BY LOOKING BACKWARD WE CAN SEE THE FUTURE
by John K. Thompson

To paraphrase one of the more frequently quoted futurists, Watts Wacker of First Matter, to be a good futurist you must be a better historian. I don't claim to be a futurist, but I do an excellent job of watching the current landscape and researching the historical events that are precursors of today's events.

What are today's events? In the data mining market, in my opinion, the weakest players out of the 100 odd firms in the market are already dead and don't know it. How many does that knock out of the race? Slightly over 50%. If you are looking to purchase an application based on data mining technology, make certain that you investigate the customer satisfaction track record and the financial viability of the vendor. That still leaves over 40 firms to choose from. Quite a lot really.

The winners in the market have moved well beyond technology. The market has voted from a customer adoption perspective as well as an exposure perspective to showcase the firms that "get it". By getting it, I mean that these firms have torn themselves from their love affair with their lovingly crafted base level technology, to a benefits and customer orientation. Too many data mining firms are obsessed with product and technology. The focus on product features, and what the competition is including or not including in the next release has completely obscured the true goal of solving problems that are affecting how potential clients operate their businesses. The firms that get it are now bringing applications to the marketplace.

In the data mining market the use of the terms applications and solutions is fast and loose. When vendors like SAP, and PeopleSoft say they have an application, I can see, touch, and have them explain to me exactly what their applications accomplish for Human Resources, Manufacturing, Supply Chain Management, etc. When a data-mining vendor tells me that they have an application to manage or reduce customer churn in telecommunications, I have to ask more questions and dig deeper to find the boundaries of this application. This is due to the fact that the majority of data mining vendors, from the executive staff to administration and all levels in between, have never seen or operated a live application in a client environment.

Now you ask how many vendors of the 40 get it? Not to be smug, but that's the wrong question. The more accurate and insightful question is, Of the approximately 40 vendors that get it, how many grouped by vertical industry/application are viable and provide enough functionality to be a true benefit above and beyond building the application or functionality using in-house development talent? That is a long question, but one that hits at the heart of the issue.

The heart of the issue is: Has the vendor built enough of the generic functionality into the application so that upon installation the client's competitive advantage is significantly improved even before the customization work begins? The majority of vendors are focused primarily on the following industries: telecommunications, insurance, financial services, retail, and Web traffic. The list is in descending order of the number of firms focused on the industry.

To answer the question, there are approximately 4 to 6 vendors leading the pack in pursuing telecommunications prospects, typically focused on customer churn, and there are approximately 2 to 4 firms in all of the other verticals pursuing a variety of applications. The high end those numbers add up to only 22. These counts reflect what I consider firms that are leading the pack. The remaining number is comprised of firms that are either just starting, and yes, there are new firms entering the marketplace, or firms that have substantial funding, but are lacking focus, or firms that have all the elements to be successful, but have not assembled the pieces in the right order.

The marginal firms will retreat from the commercial marketplace back into focus on consulting, governmental contracting, go out of business, or be acquired by other players. This is difficult time for prospects evaluating data mining technology and data-mining-based applications or solutions.

My advice to potential buyers:

1. Go slow. Evaluate the quality of the vendor organization first.
   1. Ask for references.
   2. Talk to the analysts such as Meta Group, Gartner Group, and Giga.
   3. Ask about alliances and partnerships. Quality data mining firms will have on-going and substantive partnerships with big platforms players like Sun, Hewlett-Packard, and now Compaq.
   4. Ask to have the Development plan explained to you. You need to know how this application and the technology will be enhanced during the lifetime of the implementation.
   5. Ask to see the profiles of the consultants that will be working with your staff. Write into the contract that the senior people work on your project.

2. Evaluate the base level technology. Not so much to understand the bits and bytes, but more so to understand the foundation that the application is built upon.

3. Evaluate the breadth and depth of the application. Find out exactly how this software will fit into your processes and systems architecture. Don't let the Sales team give you high level explanations, find out how the existing systems will interface with the new application in detail.

4. Focus on business benefits and Return on Investment. Make the vendor talk in these terms. If they constantly fall back on technobabble about algorithms and proprietary tools, you have a problem.

5. Inquire and delve into the on-going maintenance and how the system will evolve as your business changes. Will the models be updated in an automated fashion? Can you and your staff update the models? How often does the update need to be done? Does each update of the underlying models require a consultant from the vendor or one of their partners? How much will each update cost? Don't settle for futures or soft assurances. Get hard numbers and hold the vendor to their claims.

6. Make certain that you understand whether this application or data mining technology will be embedded into an existing system or whether this will be a stand alone system that is interfaced to other existing systems via data feeds. If it is via data feeds, how automated can those feeds be made?

7. Clearly define who are the primary users from a technical, user, and business executive standpoints.

This list could easy go on for many more points, but if you get through this short list and you still feel good about the vendor, their stated direction, the functionality of the application, and the fit in your organization, then you've got a winner.

---

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