Yesterday I attended a product roadshow for one BI industry leader: the Tableau software for visual data analytics.
Tableau is, as per the speaker, the fastest growing BI solution. The product has made it ,in rocket speed, to the Gartner’s top industry leaders list.
After the new features review, we had two customer case studies: one from Serco (for Dubai Air Navigation Services), and the other from mmi. Both presentations were impressive, and the presenters (mid-management level) were so enthusiastic in expressing their satisfaction about the product and how it really leveraged their work and benefited their organizations.
Then came the Q&A session and I posed the following question:
I have been using Tableau for almost two years for developing reports and I do agree that it will be amazing in the settings as in both customer case scenarios, where in the first scenario it is used by a professional business analyst who will utilize the IT team expertise to extract required data from the different systems in Excel format, and she would use Tableau to analyze the data, provide insight, and prepare reports and conclusions to be shared with top management. In the second case, IT had implemented a data warehouse for its various systems, and it gave access to different Information Workers (IWs) to prepare ad-hoc reports in Tableau using the developed cubes. In both cases, Tableau was perceived as a revolutionary tool for end users, because it empowered them to slice and dice their data in the most simple ways that were never imagined before. They look at their data the way they wish to answer questions they would like to investigate and take insightful decisions. It, in a sense, frees the data when it empowered the middle tier of IWs, and frees the IT team by shifting the load of slicing and dicing the data to IWs so that they can shift their focus to other innovative business solutions. But it did so with taking into consideration that data has got one single version of the truth: data files requested by a single entity and extracted by a single entity in one case, and one data warehouse that integrates all systems in the second case.
My question is, what about the third scenario, which I argue is the most existing one, in which you don’t have a data warehouse (it is a costly, complicated, and rigid project with many considerations to be taken) and you didn’t make that transition to create and empower a third tier of Information Workers, so you ended up with IT getting maybe hundreds of requests for reports that need all to read from transaction systems through SQL queries. You may end up with your reports slowing down performance in transaction systems, especially that Tableau does not support stored procedures. What is your advice in such scenarios?
We then had some discussion on the need of staging the data, a feature available in Tableau engine and it can also be handled by a third party tool such as Hadoop (as suggested by one attendee: Hadoop is a distributed processing platform that runs on commodity servers and can handle structured and unstructured data storage and analytics.)
It was a fruitful discussion. I enjoyed the session and I like the product as one powerful tool for next-generation data management.