Lenta reduced write-offs thanks to artificial intelligence

2022: Scaling an automated demand forecasting system

On June 7, 2022, Lenta announced the scaling of an automated system for forecasting demand for regular and promotional goods for stores of all formats. Over the year, the company reduced write-offs by 4% at the initial stage, the growth in the availability of promotional goods reaches 5%.

The analysis and forecasting of demand for goods is carried out by multi-parameter models, which are based on more than 100 factors: store parameters, product characteristics, checks, prices, planned promotions, analogous goods, etc. Calculations take place using the Azure Databricks dynamic computing cluster and the Azure Data Lake Storage big data model.

Centralization of the order within one service made it possible to provide a forecast-order bundle and improve the quality of feedback work. Previously, orders were manually formed by store employees.

The solution was replicated on 254 Lenta hypermarkets in 10 iterations in less than a year. 24 processes and 13 communication streams were redesigned.

Improving the quality of the forecast affects the improvement of business indicators such as availability, sales, write-offs, turnover of inventory. The peculiarity of the solution is that we managed to achieve good results in reducing write-offs not only of regular sales goods, but also of goods participating in the promo, said supply chain management director Igor Ovsienko.

The result of the project implementation was a decrease in commodity losses (write-offs) compared to 2020 for goods of the "deli" category by 4% at the initial stage and an increase in the availability of promotional goods to 5%. Automatic ordering based on a smart forecast freed up store employees. The company expects further growth in business effect as technology evolves.

Compared to the previous demand forecasting decision, the accuracy of the forecast increased by 40 percent. As part of the project, the company centralized the function of orders and abandoned their "manual" registration in stores. The recipe for success in solving such problems is a deep business expertise of a cross-functional team of five services of our company, which became the basis for successful automation, noted Director of Digital Innovation and IT Sergei Sergeyev.

In 2021, the company in two months scaled the solution in conjunction with a centralized order for new small stores "Mini Lenta" and new categories of goods "fresh vegetables and fruits" and "ultrafresh."