An Effective Methodology For BI Implementation
Sign in

An effective methodology for BI Implementation

Senior Manager - Projects (DW/BI)
Given below is an approach that one could follow to ensure an effective implementation of any BI initiative.

1. Requirements elicitation is the most important activity in a BI exercise. We need to very clearly understand the motivation behind the management's need for analytics and their short-term, medium-term and long-term objectives. You may conduct a workshop with all stakeholders to understand these requirements. Identifying and involving all stakeholders in the requirements discussion is very important to gain buy-in for the BI initiative throughout the enterprise.

2. Anyone would like to see a sample of analytics and play around a little to get a sense of the benefit/impact of analytics. Hence developing a quick prototype and showcasing the same to all stakeholders will help you get the necessary impetus for the initiative. CAVEAT: Please ensure that the prototype development does not take more than two weeks.

3. Post this prototype exercise, you have to get back to the discussion table to finalize the requirements of all stakeholders. You need to remember here that requirements keep changing over time. Hence you may use the concept of time-boxing, wherein you brainstorm the requirements with the stakeholders and freeze a set of requirements for a time window taking care to not allow scope change during this defined time window.

4. Once the requirements are finalized, carry out a study of the source systems to understand the availability of data & quality of data. By doing this study, you can identify the source systems that are ready to be used for building an analytical application. You can also understand the gaps in source systems, if any. With these inputs, you can easily prioritize the order in which you need to build the data marts.

5. Once the priority and order are decided, prepare and publish a plan for implementation. It is advisable to use an Agile methodology for implementation. So when you prepare a plan, prepare for report/dashboard releases every 3-4 weeks or so. This way, you can quench the customer's eagerness to see some results every once in a while and avoid unpleasant situations or surprises much later in the implementation, when magnitude of re-work could be high.

6. Begin dimensional modeling for the chosen set of requirements and then go ahead to build the physical schema using ETL process.

7. Ensure that you have a dedicated team (outside of your team) to do SIT (System Integrated Testing) in an objective manner once development and unit testing are completed.

8. As far as possible, you should ensure that the User Acceptance Testing is defect-free. A dedicated SIT team wiill help you achieve this goal of defect-free UAT.

9. Once the reports/dashboards are launched in production, start measuring the usage of these reports/dashboards. You may even conduct a survey on the usability and usefulness of the analytics, 4-6 weeks after the launch of the reports/dashboards.

10. After the survey is done, you may engage the respective stakeholders again in discussion to understand what changes may be required to increase adoption of analytics.