COMMENTS AND COMMENTARY FROM THE ROAD
by John K. Thompson

The comments that I will paraphrase come from attending seminars that have been vertically focused on specific industries; Magnify & Computerworld sponsored seminars, and the DCI Data Warehouse conference in San Francisco. Trade shows and seminars take up a significant portion of my time, more than I'd like to admit, actually. A few years ago I noticed that there long spells of boredom punctuated by brief flurries of interest and discussion with a wide variety of people. From these flurries, a significant and useful amount of informal survey data can be collected. The collating can be done mentally during those periods of idle time when I am mostly concerned with when I might be able to rest my aching feet for a moment or two. The top 5 themes that are discussed with a reasonable amount of frequency are:

1. The market for data mining tools has been dramatically discounted in the past six months. Not only discounted from the standpoint of price, but also in viability as an approach to building a solution in mainstream corporations. This is evidenced by the fact that the only people willing to talk about the base level technology are consultants from small boutique firms, graduate students, or garage based embryonic software companies looking for a place to license their algorithms. Bring up data mining toolkits to corporate IT or LOB professionals and invariably the answer is: Sounds interesting and we will consider it, but it would be easier for us to manage the installation of a complete application that your professional services staff implements and tailors to our needs. Do you have a complete application for...(name your favorite application here)?

2. The discussion of "complete applications" or "solutions" for specific vertical industries and business problems within those industries is racing ahead of reality. Nothing new here. Once the vendor community and press picked up on the slow adoption of data mining tools, the vendors moved quickly to announce their new vertically focused approach to building applications on top of their toolkits. Vendors have staked their claims to industries such as retail, insurance, and financial services, and delineated the applications that they will offer, but very few are being offered. When LOB professionals were queried as to which applications are they evaluating, the responses vary from; I didn't know I could buy a complete application for auto claims fraud. Who is selling the application? to We looked at a couple vendor applications, and the applications were not complete enough for us to consider an implementation. Inquiring to various vendors revealed that their plans for releasing applications were to have release 1.0 ready in late 1998.

3. Corporate IT and LOB professionals are still gathering data and looking for education on what data mining is, how they might use it as a technology and which applications make sense for them to implement. Magnify and Computerworld have sponsored a series of six seminars around the United States. Four have been completed to date and have drawn over 250 people. The program features leading analysts from the META Group and International Data Corp as well as renowned authors and practitioners in the field of data mining. The program is educational and free. In asking multiple attendees in each city, "Would you have attended the seminar if you were required to pay a fee?" I varied the amount of money when asking the question. The answer was always no.

4. In evaluating and considering a data-mining vendor, partnerships and associations make a difference to potential buyers. At the DCI Data Warehouse show in San Francisco, Magnify was invited to attend the show as a guest of Sun Microsystems, inc. During the two days that I was on the floor, I routinely asked people, "Would you consider Magnify as a vendor if we were exhibiting as a stand alone vendor?" and in come cases I added the clause, "or does it help in your decision making process that Magnify is in the Sun booth?" The answers ranged from polite to blunt but in the aggregate the answer was loud and clear. By being affiliated with Sun, Magnify was considered to be a much more viable candidate for their business. Maybe that's why no data mining vendors had their own booths and all other data mining vendors were presenting in the Oracle booth.

5. The value of data mining is widely recognized, but how to achieve that value on a large scale is not widely known. The success stories are widely known and provide anecdotal evidence of the significant value that can be achieved, but those stories are few and oft repeated, repeated too many times. The talent pool is stretched by Y2K, ERP, and data warehousing projects. Where will data mining fit in? Data mining will fit in when the vendors realize that the technology must fit into existing systems. Not only fit into existing decision support environments, but also into production systems. Data mining consists of advanced technology that must be sufficiently surrounded with ease of use functionality, and embedded into known processing flows. Valuable information is produced by data mining systems, but is useless by itself. So, until the vendors realize and begin to work on making data mining fit in where it can make a difference, it will be a long slow road to adoption.

In summary, the future for data mining technology is bright. The clients are working hard to determine how best they can utilize these advances in algorithms, data management, and modeling. The vendors need to determine how best to provide the advances in a suitable package. The journey is the fun part.

---

John Thompson, Vice President - Marketing, Magnify, Inc. I'd like to hear your thoughts. You can reach me at jkt@magnify.com