Share
Readers of business and popular press are inundated with hyperbole on the promise of machine learning and AI to revolutionize business strategy and management. There is a bewildering array of companies all claiming to unleash the miracles of advanced data science to solve all manner of problems. The reality, however, is that most companies are not prepared to realize the full or even partial potential of data science, and most solution providers do not have the combination of capabilities to make good on the spectacular promises of their marketing.
Working with hundreds of private equity investors and their portfolio companies over the past twenty years, Stax has observed a common set of challenges that inform a realistic view of the value of data science and increase the odds of success for investors and management teams.
Building a data science foundation required to drive value
There are two critical dimensions to keep in mind, one at the foundation of driving value, and one at the apex of guiding value. The foundational observation is that most middle market and even larger enterprises face data structure, process, governance, and definition issues that limit the value that can be realized from analytics and data science. While such issues are not as glamorous a topic for discussion as AI or machine learning, the value of high-end data science tool deployment is limited without first addressing the fundamental issues. Any solution, internal or external, that minimizes this fundamental truth is doomed to failure, not least through erosion of credibility within company management.
Understanding the strategic objective before deploying a solution
The second and guiding observation is that any effort to deploy data science and advanced analytics must be directed by business context and strategic value first, not directed by tool deployment. While this may seem an obvious point, often the purveyors of solutions are so enamored with their technology or methodology that they seek first to deploy a solution, looking secondarily for strategic value.
The first organizing principle for all data science deployment must be identification of high-stakes strategic questions. What, if solved or answered, would allow for valuable action toward a strategic objective? What insights, in what form, can realistically be acted upon by company management? Data science can generate all sorts of fascinating insights that are not actionable, and as a result relatively useless. Start with the right question, and organize data structure, technology, and solutions around the question, not the other way around.
The benefits are many of keeping these two observations in mind as investors and management teams seek to drive value from data and data science. The primacy of strategic value provides for effective prioritization and deployment of resources. This approach can lead to a sequential, self-funding build of data insights. A specific, limited set of actionable insights leads to value, which builds organizational credibility and value to fund the next set of actionable insights. The sequential approach, with specific, demonstrable wins, can change culture over time, empowering teams to ask and answer questions grounded in data. And the non-glamorous but critical work of building out clean, well-structured, and governed data is foundational to any success in deployment of data science.
Read More
