Man(ager Heuristics) vs. Machine (Learning): Automation for Prediction of Customer Value for Brick-and-Mortar Retailers

Abstract

Technological development has led to rich datasets, fast processing capabilities, and a large body of literature on accurate yet complex models. However, although managers see potential in becoming data-driven, few successfully apply contemporary analytics. In retailing, some of the hurdles are that (a) most applications are for online settings, while most retailing is still conducted in brick-and-mortar settings, (b) predictions of customer lifetime values are less relevant for rapid (automated) actions in real-time, and (c) there is skepticism due to the lack of empirical testing beyond large international firms, tech start-ups, and digital natives. In this study, we attempt to bridge this gap by exploring the potential benefits of automated machine learning compared to manager heuristics in predicting immediate future customer value in real-time, as applied on 338,184 grocery receipts, 179,568 beauty receipts, and 111,289 non-prescription pharmacy receipts. Our results from different retailing industries with various product characteristics in brick-and-mortar contexts show that automated machine learning provides great benefits in predicting immediate future customer value. This suggests that, even with limited know-how, brick-and-mortar retailers can implement contemporary analytics for better customer prioritization in real-time.