Marketing analytics using anonymized and fragmented tracking data

With the digitization of the retail industry, there is a growing abundance of event-based tracking data describing consumer behavior (e.g., online clickstreams and offline sensors tracking the movement of shoppers). However, stronger data privacy regulations and the growing privacy consciousness of consumers suggest that much of the data may increasingly only be available to retailers in an anonymized and fragmented form that does not identify individual consumers exactly. In response to the relative paucity of research on marketing analytics in retailing using anonymized and fragmented event-based (AFE) tracking data, this paper makes three interrelated contributions. First, we describe the relevance of AFE data in the future of retailing, contrasting it with other forms of aggregate and individual-level data. Second, we propose a methodology for analyzing AFE data, which allows us to approximately recover individual-level heterogeneity and derive meaningful variables from the raw data. Third, we validate the methodology using representative data collected by deploying sensor-enabled shelves in a field experiment within a store. We find that our approach to analyzing AFE data can help uncover interesting patterns of consumer behavior and could be applied across other online and offline retail settings in practice.