Impatient or selective? Inferring customer preference in scheduling attended home delivery from in-store customer data

Retailers have spent billions of dollars trying to speed up delivery, often hurting profits. Particularly challenging is attended home delivery (AHD), the last-mile delivery process where customers must be present at home to receive goods or services. Using data from a major furniture company that specializes in premium mattresses and beds, we study how to improve AHD by understanding customer preferences for delivery speed. We focus on the trade-off between faster service and the customer’s own availability in the absence of trade-offs with other delivery attributes such as price or delivery window precision. We develop and estimate a model of customer delivery date preferences where the selection of a delivery date is affected by the lead time, the day of the week of candidate delivery days, and the day of the week of their purchase—and interactions among these factors. We find that, even when delivery is always free, customers do not always want the earliest available delivery—a later date might be more convenient—and that assuming so unnecessarily restricts scheduling decisions. Moreover, the day of the week of the in-store purchase is a clue to the best delivery day; for instance, customers who shop on a Wednesday are more likely to be available for home delivery on weekdays than weekend shoppers. We show that, by leveraging customers’ preferences and shifting capacity, retailers can reduce fulfillment costs while maintaining service by better aligning delivery capacity with customer availability.