Service firms (e.g., Amazon, Netflix) testify that recommender systems earn repeat purchases, reduce customer churn, and sharpen competitive edges. However, little is known about why customers will keep using a company|s recommender system. We conceptualise that customer perceptions of accuracy, benevolence and process value may affect sustained usage. We also explored the roles of preference stability and privacy concerns. We surveyed customers who had used recommender systems. Surprisingly, perceptions of benevolence and process value, rather than accuracy, drive sustained usage behaviour. Results fail to support the effects of preference stability or privacy concerns. Implications for research and businesses are discussed.