A comparison of online recommendation methods: Simultaneous versus sequential approaches

Abstract

This study investigates the relative profitability of online recommendations (sequential or simultaneous) of durable goods used by various firms, such as Walmart and Amazon. To achieve this, we develop a utility model (specified in terms of product attributes) that captures the complementarity between the recommended products. To recover customers’ utility for simultaneous or sequential purchases of durable goods, we conduct incentive-aligned conjoint experiments in which participants select a combination of fully or partially complementary products. The results reveal that the product attribute associations are more noticeable for fully complementary product categories. In addition, customers tend to balance comparable attributes when buying durable goods sequentially over time, while they reinforce comparable attributes when buying simultaneously. The profit analysis based on the estimates shows that the recommendations for simultaneous purchases can yield higher profits for fully complementary product categories than for sequential recommendation purchases. However, the recommendations for sequential purchases generate higher earnings for partially complementary product categories than for simultaneous recommendation purchases. A simulation study reveals that the profits differ considerably by combinations of product attributes because customers evaluate complementarity among attributes differently between the two recommendations.