Estimating Potential Demand and Customers' Perception of Service Value in a Two-station Service System

TL;DR

This paper develops a method to estimate demand and customer perceptions in a two-station service system using observed workloads, despite unobservable customer behaviors and unknown switching costs.

Contribution

It introduces a maximum likelihood estimation approach to infer arrival rates and perceived service value from workload data in complex customer choice scenarios.

Findings

Estimator accurately recovers demand and perception parameters in simulations.

Method effectively handles unobservable balking and switching behaviors.

Provides a practical tool for operational decision-making in service systems.

Abstract

The potential demand in the market and customers' perception of service value are crucial factors in pricing strategies, resource allocation, and other operational decisions. However, this information is typically private and not readily accessible. In this paper, we analyze a service system operating across two stations, each with its own customer flow. Customers arriving at the system are informed of the waiting times at both stations and can choose to either join the local station, switch to the other station, or balk. Our objective is to estimate the arrival rates at each station and customers' perceived service value based on the observed workloads at both stations. A significant challenge arises from the inability to observe balking customers and the lack of distinction between local arrivals and customers switching from the other station, as the switching cost is unknown. To address this issue, we employ maximum likelihood estimation and validate the effectiveness of the estimator through a series of simulations.