부산대학교 도서관

Many companies assess performance by using customer feedback surveys. A primary purpose for such surveys is to generate actionable insights to improve the company’s areas of weaknesses. Asking too few questions in the survey does not provide sufficient information for root cause analysis. On the other hand, one can also ask a series of more detailed questions which will force the customer to provide specific insights. However, customers are unlikely to complete such detailed surveys and many parts of the survey may be of little interest to them anyway. Suppose instead, we focus on requesting feedback for a small number of pain points of the customer using a hierarchical decision tree approach. By doing this we can provide a more focused, smaller set of questions for that individual, based on their branching choices. This would provide the level of detail needed to determine weak points in the company. We address this problem and outline a hierarchical decision tree approach for determining such a survey, also known as a computer adaptive survey (CAS). Computer adaptive surveys have been shown to be a great tool for root cause analysis. Our approach adds to CAS by providing a relative performance score for measuring customer experience across similar companies and over time. Furthermore, we ensure that responses to the survey are appropriately weighted in order to reduce bias. In addition, we describe the procedure of updating survey questions in the future based on historical survey responses and propose how incentives can be provided to increase participation in such a way that the customer benefits only by providing quality data. In this paper we illustrate the approach using data from the telecommunications industry and compare customer experience results for two different cellular providers.