are formed by the top-notch AI-algorithms. The further communication of offers is provided by the retailer.
The solution reaches two goals of retail chains:
1. Increased turnover: by increasing the depth of the check due to timely accurate and relevant recommendations.
Profitability growth: due to the reduction of promo depth due to individual recommendations for goods without a discount or with a small discount.
2. Growth of the market share: due to the switching buyers to particular stores from competitors' stores, thanks to the forecasting of the next purchase time and the invitation to the client to take advantage of special offers on the day of the predicted visit.
The Russian company is very active in the home market and is now willing to grow internationally.
It's particularly looking for European retailers with the collected data on consumer behavior of loyalty program members for the outsourcing agreement.
The preferable sectors are FMCG, electronics, and pharmacy, but contacts from other spheres would be also welcomed.
As part of the outsourcing agreement, the foreign partner first provides depersonalized data (with hashes of contact information) about all transactions of identified customers. The format of the upload depends on the technical capabilities and the existing infrastructure of the retailer: it can be just a CSV file that is transmitted via FTPS or API.
Training of predictive and recommender algorithms on average takes up to 3 weeks. During this time the Russian company enriches retailer's data with data from 50+ sources and train models. After that, the retailer loads up the transactions that have passed during this time. Next, on a weekly basis and at a set time, the retailer will receive from the company a list of recommendations for the upcoming week. The file contains the Id-clients with whom to communicate this week, the recommended time of communication and a list of individual product promotions with a personalized discount. When preparing a file for sending, the Russian company takes into account the retailer's infrastructure, so further actions on its side are minimal — the file is made in the required format and can be immediately loaded into its communication system. After a week, the foreign partner updates the transactions for the past week, and the Russian company prepares a product promo plan for the next week. Also, the report with the obtained metrics is sent to the partner at the end of the week.