DECISION TECHNOLOGIES IN DATABASE MARKETING: PART II
by Gene M. Ferruzza, Senior VP, Decision Technologies
FUNDAMENTALS OF DATABASE MARKETING: DECISION TECHNOLOGIES
Conceptually, decision processing in database marketing consists of three components that operate with the data: the data mart, data mining, and the decision system. It is useful to refer to these components of decision processing as "decision technologies." To be used successfully, each of these decision technologies requires not only hardware and software systems, but also processes, methodologies, and expertise. These components relate to each other and to database marketing activities. Because these components must be well integrated, we cannot explore the scope of data mining without first exploring the roles of data marts and decision systems in database marketing.
The Data Mart
In database marketing, the data mart is the centralized repository for customer data. It may contain data from any number of sources identified across the corporate enterprise. Typical data sources for the data mart are systems associated with accounting, product and service purchase and usage, and customer service, as well as marketing response information. Data marts also commonly contain third-party information, such as demographic data, psychographic data, and credit history. Usually, these data are collected for individual customers. Static customer data also are kept, including name, address, acquisition date, and dates of purchases.
The data in a marketing data mart accumulate rapidly. To keep the data mart at a manageable size, dynamic data usually are summarized or aggregated, most often on a monthly basis. For marketing purposes, approximately 36 months of dynamic data need to be readily accessible by users; in some applications, a longer time horizon may be required.
Many processes are used to develop and maintain the data mart. These may include record de-duplication, customer- and household-level matching, address correction, data appending, and data imputation, among others. (These processes are explained further under "Data Mining for Data Marts.") The data mart enables business users to access a complete data profile for each customer, prospect, or former customer.
The data mart is the key data repository for data-mining operations, usually providing a more complete customer data profile than is available from any of the company's other customer databases. The data mart is the primary source of data used by decision systems in making marketing decisions. The data mart also collects new data on customer responses, which then are used by the decision system to evaluate database marketing programs.
The Decision System
In database marketing, a decision system usually is referred to as a "campaign manager." Campaign management software facilitates decision-making processes involving large groups of customers. A campaign manager should have two primary functions. First, it should pull together and use data from the data mart, intelligence from data-mining processes, and strategies from domain experts (marketing users) to execute decisions in an automated manner. These decisions may be executed in batch mode (e.g., deciding which customers will receive a retention mailing) or in real time (e.g., adjusting the content of a Web site to fit the profile of the customer, or providing information to allow instant credit approval). Second, the campaign manager should be self-evaluating: it should monitor and evaluate the success or failure of its past decisions.
Tracking and evaluating the results of decisions is essential for progressively more successful decision- making. A basic concept in database marketing is the use of data to continuously improve marketing results. However, this is possible only if all aspects of database marketing programs are monitored. Monitoring involves tracking customer data over time, identifying each customer's response to each communication, evaluating the effectiveness of strategies that drive the communications, and evaluating the data-mining results used to develop the strategies. Accurate measurements of each component incorporated into the campaign manager allows for adjustments to the program.
We can measure most aspects of any database marketing program. A typical scenario might involve measuring the reaction of the customer base to a marketing communication. The campaign manager will assign a communication or intervention to a particular group of customers. Through monitoring of these customers' activity, the communication can be assigned a response rate, a profitability value, an attrition rate, and an overall return on investment. The next step is to compare these results with the results of other communications. This information empowers the marketer to make an informed choice about where to make future investments in customer communications.
Strategic Integration
I often refer to "strategic integration" as the key to success in executing database-marketing programs. Information moves from one process to the next, and new data continually are collected from customers and stored in the data mart, providing feedback for evaluation of marketing results by the campaign manager. Many data-mining applications are used to provide an understanding of the effectiveness and profitability of the database marketing program and to guide adjustments to the program.
From this brief discussion, it should be clear how dependent each of the three decision technologies is on the others. Without a campaign manager, the results of data mining and information from the data mart cannot be used to their full potential. Without data mining, it is unlikely that the data mart will be developed properly. The data will not be leveraged for optimal decision-making, and the quality and accuracy of strategic decisions will be compromised. Without the data mart, of course, strategic marketing decisions would not be possible.
Part III of this series will be published in the next edition of D S * .
Contact Gene Ferruzza at gmf@cmsnet.com