Towards a Model for Likelihood to Recommend (L2R) based on Accumulated Experience with an Online Service

As online services become more pervasive and complex, customer experience is becoming a key competitive differentiator. Customer expectations are rising, and bandwidth requirements of online services are increasing. Internet service providers are faced with a difficult tradeoff between managing the costs of services and bandwidth requirements on the one hand, and meeting customer expectations on the other. At present there is little to guide service providers in managing the human side of this tradeoff. While there is considerable interest in quality of experience from an engineering perspective, there is a need to supplement that work with research based on a human factors approach. The goal of the research reported here is to determine how a series of satisfying experiences and frustrating events lead to longer term attitudes of satisfaction or dissatisfaction over time and, ultimately, to loyalty towards a service provider or Likelihood to Recommend (L2R). The ultimate objective of the research will be to develop models that can predict overall satisfaction, and likelihood to churn, based on network probe data (data relating to online service impairments and failures experienced by a customer), service context and customer profile. These models can then be used to optimize services design and delivery, leading to more satisfied consumers and more profitable companies. In this report we introduce the initial outlines of a model of likelihood to recommend for online services. The ultimate goal is to develop a detailed model based on data that is either available from marketing sources or that can be collected automatically from network probes.