Joint Pricing, Space Allocation and Assortment Planning: An Iterative Improvement Approach

Assortment planning determines the set of products to be sold and is central to retailers’ profitability. Despite its significance, there is no dominant solution yet in both academia and industry. The major reason is that effective assortment planning involves a combination of multiple factors: a thorough understanding of consumer behavior and demand, models that can effectively capture major concerns, and optimization techniques that can solve the models efficiently. The situation has been changing recently because of the huge amount of transactions data made available by the wide adoption of information systems. There are now golden opportunities for researchers to develop and verify their analytical solutions. In this project, we propose to solve pricing, shelf space allocation and assortment selection decisions simultaneously. We believe that the recent assortment planning research suffers from three major drawbacks. The first is that product prices are assumed exogenous and given, although price obviously has significant effect on demand. Second, the effect of display space on demand is ignored, although empirical research has shown such effect does exist. Third, the widely adopted demand model (specifically, multinomial logit model) has a few drawbacks that severely restrict its applicability.The objective of the project is to solve the assortment planning problem while taking the three factors above into consideration. We will investigate when and why display space effect should be considered in the assortment planning process and attempt to understand how it influences consumers’ choice behavior. We will take an address approach to build a demand model by endogenizing pricing and space allocation decisions. We expect such a demand model to capture consumer behavior more accurately. We fully understand that the new, enriched demand model will inevitably complicate, to a significant extent, the problem of solving for the optimal assortment. Therefore, we propose to adopt an iterative improvement approach so as to keep the problem tractable. By “iterative improvement” we mean that we will not target the ideally optimal assortment right from the beginning. Rather, we propose to continuously look for any possible improvement to the existing assortment; an improvement can be simply the inclusion of a new product, adjustment of the price of an existing product, or replacing an existing product with a new product. This allows a retailer to make only mild changes to its existing assortment before seeing the realized profit improvement. Such an iterative improvement approach is particularly attractive to practitioners because they are often very resistant to dramatic changes.At the end, we emphasize that we have locked-in support from a large local retail chain, which will provide us with test data and conduct pilot studies to verify our models and solutions. We firmly believe that the collaboration with industry will enhance the validity of our findings and ensure the practicality of our solutions.