OR/MS Today - December 2002 - Was it Something I Said?

OR/MS Today - December 2002

Was it Something I Said?

The Trouble With Data

By Vijay Mehrotra

Item: In the aftermath of Sept. 11, many OR/MS people were eager to help. Surely our skills and experience would make us welcome allies in any response to this national tragedy. Please, put us to work to help our country, be it a Manhattan Project-like initiative to move the country towards energy independence (called for by New York Times columnist Thomas Friedman), an analysis of the risks inherent in airport security (see Arnold Barnett's article "The Worst Day Ever," OR/MS Today, December 2001), or a study of migration patterns across U.S. border crossing points.

Well, the energy independence program has got "legs" in Washington, while airport security will be a hot area for some time to come. However, analyzing who is coming and going into our country — and when and where — remains a huge challenge for one major reason: data collection. It turns out there are all sorts of fragmented data out there about visas and immigration, but none of it is complete, cleansed or very useful for answering questions that are suddenly of interest.

Item: Consider Siebel Systems, who is the "elephant" in the customer relationship management (CRM) software market. In theory, CRM software helps people like the INS by supplying powerful information systems for data capture and analysis.

I am intrigued by both Siebel and the CRM market. Its message is clear and powerful: "our software will deliver major improvements to your business performance."

However, a study conducted earlier this year by Nucleus Research suggests that this message is mostly hype. The upshot is that deploying a complex system is, well, complex; even after two years most companies participating in the survey had yet to see the return on their very substantial investments in CRM technology. You can check out the details at www.nucleusresearch.com/prsiebel.htm.

The problem isn't just Siebel, or even just CRM. One aspect of this economic slowdown has been the decrease in business IT spending, which has hit tech companies hard. The tech-heavy NASDAQ now sits under 1500, or less than one-third of its peak value. Most vendors face an ROI credibility problem, and it shows in their financial results.

Item: A colleague recently described her customer's environment: "They gather all kinds of data, but when you scratch the surface at all you find out that the people entering the data don't pay any real attention. So they end up with some decisions based on bogus data, but mostly just management by 'gut feel' because no one there believes in their data at all." Aaaargh!

When I started consulting full-time 10 years ago, I naively thought that data itself was the least of our problems. I was wrong. Here's some stuff I've learned about the OR/MS version of "Quest for Fire":

1. Follow the money. For purposes of billing, collections and customer service, most organizations keep careful track of anything that involves money changing hands. With the advent of client-server computing and data warehousing, this type of data is the most accessible and reliable that you will find. Figure out what you need to do to get it.

2. Operations before optimization. Most data is captured and stored to enable operations to run, with analysis a distant second. Thus, the data needed to support your model may not be captured at all. Why not? Because the people who can get it are trying to close deals, finish phone calls and process as many transactions as possible; the orderly capture of data is not part of how they get compensated. Big problem.

Beyond that, because databases are set up to support "real-time" workflow, data is often not stored at the level of detail needed. Pet peeve: e-mail routing systems that cannot provide accurate information about time spent processing individual records, which is critical for forecasting workload and scheduling people to handle it.

3. Here's what I need and why. In the end, you often have to lobby to have specific information captured, and this will require you to face several audiences. The executives are busy with their own problems and have no idea why they need to support your project. The actual "users" are invariably frustrated before you show up to ask them to do more data entry. The database administrators get beeped at midnight with technical problems, and expect that you will only cause them to get called at home more often.

So you have to help them get there. Help executives understand the payback for what you are trying to do, and make sure that they share in the credit. Explain to users how your work will make their lives better (this can require some creativity). Work with the IT staff through their data needs and options; offer to help them actually get the solution built.

Bottom line: Every organization faces serious headcount pressures these days. An analyst who sits back and admires the data-collection problem without being an active part of its solution is an endangered species.

Hey, I know the models are the fun part, and that actually solving real, interesting problems is why most of us got into this. But just like we can't ignore the needs of our downstream customers, we also can't neglect the challenges of our upstream providers, or we'll end up with supply chain problems of our own.

Vijay Mehrotra is vice president of the Solutions Group at Blue Pumpkin Software. He can be reached by e-mail at Vijay@BluePumpkin.com

Table of Contents

OR/MS Today Home Page

Lionheart Publishing, Inc.
506 Roswell Rd., Suite 220, Marietta, GA 30060 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com