The Intellectual Capital Flowing Through the Two Rivers of Data Science

If you love to read, evaluate and implement all things data science — as we do — this is a great time to be alive. The reason is simple.  More data science is being applied by more firms in more ways for more market-relevant ends than ever before.  While much has been written about the emerging data science revolution, let’s focus today on two poorly understood facts regarding the core intellectual capital of data science — and how this capital can be turned into competitive advantage for mid- to large-sized enterprises. First, wherever it may be found — whether in academic articles, conference presentations, corporate whitepapers or even in knowledge sharing services like LinkedIn, here’s the big secret — the vast, vast, preponderance of the intellectual capital in data science lies in the public domain. Read that last sentence again. (And again!) Never before has such a vast body of truly transformative information been available just for the taking. For firms that “get it”, access to this intellectual capital not only levels the playing field for all competitors but democratizes the benefits of data science for all firms willing to meaningfully and prudently invest in it.  Second, and perhaps even less recognized, this same body of intellectual capital flows in two very distinct “rivers” — each typically managed by two very separate sets of practitioners. Let’s take a minute to briefly describe these rivers.

The first data science river is strategic data science (SDS). In SDS, enterprise-level objectives like market innovation and product design are enabled by protocols (i.e. step by-step methods) and system constructs. SDS drives change toward a single all-consuming goal — competing and winning in the marketplace.  Accordingly, SDS integrates the sundry tools of data science with other assets (strategic objectives, market research data, risk management principles, cost data and the like). These in turn are configured to achieve mission critical enterprise objectives — such as uncovering hidden opportunities and designing and innovating new products that exploit those opportunities. The mission-criticality of SDS, if it is done right, is difficult to understate.  Specifically, if SDS-driven cycles are continuously and persistently applied, SDS is one of the main contributing factors toward one of the central goals of contemporary business management – a culture of continual competitive transformation. Given this state-of-affairs, as one might expect, the practitioners, purveyors and consumers of SDS tend to be process, change, and marketing managers. In terms of their education and natural predilections, SDS professionals tend to be drawn more from the “softer” (generally less quantitative) social/human science end of the spectrum — from strategy, marketing, business, finance and even the more recent (and exotic) sub-specialties (such as cultural anthropology and social psychology). Last, but hardly least, these SDS change agents tend to be both the firm’s solution architects and integrators as well as its data science aggregators.

The second data science river is tactical data science (TDS).  In the TDS river, problem-level and within-domain level objectives are in view. Here one finds all the tech-heavy assets usually associated with the term “data science.”  This is generally the better-recognized river within the data science ecosystem because this is the “hard science” end of the data science spectrum — where the more quantitative tools of data science are selected and implemented.  In contrast to the SDS river, however, the tactical data science river itself bifurcates into two other, distinctly separate, streams.  In the first TDS stream is found the quantitative assets of data science, its algorithms, machine learning repositories, statistical/bootstrapping frameworks, and the like.  These powerful assets might be considered, in a sense, as the “hardware” of data science. Accordingly, its practitioners are mathematicians, statisticians, marketing scientists, and psychometricians on the demand side, and econometricians, actuarial scientists and risk modelers on the cost side (which, in turn, also includes risk management, governance and compliance.) The second stream within the TDS river, in contrast, is the software and data platform stream.  This stream stitches together the “hardware” assets from the first stream so that we can actually do something useful with these assets. Here is found all the myriad of data architectures and applications — both homegrown and purchased — that allow the enterprise to collect, centralize, describe, extract, predict, summarize, and discover patterns (or uncover timebombs!) in the data. Here too is where Agile and other application development constructs live — as well as where R and software repositories (API’s) weave together otherwise stand-alone data science assets into useful code. Lastly, here is where Hadoop stitches together distributed data spread across the enterprise, and where your internal software developers either gain or lose customers in the battle for digital experience excellence. In summary, the TDS river — regardless of which stream we are discussing — is where data scientists use the tools of data science to solve domain-level rather than firm-level problems. And they do it every day.

As one might reasonably infer from this landscape (and given both the rising costs and opportunity profiles of these two rivers) one of the most critical challenges facing the enterprise today lies in integrating the largely uncoordinated efforts of the SDS and TDS “rivers”.  In one sense such an effort might be viewed as simply another chapter in the ongoing saga of aligning the enterprise’s tactics with its overarching strategy. In another sense, however, this is an entirely new problem because data science is newly emerging into both the strategic and tactical realms with assets that are transformative in both.  Thus, the current challenge lies in integrating both the strategic and tactical data science rivers so that they serve the overarching goals and objectives of the organization. In our view, two implementation steps are indicated.  First, aggressive internal integration and alignment across the strategic and tactical data science disciplines are required. While the nature of these initiatives necessarily depends upon the corporate culture, such initiatives may range from required (but passive) knowledge-sharing infrastructures to actively promoted and facilitated internal data science integration events and summits.  Second, and even more critically, there is an increasing need for a strategic data science officer (typically at the EVP level or above) to work with the CTO to coordinate all data science initiatives within the organization. Such an exigency ensures that tactical data science initiatives align with strategic objectives, redundancies are eliminated, and data science knowledge capital is populated across the organization.

James A. Libby, PhD., is the CEO of Decision Support Sciences.