Motivated by our collaboration with an online platform operating in North America, we explore the joint optimization of the order fulfillment process with personalized delivery options in the context of e-commerce. Customers can choose from personalized fulfillment options to proceed with the purchase or leave with no purchase. The retailer periodically makes fulfillment decisions and relies on multiple logistic providers to perform the fulfillment operations. We model customer behavior with a general discrete choice model and formulate the joint optimization as a stochastic dynamic program. We propose a tractable deterministic approximation and develop a computationally efficient heuristic with a provable performance guarantee. We also extend the proposed heuristic to scenarios when customer behaviors are more complex and affected by fulfillment speed, cost, and order value.
Using real datasets collected from our industrial partner, we demonstrate the value of personalizing fulfillment options for the customers and jointly optimizing the options with fulfillment assignments. Our results show that demand management via personalized fulfillment options is prominent when customers favor quicker fulfillment and when the fulfillment capacity is limited. However, an optimized fulfillment operation becomes more critical when customers are more willing to wait.