A prediction model for the purchase probability of anonymous customers to support real time web marketing: a case study Euiho Suh a , Seungjae Lim b , Hyunseok Hwang c, * , Suyeon Kim d a Dept. of Industrial Engineering, POSTECH, Pohang, Kyungbuk 790 –784, Korea b LG PRC Center, 19–1 Cheongho-ri, Jinwuy-myun, Pyungtaek-si, Kyunggi 451–713 Korea c Dept. of Business Administration, Hallym University, chunchon, Kangwon 200–702, Korea d School of Computer Information Engineering, Daegu University, Gyeongsan, Kyungbuk 712–714, Korea Abstract The rapid growth of e-commerce has provided both an opportunity to create new values in the online marketplace and dramatic competition to survive. To survive in a competitive environment, Internet shopping malls attempt to adopt and use Customer Relationship Management. However, previous researches focused on navigation patterns of customers with membership. Therefore, they failed to apply real time web marketing to anonymous customers who navigate web pages without personal login. To overcome the problems noted above, we propose a methodology for predicting the purchase probability of anonymous customers to support real time web marketing. The proposed methodology is composed of two phases: (1) extracting purchase patterns and (2) predicting purchase probability. Purchase pattern provides marketing implications to web marketers while the purchase probability provides an opportunity for real time web marketing by predicting the purchase probability of an anonymous customer. The proposed methodology can be applied to the real time web marketing such as navigation shortcuts, product recommendations and better customer inducement since anonymous customers are included in marketing target and significant navigation pattern for purchase is identified. q 2004 Elsevier Ltd. All rights reserved. Keywords: Anonymous customers; Customer relationship management; Purchase probability; Web marketing 1. Introduction Internet technologies provide many competitive advan- tages such as agility, selectivity, individuality and inter- activity (Weiber & Kollmann, 1998). The Internet enables customers to search products and services meeting their needs with smaller time than before. The characteristics of the online marketplace to reduce search costs for products or services significantly affect the competition environment. Dramatic competition between Internet Companies has brought about profound changes in customer relation management. As the importance of the customer has increased rapidly in the Internet shopping mall, many dot- com companies are trying to apply web-marketing strategies that fit in the online retailer’s environment. There are many approaches related to customer relationship management (CRM) in the Internet shopping mall. Most previous researches focused on navigation patterns based on web log data and customers with membership based customer profile and purchasing information. These researches have weaknesses in web marketing. First, they paid less attention to surfing customers who do not login to websites such Internet shopping malls, and electronic auction sites. Although surfing customers are members of the Internet shopping mall, most of them maintain anonymity before purchasing and ordering products. Since previous researches uses profile and purchasing information of the customer with membership, they cannot apply marketing actions to anonymous customers. Second, they missed the opportu- nity that applies real time marketing to the surfing customer. Previous researches such as the RFM approach and clustering applied marketing actions after customers leave an Internet shopping mall. One way to overcome the above problems is to offer marketing activity according to the purchase probability and customer behavior using web log data. In an electronic commerce environment, analyzing web log data provides 0957-4174/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2004.01.008 Expert Systems with Applications 27 (2004) 245–255 www.elsevier.com/locate/eswa * Corresponding author. Tel.: þ 82-54-279-5680; fax: þ82-54-279-2694. E-mail addresses: ieman1972@hotmail.com (H. Hwang), ehsuh@ postech.ac.kr (E. Suh), sjlim@postech.ac.kr (S. Lim), my1052@hotmail. com (S. Kim).