A few years back BI was a buzz word in Sri Lanka with solution providers excited at the prospect of selling this new concept while corporate IT managers and CTO's wanting to showcase their organization as the pioneer in cutting edge technology. Fast Forward today, are we actually delivering on what was expected or are we just being content with a few glorified MIS reports?

BI vendors tend to oversell the concept to the top management who buy into the fancy demos, acronyms, jargon and surface prototypes depicting digital dashboards, KPI ect.... Well in theory it’s all fine but in fact extracting this data is the biggest issue. When I say extracting it’s not about having a powerful ETL tool, it’s where you get the source data from. BI vendors tend to patch up the data (eg. From temporary worksheets or dummy databases) when developing a prototype. The outcome is a weak implementation due to the fact that the numerous sources of data required to generate the relevant BI reports are not available. This is a key issue in local organizations where data is either not captured or ambiguous. The end result is the BI solution been underutilized, not updated with new data sources and finally become obsolete due to inaccuracies and/or availability of the same information in MIS systems.

Customers need to understand what they actually require from BI and not just what looks good. They also need to avoid getting carried away by theories and concepts which result in highly theoretical requirement specifications being submitted to vendors.

Vendors need to set expectations and not overkill the potential of BI within an organization. This means understanding the limitations of the customer in regard to data sources and more importantly identifying where BI would be actually required.

The key to a successful BI deployment would be a sound business study which involves identifying the correct personnel who have expert knowledge in their respective area within the organization.  This is the basis for identifying where the required data resides and in what form. Based on these findings we could correctly asses the deliverables. Well the rest, will hopefully fit into provided we have the right tools and technical expertise, but thats another blog entry. J