Decision Support for Home Delivery

Business-to-Consumer e-commerce has not only increased the demand for package delivery services, such as those offered by Federal Express and UPS, but has also led to the development of new home delivery service models and activities such as grocery delivery services. By bringing goods “the last mile” to the customer’s front door, home delivery is seen by many as the ultimate value-added service for consumers. Many of the companies that have started to provide such services have found, however, that consumer direct delivery poses an enormous logistical challenge and are struggling to create direct delivery strategies that are profitable and can satisfy customer expectations. For example, the difficult challenge of matching grocery store prices while incurring high distribution costs led to closing of the San Francisco-based operations for Peapod and the declaration of bankruptcy by Webvan. The fulfillment process for most consumer direct businesses can be divided into three phases: (1) order capture and promise, (2) order sourcing and assembly, and (3) order delivery. Our research effort focuses primarily on the interactions between order promise (deciding on a delivery time) and order delivery (devising efficient delivery schedules). Better integration of order promise and order delivery decisions has the potential to substantially improve profitability, especially for those consumer direct businesses offering “attended” deliveries. Attended deliveries are those where the consumers must be present and may be necessary for security reasons (e.g. expensive computer equipment) or because goods are perishable (e.g milk, flowers), and are a vital feature of many consumer direct service models. To avoid delivery failures as much as possible, it is customary in attended home delivery services for the company and customer to mutually agree on a narrow delivery window, or time slot. Thus, it is important to be able to quickly determine in which time slots each delivery request can be serviced given the already committed deliveries. We are not only interested in how to effectively and efficiently evaluate the feasibility of each request, but also how to dynamically evaluate the profitability of each request in an effort to create improved decision technology to support home delivery services. Home delivery is a fairly new phenomenon, at least at this scale, and thus few models and algorithms have been proposed and studied that help create an understanding of the complexities and intricacies of these distribution problems. We propose and analyze new routing and scheduling problems that capture key features of home delivery, including the Home Delivery Problem (HDP), which allows the vendor to select which deliveries are accepted and the time slots for the accepted deliveries, and the Home Delivery Problem with Incentives (HDPI) which allows the vendor to dynamically compute the prices presented to a customer for the different time slots. Though the issue of whether to accept an order in hand or wait for more profitable delivery later is very similar to the core issues addressed in revenue management, the solution approach must be very different due to the cost structures involved. The costs associated with an airline seat do not depend on who occupies the seat nor on who purchases the rest of the tickets in the same fare class. In home delivery, though, the costs associated with accepting an order in a certain time slot depend on the delivery address and time slot of that order as well as on the delivery locations and time slots of the other accepted orders. We propose algorithms for both problems based on insertion techniques developed for vehicle routing problems to help vendors make these decisions.