Computerworld Senior Editor Lee Copeland recently spoke with James H. Goodnight, president and CEO of SAS Institute in Cary, North Carolina, about using data-modeling tools and how best to use valuable corporate data.
Q: What is the single most important thing that information technology leaders can do to ensure their companies are in the best possible competitive position?
A: One of the biggest things is in data warehousing, establishing the structure and teaching people how to use it. IT departments can manage the actual reporting function, and then move that function out to the different departments and divisions within the organisation. Whereas the old model was to have an MIS group inside IT and anytime anyone needed a report they would call MIS and ask for a report. That's why we see MIS departments six months to one year in arrears and backed up with work to do, because they are the adjunct report generators for the whole company. The data warehousing philosophy we use sets up IT or MIS to do nightly extractions from our operational systems and moves [them] over to Unix. [There,] multidimensional databases and summary bases are created and set up online for end users to access with all the tools that are available for decision support on a daily basis.
Q: What's the best way to get started?
A: The vast majority of clients have started small with a division and provided them with the data and reporting tools that they needed. We've seen the failure of data warehousing with a two-year data-modeling approach where the roll-outs are gargantuan. It's best to work with each individual department at a time.
Q: What's the biggest problem area that your customers run into when they implement data-modeling tools?
A: When they try to do data modeling for the entire company and discover that they have different data tables out there. It's such an enormous way to do it. You can have success . . . if you start small and satisfy the needs of one group after another.
Q: Starting small sounds good, but we hear from IT shops that they do start small -- and then find that ways that they can use that data for decision support are even smaller.
A: This has been a major problem with all the ERP vendors. They are really in the plumbing business -- making sure all the internal workings of the company work. We have the analytical tools and all the graphics visualization tools that these companies haven't really thought about.
Q: It sounds obvious that ERP systems provide internal operational data, but a lot of companies were blindsided because they thought they would be able to use this data automatically for decision support and reporting. Have you found that a lot of your clients were surprised that their ERP data could not support decision making as they thought it would?
A: There were a lot of what we call "shiny shoes" people selling ERP. There's no doubt about it, because they went into to the highest level, even at the CEO level, and convinced the CEO. And quite often the CEOs made the decision without getting IT's input. The CEOs were tired of not getting anything out of IT and decided to go around them. There's no real euphoria out there because of the huge amount of code, there are huge support problems. What I like to refer to current ERP systems [as] is new legacy data. The nice thing is that we can read it all.
Q: It's got to be hard for customers to spend millions to implement these systems, then they have to go back to their chief financial officer and ask for more money to get the reports they originally wanted.
A: That's a competitive advantage to us, because a lot of companies have been burned when they put in their ERP systems. We have some of the Big Six, when we ask them to help out on an account, they say, "We can't get back there!" Because they were the ERP implementers and are no longer welcome in that shop. A lot of money has been spent and some good has come out of it.
Q: How do you think IT leaders can make an impact in the next part of the decision process and make meaningful use of the data that they invested so heavily in getting?
A: The IT shops have access to all the data available in a large company. If they want, they can take it upon themselves to do some data mining. Then go to marketing and say here are some facts about the customers that are not likely to buy from us again or say this is the profile of our best customers. IT can start initiatives like that, working with the company data to try to do better forecasting and modeling.