Risk sensitivity or social signaling? Unmasking Behaviors with Video Analytics.

In 2020, as the novel coronavirus spread globally, face masks were recommended in public settings to protect against and slow down the spread of the coronavirus. Why did people comply, or not, while shopping in 2020? Do these motivations relate to their shopping behaviors? Based on the recent literature on mask-wearing, we define three customer groups: Self-Protectors wear a mask primarily because they are sensitive to their own health risk, while Social-Signalers wear a mask primarily to signal social responsibility; meanwhile, Insensitive-Shoppers do not wear a mask. We collected four months of shopping videos in 2020 in a department store where mask-wearing was not mandated, and we determined each customer’s group from their mask-wearing choice and shopping duration. We found significant differences in the purchased items—e.g., Self-Protectors bought more healthy products and explored less variety, while Social-Signalers bought fewer popular products and more hedonic products. We trained a multi-class tree-classification model on the sales data and achieved a high prediction accuracy of 84% on a hold-out sample. This research has managerial implications for customer segmentation and targeting as well as social implications for messaging to encourage public mask-wearing.