Business Intelligence (BI) refers to the skills, processes, technologies, applications and practices used to support decision making, and is crucial component of strategy for businesses to operate successfully. MIKE2.0 has a valuable wiki article on this topic that shares guiding principles to help information management professionals develop a strong BI program.
- 1) Keep the strategy at the vision level
- 2) Use a requirements-driven approach
- 3) Develop a BusinessTime model for synchronisation
- 4) Use a well-architected approach
- 5) Investigate & fix DQ problems early
- 6) Use standards to reduce complexity
- 7) Build a metadata-driven solution
- 8 ) Store data at a detailed and integrated level
- 9) Design for continuous, increment-based delivery
- 10) Use a detailed, method-based approach
Below are the basics:
1) Keep the strategy at the vision level
Establish the Blueprint and never start from scratch – use best practice frameworks. Keep things at a strategic level while still following a diligent approach to requirements.
2) Use a requirements-driven approach
Even when using off-the-shelf information models, requirements must drive the solution. Plan to go through multiple iterations of requirements gathering.
3) Develop a BusinessTime model for synchronisation
Be prepared to handle growing requirements for the synchronisation of data in real-time into the analytical environment. Focus heavily on the “time dimension” as part of your architecture.
4) Use a well-architected approach
An analytical environment is not a dumping group for data. Data that is not integrated or conformed does not provide the value users want.
5) Investigate & fix DQ problems early
Data quality issues make it difficult to integrate data into the analytical environment and can make user reports worthless. Start with data profiling to identify high risk areas in the early stages of the project.
6) Use standards to reduce complexity
The Business Intelligence environment is inherently complex – to maximise benefits to the user the system must be easy to use. One of the most important things that can be done is to develop a set of open and common standards related to data, integration and infrastructure.
7) Build a metadata-driven solution
A comprehensive approach metadata management is the key to reducing complexity and promoting reusability across infrastructure. A metadata-driven approach makes it easier for users to understand the meaning of data and to understand how lineage of data across the environment.
8 ) Store data at a detailed and integrated level
Aggregation and integration is far easier when you store data at a detailed level. It you don’t store detailed analytical data, some users will typically not get all the information they want.
9) Design for continuous, increment-based delivery
Analytical environments should built through a “journey”.
10) Use a detailed, method-based approach
Methods such as MIKE2.0 can help provide a task-oriented approach with detailed supporting artifacts.
Need more info? Comments or additions? Please let us know in the comments section below.
