According to Gartner, 70 to 80 percent of business intelligence projects fail. The solution could be a more agile development process — and organization.Having been through this, BI is seen as vaporware until something is delivered. Users can "test" a new application, they can see mock ups, they are involved from the very beginning. Enterprise application also have something called a release schedule, which BI generally does not have. An application evolves, a BI report is the end point generally.
The issue with an agile development life-cycle that I have run into is that the people that are on the project have to be able to adapt to requirements that are not clear cut, the end result is not known, and they generally have to be free thinking and have a keen awareness of what the business is what they MAY be looking for. In a best case scenario the end user is involved not only at the requirements phase, but going forward so as to clarify, add, remove requirements as the process moves along. Given how the economy is, this may be harder than it was in the past, where users are more concerned about getting their day to day done as opposed to helping with enhancements.
Last point, what is a BI project failure? Is it the system never gets put into production, or the system does not live up to expectations? If it is the latter, I can believe the percentages, since it never seems that people use the system as they said they would when they were giving the requirements. To illustrate, we put in a module for one specific group of users that took six months to develop and deliver (not very agile), and the end result, 10 reports a week being run. I would classify that a failure in that the ROI will never be there at that rate. There is also an opportunity cost associated with that as well, those 6 months could have been used on something else that may have been more useful to the business.
