If the goal of your business intelligence project is to deliver a dashboard, you should have warning bells going off in your head.
What is a Dashboard
An easy to read graphical presentation of the current status and historical trends of an organization’s key performance indicators (KPIs) to enable instantaneous and informed decisions to be made at a glance.
Business Processes and KPIs
The key performance indicators (KPIs) of your dashboard will result from multiple business processes.
Examples of KPIs are:
- New Customers Acquired
- Revenue by Segment
- Cycle Time
- Inventory Levels
The above KPIs result from the following business processes:
- Acquire Customer
- Sell Product
- Manufacture Product
- Maintain Product Inventory
If your business intelligence project is to deliver this dashboard, you will need to completely build the analytic foundation for 4 business processes in your data warehouse. If this is your first BI project, this will be a real challenge. A common recommendation is to build your data warehouse by focusing on 1 business process at a time so that you have frequent deliverables and can show progress.
A common response to this warning is that you are only going to build summary data for these business processes so that it will be much easier and faster to build. Continue reading for a discussion on grain and atomic data.
Dashboards and Grain
If you decide to only build summary data for your business processes to support your dashboard application, you are missing out on a fundamental tenet of dimensional modeling.
You should build your fact table at the most atomic level of data possible
With summary data, the executives at your organization are undoubtedly going to ask for more details. Without atomic data, you will not be able to answer these questions. Therefore I highly recommend that you do not build your fact tables with summary data only. Summary data can be created for performance reasons, but you should still have the detailed data available.
When is a good time for Dashboards
After you have built the foundation for multiple business processes with atomic data you can tackle building dashboards. Dashboards are unique projects in that they require different skill sets for their visual and artistic nature.
For more information on business intelligence project planning for success you should buy The Data Warehouse Lifecycle Toolkit.