CW reports that Forrester has released a report that managers believe they have too many BI tools in house. Although I have not read the full report yet, I agree with the basic claim. In my own experience, where I was heavily involved in analysis in a large enterprise, there were at least a half dozen tools that were commonly used. Plus just about every other tool was being used somewhere for something.
It is true that this creates considerable chaos. Much of hidden deep in budgets and collaboration complexity. Duplication of training, interfaces, licenses and the understanding of the basic assumptions used by various tools is a costly mess for the enterprise.
So why is this the case? In part because there are no complete definitions for the meaning of BI. As a result many enterprise packages like SAP and Oracle come with their own BI capabilities. What were once purely statistical packages like SPSS and SAS that now have BI constructs. Even ubiquitous packages like Excel have become so feature rich that they too can be transformed into something that looks much like BI.
It is easy to convince management that any of the packages above have just that feature that will provide the key transformational analysis that will solve an important problem. IT may disagree, especially if the license fee is high, and may argue that the complexity added to the enterprise has its own cost, but in general they will lose the argument because they just do not understand the analytical side of the argument.
Finally BI tools proliferate because their best advocates are new-hires coming out of school have been trained in the use of new tools. It is something fresh, new and unique, like them. Which is why vendors provide free licenses to Universities. So the training of new people in the internal tool is a barrier. The overall costs are not thought through. Management usually does not understand the features and mathematics of tools already in use. Once a shiny new tool is brought in and is presented to the right management it is then very hard to dislodge.
I have seen this complexifying effect operate for many years, in fact for my entire career in the enterprise. At least now we rarely write our own tools, but the sheer number of options is numbing today.
So what to do about it? Have a core team that really understands the infrastructure, IT and mathematical capabilities of the tools in use. They also need to know how the problems have been solved in the past. The team needs to work with all the divisions: Engineering, R&D, HR and marketing etc .... It also needs to have some clout and reasonable veto power. They also need to make the overall solution more accessible and easier in most cases. Benchmark with similar firms.
Sure there may be a need for specialized vertical capabilities. But with only some work these can be kept at a minimum. Choosing a limited set of options also allows the construction of libraries of solutions that can be reapplied. Have analytical experts available that can evaluate the problem solutions. Connect all your BI users together with Web tools so they can collaborate, that is easy today.
Update: in the comments below, Colleague Stan Dyck provides some excellent additional insight into this topic.