In this paper, we consider a feedback queue with one server providing customerintensive services. The service reward increases with service time, while longer service time also leads to longer waiting time for customers. Customers are boundedly rational and decide whether to join the system based on a logit model upon their arrival. From the perspective of revenue maximization and social welfare maximization, this paper considers the trade-off between speed and quality of service and studies the optimal service rate, service price and some system performances. Under different strategies, we study the ratio of the corresponding welfare, called the “price of anarchy” (PoA), which measures the harm of service providers’ selfish behavior, and we obtain the interval of PoA and its characteristics. In addition, we conduct some numerical simulations to show the quantitative results of system operation.