Selling personalized substitutes and co-purchases in online grocery retail

We investigate how an online grocery retailer can make personalized recommendations for substitutes and co-purchases as a customer adds items to their shopping basket. Of particular interest is balancing revenue maximization with long-term customer retention, which is an important business objective but less studied in the academic literature. Methodology/results: We propose three different optimization models for this problem: (i) pure expected revenue maximization, (ii) maximization of a weighted average of the expected revenue and consumer surplus, and (iii) expected revenue maximization with a constraint on the variety of items in a customer's shopping basket, which is a measure of how much the customer relies on the retailer to meet their grocery needs. We devise computationally efficient algorithms to solve the different decision optimization models, which can find optimal solutions much faster than the brute-force method when tested out on a large, publicly-available data set from an online grocery retailer. Managerial implications: Our work shows that online grocery retailers can use fast and accurate optimization-based algorithms to recommend substitutes and co-purchase products as a customer does their online shopping, where the decision problem can take expected revenue, consumer surplus and shopping basket variety into account.