Abstract
Problem definition: We investigate the performance of human decision makers selecting which products to offer in a retail category, also known as assortment planning. In theory, this task requires the understanding of product interaction effects, such as cannibalization and inventory pooling benefits, and it involves solving a complex mathematical problem with stochastic and combinatorial components. Despite its practical importance, assortment planning is often performed by managers without any decision support system. In the academic literature, there have been no systematic studies of assortment planners’ behavior. Methodology/results: We develop a computer simulation of a market environment and conduct a behavioral experiment where subjects act as assortment planners. We analyze subjects’ decisions in different treatments by varying the decision support systems provided to them, the number of possible products to offer, the size of the expected-profit-maximizing assortment, and the default assortment. We observe systematic patterns in the subjects’ decisions; their variety choices appear to anchor at a value equal to half the size of the choice set (mean-anchoring bias) and to the size of the default assortment (status quo bias). Finally, we document some interesting effects of decision support information; its impact on performance appears to depend on how easily it can be interpreted. Managerial implications: Retailers should recognize that the size of the choice set influences the chosen assortment, leading to excessive variety when large and insufficient variety when small relative to optimal variety. Funding: This work was supported by the Naveen Jindal School of Management at the University of Texas at Dallas. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2020.0256 .