Flexible time window management for attended home deliveries

In the competitive world of online retail, customers can choose from a selection of delivery time windows on a retailer's website. Creating a set of suitable and cost-efficient delivery time windows is challenging, since customers want short time windows, but short time windows can increase delivery costs significantly. Furthermore, the acceptance of a request in a particular short time window can greatly restrict the ability to accommodate future requests. In this paper, we present customer acceptance mechanisms that enable flexible time window management in the booking of time-window based attended home deliveries. We build tentative delivery routes and check which time windows are feasible for each new customer request. We offer the feasible long delivery time windows as a standard and let our approaches decide when to offer short time windows. Our approaches differ in the comprehensiveness of information they consider with regard to customer characteristics as well as detailed characteristics of the evolving route plan. We perform a computational study to investigate the approaches' ability to offer short time windows and still allow for a large number of customers to be served. We consider various demand scenarios, partially derived from real order data from a German online supermarket.