Algorithmic Transparency in Recommendation Systems for Online Service Operations Platform

In online service platforms, economic inefficiency arises when customers are not fully aware of their preferences, since customers may choose an unattractive service among horizontally differentiated ones. With its expertise or data dominance, a platform can be more informed about customers' hidden preferences and in turn, provide service recommendations to customers. We focus on the effects of recommendations on customers' service choices and social welfare. We propose a Hotelling model wherein customers are sensitive to the delays for service while making Bayesian belief updates based on a platform's recommendations. When customers self-select their favorite service, their choices impose negative externalities through congestion, which poses a welfare gap towards ``the first best" assignment. Our results indicate that service recommendations navigate customers towards the more appropriate service, thus improving matching efficiency, reducing congestion cost, and enhancing aggregate customer welfare. We further identify the role of ``algorithmic transparency" and study how the platform should strategically release (partial) information by making personalized service recommendations to customers. Surprisingly, we identify the ``value of non-transparency'' when a customer-centric platform maximizes the aggregate customer welfare by strategically withholding service recommendations from a subset of customers. These customers turn out to be the most flexible and can correct the over-crowding service choices when set uninformed. Therefore, our managerial insights are relevant in guiding the design of recommendation algorithms.