This is the third post in my series (“four ways to change our approach to BI development”) on changing our approach to BI development. In order to break the BI backlog, move towards more pervasive BI and increase the BI business ROI, an enterprise needs to make fundamental changes to their BI efforts. These changes are interrelated and necessary for success.
For an enterprise to truly experience a BI breakout, here are four of the things we should do differently:
I’ll discuss the third one below, and the final one in an upcoming blog post.
One of the hot topics regarding IT project methodology is agile development. This type of methodology better supports the incremental and iterative development approach that has been so successful with BI programs rather than the “Big Bang” approach that enterprises often try using their systems development lifecycle (SDLC). Although I would love to recommend a pure agile methodology, unfortunately too often that scenario has led to development chaos and anarchy.
The best approach would be a BI Hybrid Project Methodology:
Check back in a couple of days for #4, “IT needs to concentrate on creating an information backbone.”
A key part of any hybrid BI development methodology is to get transparency over what the business is creating and running in operational spreadsheets. This is the melting pot from which most business needs will emerge. Armed with this knowledge one has much better chance of aligning the IT road map with business requirements.
Ralph, Great point & I agree with you.
In our consulting projects we always ask the business to review the spreadsheets that they use for reporting & analysis. We will also get their “new” business & data requirements for BI deliverable but the existing spreadsheets gives great insight (& detail) into what they need, what the gaps are and what they need to add.
Also, spreadsheets could be used (1) augment existing reports or perform follow-on analysis, (2) perform new analysis not offered by IT apps, (3) provide the superglue across data silos, or (4) be full-scale data shadow systems or spreadmarts.