With high selling prices of jewelry one can imagine the need to optimally use shelf space in stores and minimize abundant stock because of high costs. It has become extremely important to match supply and demand as good as possible. Therefore, the daily demand of 40,000+ SKU-shop combinations are predicted for all 41 Swarovski stores in Belgium and the Netherlands. The goal of this project was to increase conversion by optimizing the inventory of individual shops according to the customers needs and increase profits by minimizing lost sales and redundant stock at end of season. Besides, the solution lays the foundation of a data driven organization, while implemented in current business processes.
We created a predictive application that prescribes intelligent order quantities, per SKU per shop, for automatic stock replenishment. These quantities are based on the current stock level and the predicted demand for a particular product per individual store, versus the probability of lost sales and/or redundant stocks at the end of the season. Moreover, business rules were implemented for several product categories, such as the all-time favorite products (<1% chance of out-of-stock) and the current national best sellers (<5% chance of out-of-stock), to guarantee that these most popular products are always in stock in every single store.
The demand at Swarovski is heavily influenced by numerous factors such as seasonality and the occurrence of events (e.g. Valentine’s Day, Mother’s Day and Christmas). In this solution, we collected and analyzed many different internal and external data sources to enhance the accuracy of the outcomes at any given point in time.