This paper explores the logistics operations of instant grocery delivery services. We specifically concentrate on two widely adopted strategies: Personal Shopper Systems (PSS) and Inventory Owned Delivery Systems (IOD). In the PSS, couriers visit affiliated brick and mortar stores in the delivery area to pick and purchase ordered products and deliver them to customers. Whereas in the IOD, couriers collect products from a single distribution center or so-called a dark-store. In the IOD, the inventory of the DC is also managed by the service provider. Even though PSS strategy is asset-light, and hence attractive for investors due to the utilization of existing retailers in the area, maintaining a good level of instant delivery with the PSS is more complex than the IOD. Therefore, we introduce the Instant Delivery Problem (IDP), which explicitly addresses PSS-specific features such as real-time store selection and in-store product collection and can also be used to analyze the IOD. We propose a tailored rolling horizon framework that utilizes column generation to browse updated delivery plans for arriving customer orders. Computational studies both in real-life inspired settings and in case studies on selected urban areas show the PSS is a highly competitive strategy compared to the IOD, particularly when dealing with small-sized customer orders. We observe that the performance of the PSS is robust when varying the delivery service time frame. The case studies also suggest that the PSS becomes even more competitive in areas where the retail store density is high.
Lancaster University, Fudan University, April/2022