Oracle's recent release of Oracle Financial Analytics for SAP got more folks talking about it, Information Management's Jim Ericson points out in a column. Ericson quotes one expert, Forrester Research's Boris Evelson, who doesn't think embedding analytics in BI applications is a good idea. Rather than seeing analytics as the answer to all of their BI problems, Evelson suggests companies need to focus more on building the solid components of an enterprise BI architecture such as good data quality.
But most BI architectures focus on setting up and automating the distribution of standardized reports, not on yielding so-called "actionable" insights. IT organizations don't have the time or resources to produce new reports on a dime, so users with questions not addressed in standard reports typically take the data and transfer it to Microsoft Excel, which in most organizations has become the de facto data analysis tool.
Traditional BI restricts users' access to source data. IT organizations usually take this data and stage it in OLAP cubes to give report-reading users the ability to drill down for further details. Yes, this typically results in higher quality data, but this approach doesn't give users the flexibility to ask questions that occur to them on the fly. For that, they rely on Excel.
Relying on Excel creates problems, if for no other reason than it's darned easy to make errors in Excel. Errors typically go unnoticed and are then reproduced in other analyses. Much of the data entered into Excel also never makes it back out, which creates and worsens the information silos that are a weak link in the BI chain for so many organizations.
Excel also hasn't offered the ability to work with large amounts of data, though that's beginning to change with Excel 2010, which has an in-memory engine Microsoft says can handle up to a hundred million rows of data. When I interviewed her last November, Arlene Watson, a BI Solutions architect for Microsoft partner and consulting firm SharePoint360, told me a combination of Excel 2010 and BI tool PowerPivot can replace the need for data marts for many BI users.
A reader called Tom K, commenting on Ericson's post, insists analysis must be incorporated more directly into BI. He writes:
... The people who bring these together are going to be the winners. This is like trying to represent data without understanding the business process or trying to automate a process with no understanding of the data and little richness in the data model. A great mathematician and philosopher, Alfred North Whitehead, spoke of the world of value and the world of activity. They need to come together. ...
In his post, Ericson also promotes an Information Management article that explores the issue in more detail. The growing volumes of unstructured data, not well addressed in traditional BI architectures, are forcing the issue, writes author George Shen, an Information Management specialist master with Deloitte Consulting. Shen advocates an integrated approach that combines advanced analytics with powerful data visualization and advanced reporting capabilities.
Shen predicts vendors will move in this direction, with BI giants like IBM and SAP acquiring smaller analytics players and integrating advanced analytical tools and capabilities into their BI solutions. (IBM is already doing this.) And, adds Shen, "traditional analytics software vendors will likely push more into the BI platform territory."
These moves from both megavendors and smaller specialists will be good news for BI, says Shen, "accelerat[ing] the consolidation, standardization and adoption of analytics while moving toward an analytics-oriented BI architecture."