Stocking Algorithms

warehouseFor a large and very successful retailer, Ventana evaluated the company’s sales forecasting system, which the retailer believed to be state-of-the-art. By studying the company’s historical sales data and working with the client to understand the data better, we were able to show that assumptions built into the forecasting algorithms were inappropriate. We then showed via simulation how the company could achieve a 50% improvement in forecasting accuracy. Together we implemented a prototype system in a single retail outlet that showed the practical implications of the change. The client then proceeded to implement the new system in all of its stores.

Ventana also put the sales forecasting problem in the context of the retailer’s overall product distribution system. We showed how the tradeoffs among inventory levels, stock-outs, and product wastage or obsolescence operated in relation to the company’s overall operating policies. This showed several opportunities to improve operational performance by re-designing the supply system and its logistics algorithms and policies. With the improvements the client could determine how much of the benefit to take in decreased inventories, stock-outs or wastage.

Ventana’s contributions were made possible by creation of a simulation model that organized the client’s own data and business knowledge in a form that (a) allowed the performance of the existing sales forecasting system to be well benchmarked for the first time, and (b) gave the client a better understanding of how the business worked as a whole. This new understanding revealed that unjustified, but widely held, textbook prescriptions (and conventional wisdom) about customer behavior and inventory management were limiting the retailer’s profitability. During this project, the company implemented several of Ventana’s innovations in multi-store operational tests. These tests demonstrated the ability to reduce system-wide inventories by 50% without lowering service levels. As a result, Ventana algorithms were implemented throughout the company.