The Draw of Data - Industry Trend or Event

ENT, Sept 6, 2000 by Robert Craig

Much of the industry buzz lately has revolved around the impact of the Web on e-business, the dot-com deathwatch, and related topics, but the data mining market has quietly been doing very nicely. One indication of this is the continued popularity of the numerous data mining conferences each year. Another is the maturation and consolidation of the marketplace, as evidenced by the acquisition of Thinking Machines by Oracle last year.

Data mining is the application of software algorithms to extract previously unknown information from data. In reality, there is no single data mining software algorithm. There is a wide range of data mining algorithms that are based on either statistical analysis -- CHAID, decision tree, affinity grouping -- or artificial intelligence techniques such as neural net.

A variety of applications are amenable to a data mining solution. For instance, data mining can be useful in customer relationship management (CRM) applications to monitor, analyze, and determine customer attributes. Data mining software can be used to figure out which customers buy, and why. This can significantly improve sales efficiency.

Database marketing is another application that benefits from data mining software. Direct mail costs about $1 per piece, delivered to the post office. If you send out 100,000 pieces and get a 2 percent response rate, you've wasted $98,000. The problem with direct mail is you know you're going to hear back from 2 percent of the people on your list -- you just don't which 2 percent. Database marketing allows you to identify the factors that drive a customer's buying behavior. With this knowledge, a marketing team can deploy campaigns that are specifically targeted at the right type of prospect. This has several benefits: It can improve the response rate of a campaign, identifying more leads with fewer mailings, and it's cheaper and more profitable to sell to existing customers.

Smart companies work hard to figure out what their Web site visitors are looking for by analyzing click-stream data. This data can be more helpful when it's combined with off-line data from market research firms. You can ask users to fill out a form and reveal their physical or e-mail address in exchange for something, such as a white paper. The garnered data can be combined with off-line demographic data from one of the many market research companies that collect personal data.

Analyzing the combined Web behavior and demographic data will dynamically enhance a Web site by leading to special offers to users that match the profile identified with a predefined model.

OK, I've convinced you. So what do you do next?

First, do your homework. Buy some data mining books and learn about the various techniques and where they are applicable. An excellent book is Data Mining Techniques for Marketing, Sales, and Customer Support, by Michael Berry and Gordon Linoff. Surf the Web. There are a variety of articles and white papers available at various sites.

Second, pick one or two applications that can be quickly implemented in your organization that will give you a quick hit. The ideal project should have a high return on investment, be deployable within six weeks or less, and have the ability to generate a significant top-line or bottom-line impact. Ask around the marketing or finance departments to find out what point of pain they're dealing with that can be solved by this kind of solution.

Third, write up a short set of product requirements and contact some of the leading vendors. Have them come in and show you how they address these requirements. Your step-one homework should help you decide which vendors to call and which ones you can leave off your list.

Fourth, go out and do it. Remember, as Grace Hopper once said, "It's easier to ask forgiveness than to ask permission." --Robert Craig is vice president of strategic marketing at Viador Inc. (Burlington, Mass.), and a former director at the Hurwitz Group Inc.