Efficiency and Benchmarking Using DEA and Tessellation in Retail Stores

In today’s highly competitive in-store retail industry, it is important to compare performance of stores. Techniques to evaluate a store’s productivity and its relative efficiency involves evaluating many factors - such as sales turnover, sales per floor space, employee productivity, customer traffic, catchment area, gross margin, among other factors. Generally, store efficiency is evaluated by considering two metrics at a time, like sales per floor space or customers per employee. This paper used Data Envelopment Analysis (DEA) - a non-parametric linear programming procedure which combined multiple inputs and outputs, to get the overall performance in terms of an efficiency score, which was then used to compare retail stores. For input consideration, the paper also used Voronoi tessellation method and mapped it to census data to get the target population i.e. serviceable addressable market (SAM) by identifying catchment area for each store. The specific case pertains to 30 stores in the state of Ohio in North America for a multi-brand retail chain for the purpose of the paper. This paper presents the application of these two methods seamlessly for the business benefit of retail industry for the first time in one single use case. The findings apply to companies that are trying to compare the efficiency of retail stores and can be further extended with competitor information on product and promotions from market surveys. This approach is also applicable to other verticals of healthcare and education to compare performance of healthcare units or clinics or institutes with relevant input and output parameters.