min read
May 6, 2021



The overwhelming majority of IT executives in North America either have machine learning programs in place now or plan to have them in the near future. As in the related fields of artificial intelligence and data mining, machine learning is rapidly becoming as essential to 21st century business as email or Wi-Fi.The two leading implementations of machine learning today are: predictive analytics, which reduces uncertainty in decision-making; and recommender systems, which increase cart sizes by suggesting other items that match user preferences.

A survey presented by 451 Research and Blazent found that more than two-thirds (67.3 percent) of execs either have or plan to have predictive analytics in place. Nearly the same number of execs (66.7 percent) currently use recommender systems or have implantation projects on the books.

Once machine learning systems are in place, execs have found countless use cases for them across all enterprise. According to a 2016 Statista survey of developers who are actively creating new applications, nearly half (47 percent) are putting predictive analytics to use for more intelligent asset management.

At the other end of the scale, 11.4 percent of respondents are developing machine learning applications for the Internet of Things (IoT). These market leaders are preparing for 2020, when spending on IoT will reach $1.29 trillion according to the IDC. Between these extremes, 25 percent of execs said that network performance optimization was the central goal of their machine learning application development.


While machine learning is usually associated with new and innovative startups, in reality it has become a critical business tool in traditional large-scale enterprises like General Electric. This energy company, founded by Thomas Edison in 1889, has been taking advantage of data crunching applications to discover optimal performance profiles for jet engines, finding the best new locations for their deep-sea oil drilling operations and to prevent breakdowns by predicting where they are most likely to occur.

In the financial sector, banks in Western Europe have seen new offering sales soar by 10 percent by using machine learning algorithms instead of outdated statistical models of the market. They were able to lower their CapEx investment by 20 percent and register an equivalent drop in customer turnover with machine learning. Recommender systems boosted sales for banking clients, and predictive analytics alerted retention departments to the signs of dissatisfied customers most likely to cancel their accounts. Machine learning even makes suggestions about which types of interventions are most likely to keep customers onboard.

While AI and machine learning are normally used for number crunching applications, more complex analytical projects such as hiring were assumed to be a task requiring human reasoning. That is no longer the case. Gartner predicted that “Algorithms will transform recruitment processes by replacing reliance on both recruiters’ intuition and automated CV evaluation based on word matching with insights gleaned over time from analysis of large datasets.”


We are now living in the midst of a new industrial revolution that will change business and society in unpredictable ways. The World Economic Forum outlined the current age in relation to its historic counterparts: “The First Industrial Revolution used water and steam power to mechanize production. The Second used electric power to create mass production. The Third used electronics and information technology to automate production. Now a Fourth Industrial Revolution is building on the Third, the digital revolution that has been occurring since the middle of the last century. It is characterized by a fusion of technologies that is blurring the lines between the physical, digital, and biological spheres.”

As this digital transformation takes over, executives will need a deeper understanding of how best to apply data analytics and machine learning.


Classical statistics has served as the foundation for business decision-making for the past 200 years or so. AI and machine learning have emerged from a series of advances in applied statistics, software development and logic circuits, while the size of data sets has expanded exponentially to enable precision and scope for deeper analytics.

Although elements for AI like neural networks and parallel processing have existed since the earliest days of computing, they required the power of distributed computing and cloud-based networks to grow AI into a practical discipline.

In prior industrial revolutions, innovations took several decades to go from novelty to ubiquity, which happened to photography in the 19th century, electricity in the 20th and mobile phones in the 21st. Machine learning applications for business are still on the first part of that maturity curve, but cultural integration is accelerating.


Many companies have already discovered that there is danger in adopting machine learning without a top-down strategy to drive it. Machine learning and AI are little more than a novelty, mainly deployed to find and hold onto customers, unless the C-Suite incorporates them into strategic planning.

Consider how enterprises approach mergers and acquisitions. A company that rushes into buying up other companies without a well-defined strategic vision will end up without a core value proposition and shareholder value. Harvard Business Review concluded this occurs when “too many executives bring insufficient discipline to the evaluation process that fuels these deals — as a result, they often get deals wrong.”

Executives should approach the potential of machine learning with the same caution. This requires a commitment to evaluating the full extent of what machine learning can do in their organization, finding agreement on a machine learning strategy among all top executives and bringing in external experts to advise the company on executing that strategy.


There are two types of machine learning experts that every company needs for a successful set of deployments. The first has a depth of knowledge on terminology and methodologies, and are normally data scientists and analytics specialists. The second type has a breadth of knowledge on how to communicate the potential of machine learning, converting results into insights and visualizations that make sense to managers on the front lines.

Both types should report to C-level execs, because only the highest-level, top-down viewpoint can identify gaps in the data that need addressed. Top execs can easily see how these data insights apply across departments to prevent data hoarding. This can be the province of a CIO or CTO, but more often results in the creation of a new role: the CDO (chief data officer).

Successful machine learning initiatives achieve buy-in through solving small but persistent problems and communicating positive results across the organization. Although driven by the C-Suite and aligned with a specific data strategy, machine learning projects can’t take root without support by front line managers and staff. All stakeholders, internal and external, should have a say in the shape of the implementation.


One of the best aspects of machine learning tools is that they free up management time, reducing risk while granting greater decision-making responsibility to lower level managers.

Customer-facing managers gain vast computation and resolution resources, allowing them to take on more complex decisions without the need for approval. Machine learning tools codify best practices so upper management can reduce their involvement to special exceptions. This encourages and develops trust, which is the cornerstone of a motivated workforce.


Among data scientists, it’s common to discuss machine learning applications in terms of three levels of complexity: starting with description building, moving into prediction and finishing with prescription.

The first level is defined by databases which describe past behavior such as sales growth numbers and market penetration. Many companies are comfortable with this process and have been doing it for years.

The second level is less common but rapidly gaining acceptance. Prediction involves building models of most likely outcomes for more intelligent investments and earlier launches of preventative measures. The best machine learning algorithms can still pull valuable insights from incomplete or unstructured data.

The third level, however, is where far-sighted business leaders are now concentrating their attention. Prescriptive applications predict behavior and make recommendations about how businesses can profit from that behavior. This involves usage the most advanced machine learning done by executives experienced in data interpretation. This collaboration is at the heart of the biggest changes in the future of work. AI and machine learning algorithms can distinguish patterns in seemingly random data, but require business leaders to construct objectives that make sense of those patterns.


In the end, as the WEF’s Fourth Industrial Revolution analysis would suggest, human intelligence and machine learning will merge to create something we don’t yet have a name for. The biological and the digital each enter this equation with their own strengths. For now, what matters most is that business leaders lay down a strategic path to incorporate the best aspects of machine learning before they get left behind with now-antiquated 20th century technology.