Paul: So, in other words ...
Jamie: You were right.
Paul: But more importantly ...
Jamie: I was wrong.
Paul: There you go.

Some of you may recognize this exchange from the "Togetherness" episode of NBC's situation comedy "Mad About You."

In situations where we have a lot of personal credibility at stake, there's something about being right and wrong that drives us to not be as open to other alternatives as we could be. For instance, for at least 10 years, I've been standing at podiums telling audiences the standard sequence for developing business intelligence (BI) capability:

  1. Identify business pain.
  2. Build an incremental development of unique, politically meaningful, integrated data.
  3. Provide information distribution, access and analysis of the data.
  4. Publicize and market success relentlessly.
  5. Repeat.

Of course, in this five-step route to success, there is endless debate about the particular design of the system, especially around the architecture of data structures and the technical components of the system. There's no counting the thousands of words invested in speeches, seminars, classes, workshops, exchanges on the various industry forums, magazine articles and columns, and books on these topics.
Interestingly, there's been comparatively little discussion and debate about the order in which you deliver BI capability to the business community. I've generated discussion around building a high-end analytical system for a small user base versus bringing information delivery to a wide user base. Others have discussed the various strengths and weaknesses of the leading query, reporting and OLAP products and suites. Even so, the basic conventional wisdom remains that you start with some basic information delivery capabilities and grow from there, based on identified user requirements and needs.

What if you started with the most advanced form of information analysis – data mining ­ and delivered it as the very first thing the user community sees?

Heresy, you say? That's been my reaction in the past as well.

I've long counseled audiences, students and clients to save their longings for data mining for a later date. "Crawl before you walk," I've said. "Data mining is sprinting. Actually, it's more like driving a Ferrari."

I've told countless thousands of audience members to make data mining last. Delay and defer it until the teams have demonstrated competency and proven their value to the organization. Concentrate on delivering some basic information capability first, and save the fancy stuff with the guys in lab coats for later. I, along with my peers, have given consistent voice to this orthodox view.

Then, along comes a group of rebels to reorder our world. While judging an industry award recently, I was introduced to a BI team from a financial services company in Texas. They had won an industry award for the best data mining implementation. While their results and impact on the business were exemplary and their implementation of the tools and technologies involved was indeed worthy of their award, what I found most striking was that their data mining implementation was the very first deliverable from their BI system.

They hadn't provided thin-client reporting. They hadn't invested time, money and resources into an OLAP implementation. They hadn't built and maintained a semantic layer for a query and reporting tool. They'd gone directly for the jugular. They'd delivered a very sophisticated data mining system as their first iteration.

I was stunned. Not only had they broken all the rules, ignored the conventional wisdom and thrown tomatoes at the orthodox view promulgated by industry thought leaders, they had managed to pull it off in grand style. They didn't just stop at saving their business millions of dollars annually through this capability; they had the audacity to take the show on the road and win an industry award with it! What was the world coming to?

What did it take them to be successful? First, they had advanced statistical capability and expertise on their team. Second, they focused on a very politically important, focused problem set that they could address in a politically viable time frame. Third, they matched the scope of their initiative with their available resources. Fourth, they used tools and technologies that they could understand, implement and sustain with their limited resources. In summary, they developed a targeted, specific, sustainable solution to a specific business problem.

I later asked Herb Edelstein, DM Review columnist and the dean of data mining, if this team's success was emblematic of what was possible, or a freak of nature. Herb replied enthusiastically that with today's tools and a focused team, this was the tip of the iceberg as to what was possible. He even reinforced the point that you don't need a data warehouse to do data mining. What indeed was the world coming to?

In the end, I was forced to admit I'd spent the last 10 years telling people something that was, well, in the words of Paul and Jamie:

Paul: I was wrrrrr ...
Jamie: You were what?
Paul: I was wrrrrrrr ...
Jamie: You were what?
Paul: I was, I was, I was not so right.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access