Much has been made lately of the wave of “agile” business intelligence applications that allow business users to quickly and readily access data in meaningful ways. The non-technical user can now quickly create analytics and visualizations of complex data without relying on IT for their development and delivery. The focus has largely been on the applications themselves, as if the deployment of these solutions creates an agile delivery of analytics without the implementation of a formal process or methodology.
But are these applications all it takes to be “agile”? While they can certainly be labelled as “self-service”, do they fit the requirements of an agile delivery model? The agile methodology is characterized by shortened development cycles, iterative delivery schedules, flexibility, and direct customer involvement. The new wave of desktop analytics and visualization tools certainly fulfill those elements, but they are only the tip of a much larger iceberg when dealing with the delivery of enterprise data. To be successful in business intelligence agility, data must move quickly and accurately from transactional, operational systems to managed, analytical repositories easily consumable by the business user.
In his article What Agile Business Intelligence Really Means, Wayne Kernochan describes the model for information distribution as data entry, consolidation and aggregation, followed by information targeting, delivery and analysis. So-called agile BI tools generally only address the last two phases of this process. Kernochan refers to the first three phases as accuracy, consistency and scope, terms that generally fall under the purview of data governance. In fact, data quality has to be part of any agile approach to BI, for inaccurate and missing data in the hands of business users slows down the iterative process and creates distrust among customers.
In a truly agile BI environment, each of Kernochan’s phases would be contained within an overall delivery sprint around a specific subject area. Each sprint would be directed by the business requirements around that area, and would emphasize the accuracy, quality and efficiency of the data to be presented to the users. In fact, frequent input from the users during the sprint’s development processes is paramount to its success. Once the data is ready to be consumed at the end of the sprint, business users can then use any number of “agile” applications to analyze that data. In this sense, such applications are part of a larger agile process and methodology (which should include other applications for master data management, metadata cataloging, etc.) that truly meet the information analysis needs of the business.