Value of Learning and Acting On Customer Information

The rapid growth in demand and supply of sophisticated data mining and analytical decision tools calls for research to understand the value of learning, as well as how learning interacts with firms’ day-to-day marketing strategies. In this paper, we consider a market in which the firm can reduce its service cost when it matches the right product with the right customer. Facing uncertainty about customer types, the firm can gradually learn using observed service costs realized from recent interactions as noisy signals. On the basis of the most updated information, the firm makes matching and selection decisions to maximize its long-term profit. By solving the closed form solution to the fully dynamic optimization problem with infinite horizon, we analytically investigate the dynamic and endogenous nature of learning processes, the interaction between learning and decision making, and the evolution of profit over time. We also examine how optimal decision paths, market size in steady states and the evolution of profit are affected by parameters such as discount rate and the precision of information. Extending the model to a setting in which the precision of signals is proportional to the units of goods, we study how firms can endogenize the speed of learning. Our results shed light on the value of learning and acting on information, the time point to discontinue service to a customer, the duration and amount of short-term financial losses before learning pays off, and how the firm can modify its decisions to speed up the learning process.