Managing service systems with unknown quality and customer anecdotal reasoning

摘要

In queuing theory, it is assumed that customers choose between waiting for service in a queue or balking based on their inference of the service quality. However, in real life, customers may lack such knowledge because of scarce learning opportunities. Thus, the Bayesian rule does not hold good in uncertain situations and customers resort to simplified heuristics when estimating service quality which is termed as anecdotal reasoning. While anecdotal reasoning simplifies customer inferences of service quality, it may lead them to hold incorrect estimates when service quality is intrinsically uncertain. The article studies how the service provider should adjust price and service quality as customers gather more anecdotes and whether the service provider discloses service quality information. This was examined using an M/M/1 unobservable queue with uncertain service quality. Using an anecdotal reasoning framework, the customer equilibrium joining rate, the service provider pricing, service quality and information disclosure decisions were characterized. It was found that unlike the fully rational benchmark, the service provider uses pricing as a tool to segment customers with different service quality estimates. In particular, a low-quality service provider prices higher than the fully rational benchmark to target the niche customers who considerably overestimate service quality. A high-quality service provider prices lower to target the mass customers who do not considerably underestimate service quality. As customers gather more anecdotes, the optimal revenue may first decrease and then increase, while social welfare always increases. Consumer surplus may decrease due to intensified congestion. Apart from pricing, in practice service providers may also have the discretion to inform customers of the mean service quality. This information is characterized and found that high-quality service providers may not disclose information if customers are sufficiently boundedly rational, while low-quality service providers may disclose if customers are rational enough. Higher congestion cost makes information non-disclosure more attractive.