There is a basic first-step process that must occur in early phases of your business intelligence solution build-out. Make sure that you have business & system analysts interview, document and review with your key business stakeholders and business users, the documented process and entities to be captured to solve your business needs.
To rephrase that in a simpler, more succinct way: Model your business first with a conceptual model. This is very important as well as very difficult to achieve to a level of success in which you have accurately captured and modeled your business processes in a way that can result in a software solution. Very skilled analysts with the ability to speak in terms that clear in regards to gathering requirements and business processes, many of which the stakeholders may not have fully realized in their day-in and day-out work day.
The way that we do these through Microsoft Services and Microsoft partners for Microsoft Enteprise Cube (MEC) solutions, is to utilize our program for BI delivery models which describes best practices and techniques for understanding the customer's business models and business needs. Since MEC is most concerned about providing business value through BI solutions based on industry pain points, we devise conceptual models in Visio that are high level using ORM or UML. The first picture below is an ORM example from Visio followed by an example of UML. I have found that ORM is very much sufficient and appropriate for data modeling while UML has much broader application for capturing use cases:
What we have found with the Microsoft Enterprise Cube, BI SDM and the Microsoft Solution Framework, is that without a focus on gathering, documenting and validating models of business processes and business entities, your BI project runs a high risk of not solving the real business needs or business problems.
I am going to continue this discussion as a series throughout the summer. Next we'll dive into putting this conceptual model into a logical model and user E-R diagrams as we start to build-out the physical data models that will serve as the basis for both the relational and multi-dimensional databases within a complete business intelligence solution project.
- Mark (firstname.lastname@example.org)