In this paper, we investigate a service system where heterogeneous customers arrive with varying assessments of service reward and unit waiting cost. Additionally, we allow for correlation between a customer’s valuations for service reward and unit waiting cost. We first model the service facility as an unobservable M/M/1 queue and demonstrate that the correlation between the previously mentioned utility attributes affects various critical performance metrics. If the expected net utility of the customers, when exactly half of the customers join the facility, is positive (negative), then the equilibrium arrival rate increases (decreases) with the correlation. We also show that the expected utility of arriving customers decreases with correlation. Finally, we examine the observability of queue length within our model and calculate regions of model parameters where disclosing queue length information benefits the customers and the service provider the least or the most.