Abstract
Customer service organizations increasingly rely on newly available data sources and algorithms to inform managerial practices, potentially altering frontline service interactions. Drawing on qualitative and quantitative data from two original surveys of employees from 17 foodservice and retail companies, as well as computational text analysis of 2 million Yelp reviews, I provide evidence linking quantification with increased customer-originating mistreatment, including racism, and sexism. Qualitative text analysis shows that quantified work outputs, such as item-scanning speed, can lead employees to feel as though they appear robotic in the eyes of customers, whereas the quantification of work inputs, such as adjustments to the supply of employees based on up-to-the-minute sales data can leave employees feeling they appear incompetent to customers. In a quantitative analysis, I show that both processes are associated with higher levels of customer mistreatment, inhibiting the production of a psychologically safe climate. However, only organizational processes that lead employees to appear incompetent are associated with employees receiving higher rates of sexist and racist remarks. In a computational text analysis of Yelp reviews I explain this finding, demonstrating that the appearance of employee incompetence plays into consumer’s deep-rooted stereotypes that women and workers of color are less capable than their white and male counterparts. Together these findings demonstrate how frontline workers absorb, in likely unmeasured ways, the negative impacts of operational friction produced by algorithmic decision-making in the form of mistreatment, including racism and sexism.