Service delivery and learning in automated interfaces

This dissertation analyzes the strategic implications of customization policies available to companies that must simultaneously provide service and learn about their customers through automated interfaces. The first part of the dissertation lays out the theoretical framework within which the analysis is carried. The second part addresses whether companies should use Internet-based customization tools to design service encounters that maximize customers' utility in the present or explore customers' tastes to provide more value in the future. Good customization policies must quantify the value of knowledge so as to adequately balance the expected revenue of present and future interactions. Such policies can be obtained by analyzing the customization decision problem within the framework of dynamic programming. Interpretation of the service design policies enhances the current understanding of the mechanisms connecting service customization, value creation, and customer lifetime value. This leads to insights into the nature of the relationship between learning, loyalty, and long-term profitability in service industries. The final part of the dissertation considers situations where companies have the ability to acquire information by other means in addition to observing interactions with customers. In information-intensive industries, investments in customer retention often take the form of paying customers to answer questionnaires, or somehow acquiring information about the customers' preferences. The value of customers is convex as a function of knowledge.