PLEASE NOTE: THIS IS WRITTEN WITH SANDER HOOGENDOORN - PRINCIPTAL TECHNOLOGY OFFICER AND AGILE EVANGELIST @ CAPGEMINI.

I was asked the question: ”Is Agile the correct development method for Business Intelligence?”.  An interesting question. According to me any development method that takes BI characteristics in mind is better than a method that does not. Therefore I would say that Agile is better than most waterfall based methods.

So, what are some of the BI characteristics that we have to take in account while using any development method? First , BI projects almost always have some sort of business purpose. For example, sales insight or call center performance improvement. But the business case is not always clear upfront. So the BI contribution to the bottom line is often unknown as well. Another thing is that the business requirements are also often not very precise. At the start of any BI project getting the detailed and finalized requirements from future information product users is a challenge at best. Finally, the whole development process is pretty much linear. We get data from the source, we ETL it to our warehouse, fill the data model, ETL it to our data marts and build the reports. So before any user sees the final result 80% of the work is already done. The biggest challenge for BI development methods is to deal with the ‘advancing understanding’.

Agile can contribute to this as it works with: small iterations, used cases, fast and frequent releases, much interaction with the client, integration of testing in the development cycle and so on.

The key question that remains: ”Are agile methods being adopted and used in the BI world?”. The answer to that is very simple: Agile is all in the mind. The question is not if Agile CAN be used, the question is if BI people WANT to use it.