BI - Matching Business Needs to Data Quality

With the number of "Land and Expand" tools being offered to end users, BI Professionals are tasked more than ever to justify their development lifecycles and the time to data.

The challenge is that a properly architected and executed BI solution takes time to develop and rush jobs lead to data quality issues. If bad decisions are made because of bad data in the BI solution - you will soon find your reputation and approval for future phases of the project severely challenged.

On the other hand, making engaged, excited users wait for the completion of a full solution will erode enthusiasm and threaten your sponsors commitment towards the BI solution.

In my classes, I am looking at the same problem. Students need to understand the importance of process, but also want to "get into" BI tools and Analysis.

What can we do?

One solution to this problem is to map business processes to data sets, and then qualify those data sets based upon quality. In addition to mapping the BI solutions against data availability, impact, and feasibility - we also map against quality and trust.

Taking this approach, we can identify those business processes that use existing sources of data which met our quality criteria (sometimes third-party data sources can be a great place to start - StatsCan, etc).

This means we can make meaningful analysis available to end users without giving them access to uncleaned/transformed data that we need more design upon.

Extra Benefit - Organizational Learning

As an additional benefit to this approach, we get the organization started on learning how to use the front-end tools that will eventually serve our larger solution. Bringing in new data sets as the work is completed.

While this may run contrary to the idea that we should design and build a comprehensive solution that has all quality issues addressed, I think that it is a greater benefit to give a little under controlled circumstances rather than make users wait and run the risk of them slapping a tool directly onto a data source!