Without exception, my clients want to discuss self-service BI (“SSBI”). This isn’t surprising, as it’s both a buzz-word and mainstay of the analytics landscape. SSBI is universally recognized as a strategic driver and a business need, but some perceive it to be ambiguous or simply hype. I’ve had the opportunity to address this several times recently, and I’d like to capture my thoughts on the subject here. This will be the first post in a series.
At its heart, SSBI is about empowering users – data consumers – to analyze data their way, without IT prescribing the content or presentation of the output. Ideally, data should be available in a format and through a mechanism that eliminates the need for analysts to have technical backgrounds, but unfortunately, this isn’t always the case.
No matter how mature an organization’s SSBI platform may be, there are always opportunities for growth – but how do we identify and address them? If we adapt a general maturity model to evaluating SSBI capabilities, we can begin to define achievable objectives that score and help guide incremental improvements to an SSBI initiative.
Similar to the Capability Maturity Model (CMM), this framework has five levels:
- The business must ask IT for what they need, and IT pulls and delivers the data. Most organizations start here, but without advancing beyond this level, both IT and the business will struggle and ultimately fail to sustain growth. This really isn’t SSBI.
- This is where SSBI begins, and unfortunately tends to languish. In this scenario, IT provides access to data, possibly even a data warehouse with some standardized reports to expose the data, but the content and presentation becomes stale over time. In the absence of updates to data structures and reports, the business ends up pulling copies of the data and transforming it to meet their needs. This leads to the formation of “data silos” (often “spread-bases” or “spread-marts”) localized to individual departments. As the SSBI capability becomes more fragmented across the enterprise, there are some typical and unfortunate victims – among them, enterprise data definitions, the “single version of the truth”, and any semblance of change management or data governance.
- This level is achieved when an organization introduces a common framework for analytics. This usually includes an attempt at data governance, a standardized set of visualization tools and a commitment from IT to align data structures with business needs. These advancements give rise to new technical challenges, however, when it becomes clear that the “plumbing” that feeds the data structures has unresolved problems and major inefficiencies.
- In this state, the plumbing has typically been gutted and replaced to include features that ensure streamlined data movement, definitional consistency, and data fidelity. This may include a transition from a traditional data warehouse to a data lake, and is certainly earmarked by the presence of an enterprise data dictionary, change control, and data governance – all of which enable data to be “certified” and trusted by analysts.
- This is a philosophical ideal, serving more as a strategic goal than a tangible or attainable state. Very few organizations will ever reach the point where they will describe their SSBI as optimized, but this shouldn’t stop us from pursuing perfection for the platform.
Of course, this raises questions. How would you rate your organization? How can you intentionally pursue advancement within this maturity model? What challenges will arise along the way, and how should we address them? And, a big one – what does good data governance look like?
I’ll continue to explore these topics, and I welcome feedback along the way.
About Kevin Mintmier
Kevin is a consultant with over a decade of experience leading successful business intelligence engagements. He believes BI is a pillar of organizational maturity, and his mission is to strengthen relationships and BI capabilities wherever he goes. Kevin enjoys writing and spending time with his five-year-old son.