Heard Around the Water Cooler
What do you mean we don’t have that data in our new data warehouse?
Every Friday I meet with the key BI stakeholders for one my clients. In those meetings we discuss current issues, questions and concerns within the BI team. We also explore any new issues that someone has discovered and talk about new requests that might help meet business objectives.
Last week the group received a new request that became a cause for concern when they realized the new request would require data that was not yet available in the new data warehouse. Incorporating the necessary data into the new data warehouse architecture would take some time—definitely was not something that could be done immediately as per the request.
Hoping to bridge the gap, I thought back to one of the eight rock star principles that Bernard Wehbe (StatSlice Systems Co-Founder) talks about in one of his popular articles. Principle #6 is learn how to develop prototypes quickly. Developing prototypes quickly is something that has been an ongoing focus with my co-workers, and I was pretty sure I could put something together to meet the stakeholders’ needs, at least to the point of delivering the requested data on a temporary basis while simultaneously pursuing a long-term solution. I proposed that we create a quick SSAS cube and then publish the data in an Excel Pivot Table. I could grab what was needed to populate the cube from staged data, so no need to query the source systems. Since there were no true requirements yet other than access to specific data elements, there wouldn’t be much point in creating a standardized report, or calculations and complicated business logic. End result should be an ad-hoc cube with less than a day of effort.
The prototype went according to plan and addressed the situation effectively. Utilizing Analysis Services for the business layer, we were able to control the aggregations to ensure fast response times to user queries, as well as creating some rudimentary hierarchies very quickly. The presentation layer was handled with an Excel pivot table, giving the users access to all the fields in the cube prototype, as well as some initial understanding of how the fields might interact in different combinations.
In the end, the prototype provided exactly what they wanted to see at that point in time, not only to meet their immediate analysis needs, but also to refine their understanding of what they really wanted as a long-term implementation. In other words, the prototype provided the business users a way to see the data and THEN determine what type of analytics they wanted to see going forward. This in turn provided me with a better blueprint for the more significant data warehouse changes that were needed to meet their long-term needs. We were able to push it out with minimal development costs and team stress, and most importantly, maintain the quick turn-around times that the stakeholders’ business users expected. Score one more for quick prototyping.
By Brett Neuman, Manager Consultant, StatSlice Systems
Brett has been developing business intelligence strategies for more than eight years at several fortune 500 companies. He has implemented data warehouse solutions and analytics strategies for many verticals including Health Care, Finance, Retail and Marketing. He has a strong background in application development, data warehouse architecture, data modeling, ETL, OLAP, reporting and dashboarding utilizing both the Business Objects and the Microsoft suite of Business Intelligence platforms. Brett joined StatSlice for the opportunity to empower organizations to make more effective business decisions through the availability of accurate, mission-critical information.
You can subscribe to our RSS feed.