In Unplugged, a recent article in the Information Management Magazine, George Shen asks whether Business Analytics (BA) is overtaking Business Intelligence (BI). This question seems only poised to get our attention because what Shen is really promoting is the need for an analytics oriented BI architecture that would incorporate better advanced analytics and modeling. I don’t think he’ll find too much opposition to that pitch. However, I do see a problem with the placement of the Advanced Analytics before the business semantic metadata layer. Shen is clearly not unaware of the problems with data cleanliness nor is he a stranger to the challenge of more user friendly design. However, in the article I don’t see a connection between these two problems and the need for a well developed business semantic layer or an indication that the layer is required to be accessed by the advanced analytics tools. Shen places BA beneath the semantic layer and only shows a push upwards. The other problem with the architecture diagram is the lack of emphasis on data that is external to the corporation. Even though Shen discusses examples of valuable data available outside the corporation he does not show how this would be incorporated into the architecture.
I like Shen’s statement “BA needs to be integrated and embedded in business process to be effective”. Too often we see that this is not the case and training for Business Analytics is divorced from the business process. Business Intelligence systems are often a very small percentage of the budget of a much larger ERP project and as such is treated much like an after thought. To compound the problem training is often treated as an IT function and provided as ‘systems training’ with an emphasis on standard reporting. There is often very little done to make the learning experience authentic by situating it within the current business process and relating it directly to real business problems. While I agree with Shen that the BI vendors need to step it up a notch in terms of their offerings and integration of advanced analytics, I think there’s an even bigger problem with the typical BI training approach.