“What is the ROI of my Business Intelligence platform?”

The irony is cutting! And yet, it’s a shockingly common refrain amongst Business and Technology leaders – the motivations ranging from genuine curiosity and shaken faith to baffled guilt or plain resignation. It’s particularly disturbing given the volume of attention and investment in the Business Intelligence space. To provide a sense of scale, Gartner reports forecast Global Revenue of over $18 Billion in 2017 for BI and Analytics software market. So, what gives? Why do we have this conundrum in, inarguably, the most strategic driver of business?

There are several contributing factors, but the most fundamental is a structural flaw. Unlike most transformative ideas in the past, there has been a singular lack of formality about framing Business Intelligence. Definitions are loose, competence is dispersed and Accountability remains diffused. So, while short-term recognition abounds (the “hero” mode or Noise), long-term sustained value creation is severely compromised.

Business Intelligence Tapestry (BITe) provides a structure to crystallize the components, process and semantics of Business Intelligence. This is critical in optimizing resource alignment & investments.

There are four principal elements of BITe (Figure above):

The Data Framework: Focusses on data structures, semantics and sensing/movement/storage of Data.
The Visualization Framework: Focusses on information modeling and dissemination efficiency.
The Analysis Framework: Focusses on domain modeling, applied statistics & Operations Research.
The Response Framework: Focusses on developing Business rules and activation processes. (An additional note: This is the area that receives the least attention & investment, so, it’s potentially where the biggest “loss” occurs.)

An important, and formal, distinction is that each component is treated as a Framework. Most existing implementations follow one of two extreme models. In less formal settings, these four capabilities are not distinguished into individual tracks at all, resulting in a Master-Slave mode (typically, either Visualization or Analysis, becomes dominant & the remaining ones are developed on an ad-hoc basis, as enablers). This is not a scalable or sustainable model. In larger and more formal organizations, the pendulum swings to the other extreme where each capability is developed independently resulting in an Empire mode (each capability is considered as an end). This approach compromises value creation because of its inherent lack of throughput velocity and dispersed accountability.

The frameworks allow for a more structured approach to identify and develop each of these capabilities, proactively. The focus is on crafting a blueprint or reference architecture to build upon. The main departure from the traditional approaches, is that physical realization of these plans is driven by an external (but collaborative) force & not from within a silo.

The Data Science Spiral is that force or the fifth element. Metaphorically speaking, it weaves together these distinct fabrics into a tapestry. Essentially, it’s the architecture that aligns these four distinct disciplines into a common business initiative. The integration design pattern adopted is the Spiral methodology. Each spiral arc represents a business goal and the capabilities required to operationalize it. As each arc of the Spiral is defined, the specifications for the physical realization of each of the four frameworks is crystallized. Thus, the resources are aligned towards throughput or high-velocity value realization, but are building blocks of an existing blueprint. The organizing principle of this approach is to address weakness in two main business drivers: Lag and Weight. The collaboration minimizes delay between conception and activation. The incremental build-up optimizes TCO. This is a fundamental shift in the conception of BI programs. Another key paradigm shift is in evaluating programs. Typically, most organizations evaluate the maturity of their programs by measuring the sophistication of each area separately and assigning ad-hoc weights. These numerous linear models are inherently static and hence, limited in their usefulness. In a Spiral methodology the maturity is measured in “throughput” – the radius of the arc & the flow velocity.

While the spiral philosophy, itself, is not unusual in software development (it’s at the heart of Agile movement), it’s worthwhile to note the much larger organizational scope and purpose when it comes to Business Intelligence programs. This strategic rethinking is the core of “framing” BI and its role in an organization. That begins by revisiting the original ROI question – the collaborative framework exposes it as ill-conceived or malformed. Value isn’t discretized in individual parts of a program but, it’s a continuous accumulation as the arc spans all functions. The sum is greater than its parts. So, throughput based evaluation seems inherently more relevant than a capacity based one. But, the most fundamental outcome of this exercise is the shift towards a different line of inquiry “What is the measure of Business Transformation?”.

Business Intelligence and Data Science arc

dCAF: Refresh your customer brew

BITe: The Curious Incident of ...