The optimization of store and warehouse based e-grocery fulfillment

Abstract

After widespread adoption of e-commerce in the retail industry, online grocery sales by e-grocers are getting more popular as well. These e-grocers are struggling to optimize their fulfillment operations. In practice, two main modes of fulfillment can be identified: store and warehouse based fulfillment. In both modes, the requested products are first picked, i.e., collected from their storage locations, and afterwards delivered to the customer. In store based fulfillment, the existing network of grocery stores is used to pick the items, while in warehouse based fulfillment a dedicated warehouse is used for the picking operations. Both store and warehouse picking have benefits and downsides. Grocery stores are usually located in the middle of the city, allowing for fast delivery after picking. However, order picking in stores is less efficient than in a warehouse, due to the layout of the store and the interactions with physical customers. While the warehouse allows for very efficient picking, it is usually located out of the city’s center. As such, the average delivery distance to customers is relatively large. In our research, this trade-off between efficient picking and delivery is studied. We consider an integrated problem in which customer orders should be assigned to one of the fulfillment locations and the delivery routes, starting from an order’s assigned location, should be constructed. First, a mathematical model is used to optimize the operations on small problem instances. Next, a metaheuristic algorithm is developed and used to solve larger problem instances. To obtain practically relevant insights, a real-life data set is used to generate the e-grocery orders. Moreover, the algorithm is used to optimize the fulfillment operations in different problem settings to identify the critical parameters influencing the order assignment decision.