Abstract
Problem definition: We investigate the performance of human decision makers doing assortment planning, that is, selecting a subset of products to offer in a retail environment. This task requires understanding of product interaction effects, such as cannibalization and inventory pooling benefits, and involves solving a complex mathematical problem with stochastic and combinatorial components.Academic and Practical Relevance: 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: We develop a computer simulation of a market environment and conduct a behavioral experiment, where subjects act as assortment planners. We analyze subjects' decision and clicks in different treatments where we vary the decision support systems provided to the subjects, which we imitate by providing intermediate calculations to simplify their task.Results: We find that subjects make intelligent decisions, and a substantial number of them figure out the optimal solution even when it is not intuitive. At the same time, we observe certain systematic patterns in the subjects' choices such as that the assortment planners' choices may be affected by the total number of possible products to offer. We also find that subjects who receive additional information do not always exhibit better performance, and in some treatments their performance is actually worse compared to the subjects who receive no decision support.Managerial Implications: Assortment planners should be made aware that they have a bias towards offering more products when more products are presented for their consideration. Also, managers should exhibit caution when implementing assortment planning decision support systems and verify that they indeed add value to the decision process