|
Abstract:
|
On a competitive market, customers may buy a product/service from one or more providing companies. To any particular company, a customer's total spending, called the customer's wallet, is of great interest, because it provides the focal company an opportunity to see to what extent, their revenue can grow, and it also serves other purpose, such as evaluation of marketing strategies. For most products/services, a focal company only knows a customer's spending with itself. There are different approaches to accessing a customer wallet, such as a survey, sales-persons' self-reporting of their assessment, and statistical modeling. Each approach has its advantages and disadvantages. We will consider building a statistical model using a focal company's customers observed spending in which some empirical Bayes inference can be done. Simulation is used to assess the proposed methods.
|
