Popularity prediction and optimal location: Application to restaurants

Abstract

Location is one of the most important strategic decisions for a retail firm. It is often a high-stakes decision for the firm insofar that it involves a large investment, is very difficult to rectify, and affects profits and operations for many years in the future. In this paper we consider the location problem for a retail facility, specifically a restaurant, a consumer-facing storefront that provides a service and competes with other establishments. The location decision is difficult for the following two reasons: (i) Forecasts of demand and revenue are difficult to make as the firm has no historical demand data of its own. This demand is highly dependent on the population of the catchment area and its tastes, what other establishments are in the area, as well as the features of the proposed establishment. (ii) Future entry and exits of competitors would affect demand and the firm’s long-term profitability, but predicting such future sporadic events is very difficult. In this paper we use online reviews of the incumbent establishments to forecast the former; incumbent facilities have observable as well as latent characteristics (such as quality of service) and customers have taste and travel preferences. We show how publicly available data on demographics and density can be combined with online review volume and valences to create good forecasts, which we use then in a simple model of
forward-looking competitive entry and exit for predicting locations that will survive over time. Apart from a tractable toolkit to help their decision process, we show via counterfactuals, that optimized location decision making can increase chances of survival (which we define as existing in a steady state) by up to 37.5%. Our estimation results show that customers differ significantly in their willingness to travel and review rating sensitivities across restaurant types. Managerial insight into the nature of competitive location dispersion is also provided.

Keywords: Popularity prediction; retail; location; analytics; online reviews

JEL Classification: C51, L83, M31