Off the Shelf: The Retail & CPG Blog

There has been a lot written about the necessity for better Business Intelligence (BI), the how, what and why to build a Data Warehouse and the impact it can have on driving a data driven organization as opposed to a gut based decision making organization. However, one of the challenges in some of the best executed BI programs is how to increase the usage of the BI solution. The solution is built on a solid foundation of understanding the business, bringing in the relevant data sources, building the right hierarchies etc. BUT it fails to penetrate far and wide in the organization.
 
My thought is that while building the BI solution, a lot is focused on what the end KPI's will be how to get them into the BI solution. This I call the science of BI. This is really useful when the objective is to be able to bring the organization to be driven by KPI's. It is useful to track sales, financial & other KPI's. It also enables the user to have the power of having these KPI's at their finger tips. But the final interpretation of the data is done by the business user. For example, a dashboard that tracks the sales in an organization is a very good tool that allows the user to look at the sales, shows them whether it is increasing or decreasing, maybe drill down to the level of details on KPI's to determine why it is increasing or decreasing. But this guided analysis always doesn't yield results.

• The Sales Manager might think - 'Maybe the sales are increasing because of increased promotional activity by us'
• The Brand manager build a hypothesis - 'Maybe the sales are increasing because the launch of the new product (which was bad) by the competitor has turned their loyalists away'
• The General Manager looks at the macro picture - 'Maybe the economic stimulus flowing into the stores in this city is pushing up the sale'

The end user who knows the art of interpreting the data will develop a hypothesis and then use the data to either prove it right or wrong. If right, it makes its way into the explanation of numbers, if wrong – another hypothesis needs to be developed. The questions remain – Is the BI Solution easy enough to allow the end user to answer the hypothesis that they have developed? Is it solution wide enough to provide the answers for a majority of the hypothesis that come up ? The availability of data yields to further need for data so the BI solution is always ever expanding. To cater to the needs of end user, we need to better understand how they think. The science of BI needs to understand the art of building the hypothesis.