Internet-based marketing strategies generate extremely large data sets from customer interactions. Purchase histories, financial records, customer service records, and Web site usage are just some of the data that reside in customer databases. In order to transform this mountain of diverse data into operationally useful information, marketers are increasingly using data mining procedures. Data mining is the computerbased exploration and analysis of large quantities of data in order to discover meaningful patterns and rules for the purpose of improving marketing, sales, and customer support operations. The combination of data mining procedures with data warehousing enables the MDSS to move beyond just support for the operational processes in the marketing organization and to focus on actual customer behavior. Data mining and data warehousing provide the means and the infrastructure for extracting strategic opportunity from knowledge of the customer.
a. The Data Mining Process
Large, multinational organizations produce much more marketing data per day than its managers can assimilate. The Internet facilitates the rapid growth of data on a worldwide basis. However, exponential growth of data can, paradoxically, lead to a situation where more data leads to less information as managers become swamped by the flood of data that defies ready interpretation. Marketers need to develop procedures for processing, filtering, and interpreting this data for strategic marketing purposes. Data mining is essentially the engine for a knowledge-based marketing strategy. It provides the ability to collect, process, disseminate, and act upon information more rapidly than the competition which is essential for the creation of first-mover advantage.
The first step in the process is to collect data on what the customer does. On-line transaction processing (OLTP) systems do precisely that. Virtually everything a customer does when purchasing a product or service generates a string of transaction records. If the customer calls an “800” number to order a product, the phone company will capture data on the time of the call, the number dialed, and the duration of the call. The marketing company will generate similar data in addition to that on products and services purchased, catalog referenced, special offers, credit card number, order size, and time since last purchase. Further transactions are generated by the order entry, billing, and shipping systems. The bank and the shipping company will log further transactions. The customer may need to call customer service to solve postpurchase problems. Internet transactions can generate even more data as the customer’s purchase behavior can be linked to Web-browsing behavior within a site and throughout the Web. This data can then be linked to purchase histories, financial history, and other personal-identity information.