Experiments conducted as part of the study show as much as a 25% gain in profits using the models, which help retailers identify strong and weak selling products and make appropriate substitutions. The models also help retailers quantify lost sales from poor category management.
Significant improvements in profit for a category of food does not require a major revamp of the assortment, the researchers find. In experiments known as reconfiguration exercises, the authors found that the number of replacements range from 2 to 32 with an average replacement per brand of less than 1.
"The complexity of managing an assortment on the store shelf has grown tremendously in recent years," says Prof. Christopher S. Tang, Anderson Graduate School of Management, UCLA. "Increased product turnover and proliferation rates in most product categories make it difficult for managers to gauge consumer preferences and to utilize their limited shelf space optimally. Our research helps retailers manage that complexity."
The study, "A Modeling Framework for Category Assortment Planning" is by Prof. Tang; Teck-Hua Ho, The Wharton School, and Juin-Kuan Chong, NUS Business School, National University of Singapore. It appears in a special issue of Manufacturing and Service Operations Management (M&SOM), an INFORMS publication.
An abstract of this study is available online at http://www2.informs.org/Press/category.pdf
Product Proliferation
The study presents a modeling framework for managers to assess the revenue and lost sales implication of alternative category assortments. It also generates an alternative category assortment with higher revenue.
One model, called a purchase-incidence model, predicts the probability of an individual customer who purchases from a given product category during a shopping trip. The authors developed what they term a "no-purchase probability," which enables them to estimate lost sales due to assortment changes in the category.
Another model, called the brand-share model, extends earlier research by predicting which brand a customer will choose if a purchase incidence occurs in the category. It employs new "brand-width" measures that quantify the similarities among products of different brands within the same category.
Data
The authors’ analysis of the data shows that a reconfigured category assortment can have a profit improvement of up to 25.1% with 32 products replaced.
For the data examined, they find the level of lost sales due to poor assortment changes could range from 0.9% to 10.2% for a period of 26 weeks.
The authors examined scanner panel data from a single market in a metropolitan area in the US. The data was provided by Information Resources, Inc. The data contains shopping information from 548 households over a two-year period, from June, 1991 to June, 1993.
In addition, the data set contains purchasing information in eight food categories at five stores located within a two-mile radius. These eight food categories are ice cream, coffee, frozen pizza, hot dogs, potato chips, regular cereal, spaghetti sauce, and yogurt.
For each consumer, the researchers tracked every shopping trip, whether the shopper bought in a category and what brand was purchased.