With massive growth in data availability and development of more sophisticated algorithms such as deep learning, many organizations are using data-science methods to support operations. The typical approach is to identify an application area and then develop some models based on data that are internally and externally available. In this article, I argue that this constitutes a very limited application of the power and scope of data-science methods.
What determines competitive advantage for an organization is its strategy. A strategy is a series of interrelated decisions that guide other decisions within the organization. This set of decisions fundamentally shapes creation of value both for the organization and for its stakeholders. Because strategy is a series of decisions, data-science holds tremendous potential for shaping these critical decisions.
Data-science is the use of statistical and computational techniques to gain insights from real-world data. The scale and depth at which data-science methods can extract insights is often counterintuitive, mainly because these insights are not shaped by social conditioning and prior non-empirical beliefs.
Thus relating an organization’s strategy to its data-science capability can be super-additive to value creation. This raises the question, what is an organizational data-science capability?
An organizational data-science capability can be decomposed into the following elements:
Leadership
Decision-making culture
Data infrastructure
Human assets
Technology infrastructure
Analytics technologies
Leadership and its motivation to link organizational strategy with data-science is the foundation of successfully utilizing data-science. Leaders have to be evangelists, as well as, technically grounded in understanding the strengths and weaknesses of various data-science approaches.
How an organization makes decisions is a central component of the culture of the organization. It is much easier to link strategy with data-science if the organizational decision-making process is already small-data driven. However, if organizational decision making is overwhelmingly guided by the intuitions of a small group of leaders, implementation of a data-science capability will face tremendous challenges.
Data-science operates on well organized and accessible data. Hence, creating systematic ways to acquire, store and retrieve data are central to building an organizational data-science capability. If the organization’s data is in fragmented spreadsheets stored across various computers it is likely that organizing that data into usable stores will take a long time.
Human assets that have the motivation, training and ability to learn new methods are central to building a data-science capability. Given the rate at which knowledge and methods are developing within data-science and the multidisciplinary nature of the field, it is impossible for any one person to be a “full-stack data scientist”. Therefore, an organization has to create a “full-stack team” to build an effective data-science capability.
Technology infrastructure and analytics technologies are the easiest rung to climb in the journey towards building an organizational data-science capability. Most technology is either open source, with plentiful online training resources or offered by cloud vendors who have created their own training and educational resources. Once an organization has the right team in place they will adapt these technologies for use within the organization.
In this article, I have laid out a blueprint for how to create an organizational data-science capability that can underpin competitive advantage.
https://medium.com/@omalik_38149/do-you-have-an-organizational-data-science-capability-8cde74993776
Comments