In-Store Customer Behavior Analytics Using Computer Vision

Abstract

In the competitive landscape of modern retail, acquiring precise and real-time insights into customer behavior is a vital prerequisite for achieving sustainable competitive advantage. Nevertheless, methodological limitations inherent in conventional self-report instruments—such as surveys—and retrospective transactional data fail to capture consumers’ subconscious actions, thereby creating a pronounced gap between stated intentions and actual point-of-sale behaviors. Addressing this gap, the present study establishes an innovative analytical framework leveraging computer vision and image processing technologies to monitor and analyze shoppers’ physical behaviors in an objective, continuous, and non-intrusive manner. In this observational study, an automated pipeline integrating deep convolutional neural networks and multi-target tracking algorithms was deployed in an operational retail environment, capturing and processing spatiotemporal trajectories from 3,210 valid customer visits over a one-week period. The empirical findings reveal that a deceleration in shoppers’ walking velocity along product aisles serves as a robust behavioral indicator of the transition from spatial navigation to focused cognitive processing. Statistical analyses demonstrated the pronounced superiority of eye-level shelves in driving visual engagement and dwell time (p<0.001) relative to other shelf tiers. Furthermore, spatial heat-map analytics identified that 21.8% of the store floor area constituted “spatial blind spots.” Multiple linear regression analysis further indicated that the store spatial coverage ratio (β=0.41) and total dwell time (β=0.22) acted as significant positive drivers, whereas checkout delay (β=−0.28) exerted a detrimental effect on behavioral loyalty (repurchase frequency). By shifting the paradigm from post-hoc transactional audits to real-time behavioral intelligence, this study provides actionable implications for store layout re-engineering, physical space optimization, and loyalty strategy personalization, offering an evidence-based foundation for decision-making in next-generation smart retail.