By Jeff Rosenbeck and Kristen St. Jean
The world of analytics is constantly moving and evolving. It never stops. Even the most successful analytics programs will eventually reach an inflection point when end-user demand outstrips near-term capacity. Rather than let this increased demand sneak up and catch you off guard, it is better to be proactive and prepared.
What can you do to prepare before your organization reaches this tipping point?
It is essential to keep the theme of constant adaptation top-of-mind, regardless of whether you are going down the road of an in-house build or relying heavily on partners. According to Steve Jobs, “If you don't cannibalize yourself, someone else will." This highlights the perils of approaching analytics with an unbending certitude instead of recognizing the need for flexibility. A strong, adaptive analytics program does not just satisfy the current faction of analysts; it also opens new doors for them. More importantly, it creates net new end-users, or at least business units that want to be new stakeholders. With new users come new use cases, demand for more and better data, new ways to leverage data, and new methods that may not have been anticipated initially.
It is not easy to be willing to change the processes that extract, transform and load data, data models, semantic layers or end-user UX layers. Willingness to continually adapt requires open mindedness to facilitate creative dialogue around past decisions that do not fit nicely into traditional development life cycle processes, alongside rigor and governance. One approach is to evolve the conversation from “Can we do it” to “How best to do it.” Once this subtle shift in thinking has been internalized, it becomes easier for the team to embrace change and adapt.
What are the other important considerations when establishing an adaptive analytics program?
Q&A
If you are relying heavily on partners or looking for a comprehensive “buy” strategy, think about how adaptable your partners and/or platform can be. Discuss how use cases that diverge dramatically from the current or initial domain would be supported. Consider hypotheticals around new business units coming on board, M&A situations or major changes in regulations.
While in-house build-outs allow for more control, asking internal teams these same questions can ultimately lead to a better end product.
Iterations
Commit to manageable iterations. The key is to break from a mindset that the initial deployment – whether it is a new data domain or just a new reporting method – has to be perfect. Instead, put something out there as a first step and build on it. This approach has two benefits: results will come faster and more regularly while reducing risk; and your user community will be trained to expect these iterations. This last point cannot be underscored enough as it sets a cadence and offers relief when managing a long-term roadmap.
Velocity
The third and possibly most important point involves gaining velocity by multi-threading iterations – in other words, being able to run two or more iterations in tandem. The more iterations that are moving at any given time, the better you can now address more demand points simultaneously. This can turn into a fairly complex process given a mix of large and small effort iterations. Easing into it, ideally with the help of a dedicated project manager, is recommended.
The closer business stakeholders work with the technical team, the greater the chances of success. Only when business and technical staff are dedicated and embedded together – thru direct or functional reporting – can you start to generate velocity. Having business staff dedicated to the analytics program while also developing technical knowledge that allows them to be directly conversant with their more technical counterparts can be hugely beneficial.
Driven by the need to continually adapt to increasing demand, this approach yields not only more velocity but also a growth-oriented workplace that benefits everyone.
Jeff Rosenbeck and Kristen St. Jean are Team Leaders of Enterprise Analytics & Business Intelligence at PSCU. For info: www.pscu.com.