With Andrew Kellett, a Butler Group [www.butlergroup.com] analyst.
Question: Your research found that many companies are taking a departmental, rather than an enterprise-wide, approach to business intelligence. Why is this so?
Kellett: We were a bit surprised by the number of business intelligence deployments we found in the average organization; we reckoned it to be a dozen-plus. A big part of the reason is that the history of BI has been based on organizations saying, "I've got a particular information problem I need to deal with" and then applying solutions to those problems.
Question: What problems can departmental BI cause?
Kellett: Typically, when you have systems that depend on secondhand operational data, you have initial data quality problems. That gets even more complicated when you are talking about multiple departments. You end up with lots of different versions of the truth. So if you are in financial services, you might end up with your marketing department sending out a loan promotion to a customer on the same day that they get a notice from your operational department telling them that they've exceeded their overdraft limit. This is poor customer service; it makes you seem as if you don't know what you are talking about. If the departments had access to the same data, they could get together and send a message along the lines of, "We know you've exceeded your overdraft limits, but we still feel you're a good customer. So we'd like to help you out by giving you a loan."
Question: How can companies move toward an enterprise, rather than a departmental, approach to BI?
Kellett: Reducing the number of vendors obviously can help, but most organizations are a long way from bringing their business to one vendor. The finance department has gotten used to the comfort of its BI tool, and they know theirs is best. The problem is, the marketing guys feel that way as well. And so on. What's needed is someone to take an independent view of what your information requirements are and which tools will work best to meet those requirements. The good news is that BI vendors are beginning to extend their products into operational areas. So the tools aren't just about data mining or analytics, they're also about delivering information on operational issues in a way appropriate for the end user, whether it's dashboards for managers or actual business applications on the shop floor. And it's no longer just the pure-plays like Cognos and Business Objects; the enterprise vendors are all becoming interested in this space as well.