Managing Demand Uncertainty for Perishable Products Through Omnichannel Grocery

This dissertation presents a collection of studies on the value of demand information in a grocery retail setting. Mutually aiming to improve operating outcomes and reduce food waste, the studies tackle the retailer’s daily challenge in deciding replenishment quantities while facing uncertainties in demand. Two central methods of acquiring future demand information are explored, including receiving advance orders from customers, and recapturing unsatisfied demand. With the rise of online grocery following Covid-19, customers are now placing more delivery and pickup orders in advance of the fulfillment date. This mode of interaction allows the grocer to collect advance demand information, which has been shown to reduce inventory costs for retailers of non-perishables. For perishable grocery items where spoilage is a major cost driver, the extent to which advance ordering can improve profit, or offset increased fulfillment costs, is not yet clear. To investigate this, a model that solves for the retailer’s optimal ordering policy is de- veloped to reflect a modern grocer’s inventory management operations. In addition, another mode of interaction, where customers are presented with the option to place backlog orders on out-of-stock products, is also considered. The economic impacts of implementing either or both omnichannel fulfillment options are analyzed through Markov Decision Process-based models in terms of changes to expected profits, outdating events, product freshness, as well as in-store and online product availabil- ities. A rich selection of independent variables are reviewed throughout when applicable, such as demand lead time duration or discount given to customers. Extensive numerical studies are conducted for short shelf life products at various price points, and the results indicate that a combination of the two sources of demand information can increase the grocer’s profit, reduce inventory and food waste, while maintaining the overall availability level. These benefits increase as the replenishment products’ mean shelf lives decrease, and continue to be observed even when advance and backlog orders receive price discounts. It is found that the average cart value, proportion of short life items, handling time, and order pick rate are strong determinants of the extent that two-day orders can offset omnichannel fulfillment costs. In many cases, there exists an attainable breakeven pick rate, such that the grocer’s net expected profit is higher with curbside advance orders than with traditional in-store purchases