Selecting Profitable Customers for Complex Services on the Internet
研究了互联网上复杂服务（如抵押贷款、保险）的客户决策过程，利用潜在类别概率模型预测购买行为，帮助公司筛选有利可图的客户以提升利润。
In contrast to books and compact discs, the number of complex services offered on the Internet is still small. The decision-making process for complex services is different because it has an additional intermediate step of “indication of interest.” The Web site is (a) visited and searched for information; subsequently, (b) a request for the service is made, which may lead to (c) a purchase. The authors acquired a unique data set from an online Dutch financial service provider, which offers services such as mortgage loans and insurance on the Internet on behalf of financial institutions. They also obtained information on whether the request for the service resulted in a purchase. The authors used the available information to predict the purchase using a latent class probit model. A direct managerial application of this model is the ability to identify and select profitable applicants, resulting in significant profit improvements for the company.