Abstract: Due to the heterogeneity of customers'waiting cost service providers often adopt customer segmentation in queueing. However, the customer segmentation would produce the differernt of waiting time (W₂>W₁) and service value, which may cause the transfer or switch of customers between different queues in the system, so did affect the revenue, social welfare, and customers' utility. The first exact analysis of a service provider offering two classes of non-preemptive priority service is provided when customer perception of fairness is explicitly modeled. Customers differ in their waiting time cost and service value (R₁'>R₂') in different queues (regular and priority), but that information of waiting cost (H) is private to each customer. The service provider cannot schedule services so she offers a regular queue and a priority one-with additional charge-to induce customers to self-select into the two queues, in order to maximize revenue. The customers' fairness perception is modeled as a negative utility on the regular customer that's proportional to the waiting time difference between the two queues (α(W₂-W₁), α represents the fairness preference of regular customer). Analyzing a stylized non-preemptive M/M/1 queue system, important results are derived some of which reaffirms existing research results about the benefit of offering differentiated service and pricing to the service providers.
In this paper, it is first proposed that service providers should adopt customer segmentation, utilize two queues in the system and charge a concierge fee to obtain more revenue. Through intuitive perception and comparison between two queues, the fairness preference can be further stimulated. That is service providers should establish the service windows of the priority queue and regular queue in relatively close locations to more clearly present the advantages of shorter waiting times in priority services to all customers and to advertise the service experience associated with priority services in order to attract more customers jioning the priority and increase the priority fee used to customer segmentation. Then, service providers are suggested to adopt customer segmentation from the angle of social welfare maximization, which constructed by the revenue and utility of customers. In order to reduce the stimulation of the regular customer's fairness preferences, and minimize the negative psychological utility caused by the fairness, service providers should guarantee e the service windows of the priority queue and the regular queue far from each other. Later, it is found that the service providers should cancel the classification of customer service, and adopt the mechanism to keep one queue (regular customer only or providing customer segmentation but cancelling the priority fee then all the customers will join the priority queue) to get the maximization customers' utility. Finally, the properties of optimal results are producted, analyzed with the existed results in the above three different perspectives, then the extension research is proposed. This research shows important reference value and practical significance for service providers who adopt scientific mechanism of classification service and the corresponding optimal service pricing.

Key words: service operation, fairness preference, revenue, social welfare, utility, optimization, heterogeneity