Gartner has reported that by 2021 75% of the organizations will have a Chief Digital Officer. There is no gainsaying to the fact that most organizations are focusing heavily on ‘data’ – the new oil. However, the progress so far has been slow. SiliconANGLE has reported in January, 2020 that “the 38% of organizations that call themselves data-driven has changed little over the past four surveys”. This blog is an attempt to look at the data science space from a different perspective and offer a point of view on a probable solution. It does not cover a detailed description of how a step-by-step data space roadmap could be created for an enterprise.
Assuming data as an organizational asset, the traditional enterprises looked at data governance in the asset value-chain in the ITIL-based “DIKW” way, as shown below:
Many organizations have been successful in being data aware in the traditional approach that they know their data governance well enough to bring about business impact defined by their own KPIs. So far so good. In fact, Big Data technologies have also been adopted well by many enterprises. But, most of them lack a holistic approach towards deriving intelligence from the dynamic and ever-changing data engineering landscape where human reactions, interactions, evaluations, interpretations, and feedback are critical to capture, with respect to their user and asset value chains. Once we appreciate that the data engineering landscape is a continuum comprising multiple moving parts, we get a sense of the hiatus that enterprises need to bridge.
While data-driven organizations have done well in bringing people, process, and technology together and managing and governing data science from multiple sources, e.g. devices, IT systems, external sources, LOB apps and portals, through optimal data operations, two aspects seem to have been overlooked. One, market ecosystem and two, human-centricity. The following diagram depicts how the market ecosystem affects business process changes and plays continually and critically into every organization’s business strategies and resultant decisions:
Now, to human centricity. Attention to industrial-organizational (IO) psychology (focusing on workforce selection and training, policy planning, improving management, overseeing analysis, and synthesizing work styles), along with end-user well-being and enablers such as organization change management, multivariate networking, and ethical AI could come to rescue. Human-centered strategic goals need to be added to the business KPIs already talked about. I have borrowed Martin Seligman’s model for human happiness called ‘PERMA’ and have added those parameters at an enterprise level, my logic being humans make an enterprise and it is humans that businesses serve. So, if an organization wants to move from being data aware to data conscious, connecting these dots is a must. The following picture is an attempt to highlight the difference between a ‘data aware’ and a ‘data conscious’ organization with respect to human-centered business experiences.
Let us now see why we need the ‘data consciousness’. In short, organizations need to be resilient (with the focus shifting from robustness).
What this means is that unlike predefined set of analytical approaches, businesses need to remain vigilant and nimble to change course. I have drawn upon Dave Snowden’s Cynefin framework for this purpose. While traditional approaches can address ‘simple’ and ‘complicated’ challenges where good or best practices are available (as defined by Snowden), they fail to address the challenges in the ‘complex’ or the ‘chaotic’ domains, where predefined ‘analysis’ does not work. ‘Sense’ being critical to both of these domains, if we are able to apply Data Science faculties to alert us about early signals of such ailments through predictive and prescriptive models, organizations would be much better prepared to display alacrity in face of adverse business situations. These models would work much better (since they demand data richness and volume), if organizations are able to capture the ‘broken window’ syndrome through internal and external ecosystems. These are hard to catch through traditional ETL/ELT, data wrangling, cleansing, and munging techniques. This is where the necessity for ethical AI also crops up so that the whole initiative is not counterproductive. It is through bringing together of all the data workstreams, e.g. data engineering, data science, and data governance (that focusses on data monetization through data operations, privacy, and democratization through data virtualization and visualization techniques) in tune with the market ecosystem as well as IO dynamics that organizations would be able to achieve the desired level of athleticism and agility.
The point of view shared in this blog, albeit derived from first-hand customer experiences, is novel. Any new challenge warrants a ‘novel’ approach as suggested by Snowden. Experimentations and refinements (e.g. common industry nomenclature, would surely be necessary for different industries and domains but I am positive that adopting the approach proposed above, organizations can go a long way into being truly digital and data-led, especially in the post-pandemic era where they would be valued more for human-centered approaches to their products and services rather than their business KPI-led achievements.