Tools that make BI agile

“Back to the Future III” has the scene where Marty McFly asks the train engineer in the wild west of 1885: "Do you think it's possible to get it up to... 90?”

The engineer's reply was: “Ha! 90? Tarnation, son, who'd ever need to be in such a hurry?”

How does the BI delivery team catch up to this runaway train of demand for data and advanced analytics?

Yolanda Smit is senior BI business analyst for PBT.

IT has only recently started to master the art of agile software development, and already the call for agile business intelligence (BI) and even agile predictive BI is ringing louder and louder. So, how does the BI delivery team catch up to this runaway train of demand for data and advanced analytics?

My trials and errors in agile analytics have led me to pick some generic tools that are critical for agile BI success. These tools should be non-negotiable, regardless of which agile methodology users subscribe to. I categorise the tools into three sets: (1) teamwork and planning, (2) BI best practice techniques, and (3) automation.

Teamwork and planning tools are all about how the team goes about working together. In “Back to the Future I”, Marty converts a boys' push-scooter to a skateboard for a “high-speed chase”, but in 2015, this is upgraded to a hover-board to stay ahead of the competition. In a similar way, these tools are all about progressing faster and better. Examples of tools in this category are collaboration and co-location.

* Without collaboration, the end result might still be achieved, but with a lot less agility and far more casualties. Collaboration is not something the team leader just commands the troops to do and it happens. It is the end-goal of a road of establishing a trusty relationship, and trust is earned. The waterfall relationship is almost forced by compliance guidelines. But in agile, as trust matures between all players, the relationship evolves until it is characterised by shared ownership, mutual accountability, minimal ego, and collective energy.
* The technique of co-location is a simple technique designed to promote collaboration. It subscribes to locating the developers and key stakeholders in close physical proximity to foster better collaboration.

The second set of tools is development practices focused on achieving a best practice solution amid agility. It's like exchanging the nuclear power of the Delorean for organic fuels using Mister Fusion. The components of Mister Fusion are agile design, refactoring, and testing principles:
* The first one referred to is evolutionary data modelling. A new set of modelling techniques is not needed. On the contrary, users should continue to subscribe to Kimball, Imnon or whichever founding father's design they subscribe to. Agile development merely requires a new way of applying good modelling methods in an incremental, iterative, and evolutionary manner.
* Building on evolutionary design, database refactoring is a set of techniques that enables users to make changes to the solution “without breaking previously working features and components”.
* Rigorous unit, integration and regression testing is the final tool in this category that ensures success. Test everything. Be confident that what is presented for deployment is actually 100% correct, in order to maintain trust in the collaborative partnership.

The last set of tools is a key technique for developers more focused on empowering agility... automation, automation, automation, just like Marty's jacket in 2015 that auto-resizes and auto-dries.

* Automated testing aims to optimise the continuously recurring process of regression testing. If users are testing data accuracy of a unit, they might as well script the test and save it, because they will need it again in the future.
* Automated version control is essential with smaller deployments more often in agile BI. If it's not automated, version control may become an overhead that deteriorates agility early in the process already. Proper version control provides several advantages to the team, such as the ability to roll-back, shared development without risk of overwriting, maintaining a proper audit trail, eliminating the need for code-freezes prior to a release, and ultimately, to fearlessness. Developers can feel free to experiment with solution alternatives, explore new ideas, and make changes without fear of adversely impacting others. Ultimately, this fearlessness stimulates innovation and collaboration.
* Automated deployments, aka push-button releases, are all about automating the whole deployment process down to a single procedure to execute. Ideally, users will want their deployment so simple that no time is required from developers to hand-over to deployment teams; and they don't want to have to write lengthy instruction documents to accompany deployments.

So, in “Back to the Future 1”, Marty McFly finds himself stranded in 1955, and is faced with the challenge of helping his teenager dad win the heart of his teenager mom to ensure the continuation of his own existence. They both conclude the best way to do this is to convince mom that dad is brave. They plot to stage a scene where dad can act brave... they were planning to 'do' brave. However, their plans are foiled and dad McFly is forced to 'be' brave, and punches out his arch nemesis, Biff.

Now, the question may be asked, how is this at all relevant to BI? Similarly, agile BI is not about doing agile, but about 'being' agile while doing BI. When employing the tools described in this Industry Insight, the BI team's DNA, nature, and culture will gradually change as they are empowered to become agile while doing what they are ultimately paid for: BI!