Neural Networks for Target Selection in Direct Marketing

INTRODUCTION Nowadays, large amounts of data are available to companies about their customers. This data can be used to establish and maintain a direct relationship with customers in order to target them individually for specific product offers and services from the company. Large databases of customer and market data are maintained for this purpose. The customers to be targeted in a specific campaign are selected from the database given different types of information, such as demographic information and information on the customer’s personal characteristics (e.g., profession, age and purchase history). Usually, the selected customers are contacted directly by mail promoting the new products or services. For this reason, this type of marketing is called direct marketing. Among others, a growing number of bank and insurance companies are adopting direct marketing as their main strategy for interacting with their customers. Apart from commercial firms and companies, charity organizations also apply direct marketing for fund raising. Charity organizations do not have customers in the regular sense of the word, but they must be able to trace people who are more likely to donate money in order to optimize their fund-raising results. The targeted individuals are then contacted by mail, preferentially in relation to other individuals in the database. Thus, direct marketing has become an important application field for data mining. In the commercial field, various techniques, such as statistical regression (Bult & Wansbeek, 1995), regression trees (Haughton & Oulabi, 1993), neural computing (Zahavi & Levin, 1997), fuzzy clustering (Setnes & Kaymak, 2001), and association rules (Pijls & Potharst, 2000) have been applied