Pricing considerations for delivering e-content on-demand

Internet usage, available bandwidth, and the multimedia capabilities of end-user computers have increased tremendously over the last decade. These have enabled powerful ways to exchange data, and spawned new business ventures based on innovative technologies. Content providers can now sell multimedia-rich software and services using the Internet. Customers can download the e-content immediately after online transactions. Alternately, the content provider can stream the content to customers. For the success of such enterprises, content providers need to answer two important questions: (1) How to maximize revenue/profit? and (2) How to manage resources efficiently? In this thesis, we argue that both issues are addressed by choosing the “right” price. We illustrate how resource management is related to revenue maximization, and study the impact of price on system performance. We develop a theory to choose intelligent prices based on the available system resources and on models of customer behavior. We formulate revenue maximization problems for a number of service models, where the services differ in how system resources are utilized. We also construct a Finite State Machine to perform controlled price experiments. The data gathered from such experiments can be used to estimate approximation models of customer behavior in near real-time. The framework we develop is robust to changes in customer behavior. Based on our analysis of revenue and system utilization, and the framework to learn customer behavior, we develop dynamic pricing algorithms that not only generate high revenues but also reduce the number of requests denied service due to resource constraints. Using these pricing algorithms, content providers can develop services that generate more revenue while also being more compelling to users.