Predicting likelihood to recommend and likelihood to churn based on cumulative experience of online services: modeling transitions in customer attitudes and behaviours

点击翻译按钮获取中文摘要

Customers want high quality, and high bandwidth services, at competitive pricing. In practice there are engineering tradeoffs between guaranteeing increasing levels of quality and the associated costs. What is the relationship between cumulative service quality and resulting customer satisfaction as reflected in attitudes and behaviors such as the net promoter score (or likelihood to recommend, L2R). We will carry out a series of experiments and observational studies where our goal will be to develop a family of models, ultimately for each family of wireless services, that predict key attitude scores and L2R based on patterns of quality of experience over time. Using controlled experiments and field and longitudinal studies we will investigate how cumulative experience translates into attitudes towards services based on the history of frustration and satisfaction experienced. We will develop models that predict the impact of cumulative experience in different contexts and will examine how changes in quality of experience lead to changes in attitudes, and finally decisions, that are reflected in L2R judgments (and discontinuation of service in some cases). An important part of the research will involve controlled experiments that can test hypotheses about causal relationships. Thus our strategy will be not only to develop models, but also to test their predictions in formal experiments, with analysis of variance carried out on the experimental results to identify causal factors that influence L2R. The outcomes of this research will also include guidelines for monitoring and managing service delivery so as to increase satisfaction and loyalty, and reduce frustration and churn on the other.