What Has Machine Learning Got to Do with Leadership? (Quite a lot, actually.)By: Reece Akhtar
Organizations invest money in assessing, measuring, and modifying a leader’s behavior in the hopes that a few nudges can push the needle. This is not a wasted effort; there is much evidence to demonstrate that this is a worthwhile exercise. But can we do a better job at identifying and understanding a leader’s talents? Yes. And machine learning may be the key.
You may have heard of studies that have used algorithms that translate your Facebook likes, Twitter feed, or even your facial expressions and voice into highly accurate models of personality, personal values and talent. These models are remarkably accurate due to machine learning algorithms that can spot patterns in data that are too complex for humans to understand. There is no reason we cannot leverage similar approaches to better understand leadership talent. The more we can utilize data and empirical models of behavior, the better our predictions and evaluations of leadership potential will be.
In light of this and inspired by a recent paper that used machine learning to demonstrate that the “Big Five” personality profiles (i.e. Openness, Conscientiousness, Extroversion, Agreeableness, and Neuroticism) could be organized into four discrete types, we sought to identify typologies within our Readiness for ScaleSM (RFS) model of leadership. The RFS model contains nine behavioral dimensions that can be organized into three domains: the ability to lead the business, lead people, and lead oneself. You can learn more about this model and its dimensions here. With thousands of possible configurations, it is a comprehensive model of leadership. So, we wondered: can machine learning methods reveal the core roles a leader can play? To answer this question, we leveraged a dataset containing more than 600 leaders who were assessed against the nine RFS model dimensions. Using a popular machine learning algorithm1, we can describe leaders as playing one of the four roles:
Strategists score high on the Leading Business domain, while their scores are somewhat average on the Leading People and Leading Self domains. These leaders can be described as bright, technical, business-savvy thinkers who do a great job at following industry trends to shape strategy, are capable of making decisions that positively impact the entire business, and can balance pursuing short-term goals with building capability for the future. While their orientations toward data, systems, and processes can help create growth and progress, it does get in the way of influencing and connecting with others.
Influencers score high on the Leading People domain, below average on the Leading Business domain, and average on the Leading Self domain. They can be described as interpersonally skilled and relationship-focused individuals. They are effective at driving clarity and alignment among followers and key stakeholders, delivering compelling messages that galvanize and inspire action, and creating cultures that positively shape behaviors and decisions throughout the organization. Influencers have a complementary style to Strategists: despite an affinity and connection with people, they struggle with strategic and technical thinking.
Executors typically score in the mid ranges across the nine RFS model dimensions, meaning that they are steady and well-rounded leaders. While they are neither rock stars nor truly incompetent, Executors can be trusted to achieve their goals and get stuff done yet are likely to struggle to influence and set direction for the entire enterprise. They do their best work managing small- to medium-size teams where they can implement another’s vision and strategy.
The Change Agent
Change Agents score above average across all RFS model dimensions. Such individuals have a style that enables them to lead at scale and create organizational growth. That is, they set an ambitious and achievable strategy for the organization, inspire and galvanize action in others, and remain emotionally aware and productive when under extreme pressure. These leaders are driven, interpersonally sensitive, and highly curious. That said, they are not without their flaws: they tend to be fairly bold and mischievous, which can lead to overconfidence and harmful rule breaking if left unchecked.
As you read through these descriptions, you can agree that they are intuitive and make practical sense. We are sure that you can identify with one of the four types—and certainly sort your colleagues among these types. The takeaway, however, is that machine learning methods were used to churn through a large and complex database to identify a handful of core leadership styles that were not immediately obvious. The more data we collect from leaders and from more diverse places, the greater the role machine learning can play in our leadership assessment and development efforts. Such technologies can help us become less biased and better judges of talent and ultimately improve our ability to help organizations achieve their goals.
1A full description of the methodology is available upon request.
Readiness for Scale℠ is a service offered through Executive Bench®. To find out how RHR International's Executive Bench® services can help your organization plan for critical leadership transitions, please contact Jessica Foster.
SIGN UP TO RECEIVE BLOG POSTS.
Our insights and thinking, direct to your inbox.