Effectiveness of Product Recommendations Under Time and Crowd Pressures

Abstract

Understanding the effects of contextual factors is crucial in designing context-based marketing. This paper focuses on product recommendations and studies how time and crowd pressures — two prominent contextual effects in the consumer behavior literature — can impact the effectiveness of recommendations. Measuring these effects is not straightforward because the joint distribution of consumer choice, time, and crowd pressures is rarely observed outside the laboratory, and recommendations are often endogenously determined. We overcome these issues using data from an experiment conducted with vending machines in railway stations across Tokyo. The machines are equipped with a facial recognition system to make recommendations, and recommendations are changed exogenously in the experiment. This setup provides us with well-measured variables of the time and crowd pressures that affect the effectiveness of recommendations. After showing that recommendations increase the sales of both the recommended and non-recommended products, we show that time pressures moderate the effectiveness of product recommendations for both recommended products directly and non-recommended products indirectly. Crowd pressure weaken the direct effect on the recommended products although its impact on the non-recommended products is small and not robust in some cases. These results indicate that, when marketers make context-based recommendations, they should be mindful of the consumers under time pressure.

Keywords: Product Recommendation, Time Pressure, Crowd Pressure, Spillover Effect, Context-Based Marketing