Retailers cannot capture customer demands unless their locations can be easily accessed by customers. Though analytical models are available for dealing with the location problem, no empirical progress has been made in identifying attractive locations to maximize the revenue. This article deals with an online grocery retailer who uses a buy-online, pick-up-in-store (BOPS) fulfillment method. The retailer uses delivery trucks parked at various locations on specific days so that customers can collect their orders from the nearest truck. These pickup locations include schools, businesses, gyms and parking lots and operate once or twice a week. Usually the retailer will not change the week days and hours when the truck delivery is done to maintain consistency. As the business expands, the retailer would like to close down the under-performing delivery locations and introduce new ones. These settings provide voluminous data that can be used for relocating and introducing new locations in both time and space dimensions. The brick-and-mortar format may be more attractive to customers than the current method because of limited availability. However, since grocery purchases are done ahead of the requirements, customers can adopt to BOPS. It helps the retailer to access multiple markets with limited investment. The retailer tries to identify locations with high sales potential due to factors such as other stores nearby, income level of the people in that location and the population and its density. In spite of all these factors, the orders picked up at the different locations vary considerably. The objective of this study to optimize the retailer's location and operating schedule using a combination of machine learning and econometric methods.
展开▼
