Despite decades of scientific research, incredible advances in deep analytics and AI, and no shortage of good intentions, many organizations still struggle to select and develop the leaders they need to navigate increasingly complex and unpredictable business challenges. Markets are volatile, uncertainty is constant, and leadership quality matters more than ever. Yet many firms still fail to identify and elevate the best (or at least right) leadership talent available.
Contrary to what many people think, more often than not, the problem is not a shortage of capable leaders. Rather, it is a failure of the systems designed to identify, develop, and advance them, which simultaneously results in selecting and investing in the wrong candidates, with tragic consequences for organizations.
Consider a glaring example: women. For several years now, women have represented more than half of the workforce in most advanced economies. They also outperform men in higher education across the OECD, earning the majority of university and postgraduate degrees. In many countries, women also achieve higher average grades and completion rates.
Research further shows that women tend, on average, to score higher on several of the leadership capabilities organizations claim to value most. These include emotional intelligence, self-awareness, collaboration, and integrity, and they have proven links to stronger leadership performance. Women are also less likely to display certain “dark side” personality traits such as narcissism, impulsivity, and excessive risk-taking, characteristics that are frequently associated with destructive leadership.
And yet, paradoxically, as careers progress, the leadership pipeline narrows dramatically in favor of men. Across most industries and regions, women remain significantly underrepresented in executive roles, board positions, and CEO seats. The economic implications are considerable. Various estimates suggest that narrowing gender gaps in leadership participation could add trillions of dollars to global GDP. Even for those who approach the issue purely from a commercial perspective, the cost of underutilized leadership talent is difficult to ignore.
Leading with science
One obvious way to address this paradox is to align leadership selection and development with science-based approaches, using data and evidence to select on leadership potential and develop the right leadership qualities so that performance actually increases and leadership talent improves.
If organizations adopted more rigorous, evidence-based approaches to identifying and developing leadership talent, two outcomes would likely follow simultaneously. The overall quality of leadership would improve, and the proportion of women in senior roles would rise. In other words, better systems would produce both stronger leaders and more balanced gender outcomes.
A science-based approach to leadership would also address the current lack of enthusiasm for many traditional diversity initiatives over recent years. Too often, such popular initiatives emphasize good intentions rather than measurable results. For instance, programs designed to encourage women to “lean in,” be more confident, network more strategically, or emulate traditionally male leadership styles may offer useful advice in some contexts.
But these approaches focus on adapting individuals to flawed systems rather than improving the systems themselves. In a meritocratic environment with systems designed to accurately detect leadership potential, there is surely no need to teach people (women or men) to self-promote, manage up, take credit for others’ achievements, and become successful corporate politicians. Similarly, workshops on unconscious bias or well-meaning diversity pledges may raise awareness without meaningfully changing how organizations identify and promote leadership talent.
So, what would a science-based approach look like?
1. Make selection gender-blind (focus on talent, not gender)
Many diversity initiatives begin with targets rather than diagnoses. Organizations decide they want more women in leadership roles and then design programs intended to move the numbers.
The difficulty is that the underlying systems used to identify leadership talent often remain unchanged.
Promotion decisions in many organizations still rely heavily on informal nominations, personal sponsorship, subjective impressions, and traditional archetypes of leadership. These processes tend to reward visibility, confidence, and political fluency rather than actual leadership effectiveness. When talent identification depends on intuition or familiarity, leaders tend to select people who resemble those already in power.
A more effective alternative is to remove gender from the equation altogether and focus instead on measuring leadership potential and performance with scientific rigor. Decades of research in organizational psychology show that structured, evidence-based assessments are far better predictors of leadership effectiveness than intuition or informal recommendations. Validated psychometric assessments, structured interviews, behavioral simulations, and performance analytics provide a far more reliable signal of who actually has the capabilities required to lead.
Today, advances in data science, analytics, and (well-designed and validated) AI make it possible to take this approach even further. Organizations can analyze large datasets on performance, leadership behaviors, decision patterns, team outcomes, and career trajectories to identify the attributes that truly predict leadership success. When these models are applied consistently and objectively, they reduce reliance on subjective judgments that often distort talent decisions.
In effect, this approach creates a gender-blind evaluation of leadership potential.
The outcome is not only fairer but also more effective. When organizations measure leadership talent rigorously rather than relying on human intuition, two things tend to happen simultaneously: leadership performance improves, and the candidate pool becomes more balanced.
An analogy can be found in professional sports scouting. Teams that rely on reputation and instinct frequently overlook high-performing athletes. Teams that rely on objective performance metrics often uncover talent others missed. Leadership pipelines operate in much the same way. To be more specific, if men and women do not differ meaningfully in leadership potential, the solution is not to “fix” one group or train individuals to mimic traditional leadership stereotypes. The solution is to improve the systems used to identify leadership talent in the first place, irrespective of gender.
2. Measure ROI and impact
Another common mistake is treating diversity initiatives as symbolic commitments rather than operational experiments. Organizations frequently introduce mentoring programs, leadership accelerators, or training initiatives designed to support women. Yet, they rarely measure whether those programs actually improve leadership outcomes. In most areas of business, initiatives that fail to produce measurable impact are quickly reassessed. Diversity initiatives should be subject to the same discipline.
A more effective approach is to treat these programs like strategic investments. Track outcomes such as promotion rates, leadership performance scores, employee engagement, retention, and business results. In other words, evaluate whether interventions improve leadership quality, not just representation. When organizations measure outcomes rigorously, two things tend to happen. Ineffective initiatives are identified and discontinued, while successful ones can be expanded and refined.
In short, evidence replaces ideology, science replaces intuition.
3. Remove structural barriers
Identifying leadership talent is only half the battle. Even the most rigorous selection systems will fail if the environment into which leaders are promoted contains structural barriers that distort how performance is evaluated and rewarded.
In other words, attracting capable candidates or selecting individuals with genuine leadership potential is pointless if the evaluation system itself is biased, or if the organizational context prevents people’s talents from being fully expressed.
Many such barriers are not deliberate or visible. They tend to emerge from the way organizations measure success, assess reputation, or distribute opportunities.
Consider a few common examples. Research has shown that identical leadership behaviors are often interpreted differently depending on who displays them. Assertiveness in men may be described as confidence, while the same behavior in women may be labeled abrasive or aggressive. Similarly, collaborative leadership styles may be undervalued in environments that equate leadership with dominance or visibility.
Another source of distortion comes from reliance on reputation-based evaluation systems. Leaders are often promoted based on perceived status, visibility with senior executives, or informal sponsorship networks. These mechanisms tend to favor individuals who resemble those already in power, while disadvantaging unconventional candidates or those operating outside dominant networks.
In addition, traditional performance metrics may capture outputs without adequately measuring the value leaders actually create. For instance, individuals who inherit high-performing teams or favorable market conditions may appear more successful than leaders who quietly transform underperforming units. When organizations evaluate outcomes without accounting for context, they risk rewarding circumstance rather than capability.
This is where scientific approaches and modern data tools can help. Advanced analytics and AI allow organizations to examine performance data at a much deeper level than traditional evaluations permit. Instead of relying primarily on reputation or visibility, organizations can analyze how leaders influence measurable outcomes: improvements in team performance, retention, innovation, decision quality, or long-term value creation.
These tools can also detect systematic biases embedded in evaluation systems. By analyzing patterns in performance reviews, promotion decisions, and feedback language, organizations can identify whether certain groups consistently receive different types of evaluations, slower career progression, or narrower development opportunities.
Equally important, data-driven approaches allow organizations to move beyond measuring success as a static outcome and toward measuring value added.
For example, rather than simply rewarding the leader of the highest-performing business unit, organizations can examine how much improvement that leader generated relative to where the team started. Instead of evaluating managers solely on revenue or output, companies can analyze the extent to which they improved employee engagement, reduced turnover, strengthened collaboration, or accelerated innovation.
When evaluation systems focus on value created rather than status accumulated, leadership assessments become both fairer and more predictive. Technology, often viewed as a potential source of bias, can therefore become an important tool for reducing it. When combined with scientific measurement and transparent metrics, data-driven evaluation systems help ensure that leadership advancement reflects genuine capability rather than familiarity, reputation, or historical advantage.
In short, identifying talent is essential. But removing the structural barriers that obscure and constrain that talent is what ultimately allows it to flourish.
4. Redesign succession systems
Leadership pipelines are shaped not only by evaluation systems but also by succession processes. Too often, succession systems are designed to preserve the status quo. They unintentionally favor candidates who resemble existing leaders or who followed traditional career paths. If organizations want to broaden leadership pipelines, they must rethink how those systems operate.
First, they need to create more pathways to the top. High-potential leaders should gain enterprise-level experience earlier in their careers. Many women, particularly those coming from functional roles such as HR, legal, or marketing, are less likely to receive early P&L responsibility or cross-enterprise assignments. Without those experiences they may later be perceived as lacking the experience required for senior roles. Deliberately assigning stretch roles earlier helps develop enterprise leadership capability and expands the pool of future executives.
Second, succession processes should default to inclusion rather than self-selection. Research suggests that men are somewhat more likely to assume they are ready for senior roles, while women may wait to be invited. When succession requires individuals to nominate themselves, this dynamic can unintentionally reinforce the gap.
When qualified leaders are automatically considered for succession pipelines, organizations signal that a broader group of individuals is seen as potential leadership talent.
Third, organizations should make the informal rules of leadership advancement more transparent. Many talented leaders do not lack capability; they lack access to the unwritten norms that influence leadership selection.
Mentorship can help bridge this gap by making these norms visible.
5. Embrace cognitive diversity
Another effective way to improve gender diversity and leadership talent is to focus on something broader: cognitive diversity.
Cognitive diversity refers to the range of thinking styles, perspectives, skills, and values that individuals bring to decision-making. Teams composed entirely of people with similar backgrounds and temperaments often fall prey to groupthink. By contrast, teams that incorporate a wider variety of perspectives are better able to challenge assumptions, identify blind spots, and improve the quality of decisions.
Importantly, cognitive diversity often leads naturally to demographic diversity. When organizations deliberately assemble teams with varied perspectives, leadership styles, and problem-solving approaches, gender balance and other forms of diversity frequently follow.
Some people argue that emphasizing cognitive diversity is merely a way of sidestepping demographic diversity. In reality, the logic runs in the opposite direction. One of the main reasons organizations value gender diversity in the first place is precisely because individuals with different backgrounds and life experiences tend to bring different ways of thinking. The purpose of diversity is not simply representation, but the broader range of perspectives that improves judgment and decision quality.
Rather than treating diversity as a compliance objective, organizations should therefore treat it as a strategy for improving organizational effectiveness, innovation, and long-term performance.
6. Treat leadership development as a continuous system (rather than a one off treat)
Finally, organizations must recognize that leadership development is continuous rather than episodic. No single workshop, mentoring program, or leadership course produces a CEO. Leadership capability develops over time through repeated stretch experiences, feedback, and reflection.
Efforts fail when they are treated as isolated initiatives. They succeed when development is embedded in the broader system through which talent is identified, developed, and advanced. Organizations that make meaningful progress approach leadership development the way engineers optimize complex systems: they introduce changes, measure outcomes, refine their approach, and repeat. Like with compound interest rates, small improvements accumulate over time.
From intentions to outcomes
Discussions about gender diversity often drift into predictable camps. Some frame the issue primarily as one of fairness or representation, while others dismiss many diversity initiatives as symbolic gestures that achieve little in practice. Both positions, however, tend to overlook a simpler and more practical point.
Organizations perform better when they are able to identify, develop, and deploy the best leadership talent available. If a significant portion of that talent remains overlooked or underutilized, the system is not working as intended. The cost is not only social or reputational. It is operational. Seen from this perspective, improving gender diversity is less about correcting outcomes directly and more about improving the systems that generate those outcomes in the first place.
Paradoxically, one of the most effective ways to achieve more balanced leadership may be to focus less on gender itself and more on the quality of leadership selection and development. Measure talent with greater precision. Evaluate interventions with evidence rather than aspiration. Remove structural barriers that distort how performance is judged. Broaden pathways into leadership roles. And design teams that benefit from genuine diversity of thinking.
When leadership systems become more objective, data-driven, and evidence-based, two things tend to happen at once: leadership quality improves and representation becomes more balanced, not just in terms of gender, but also in terms of cognitive diversity.
In the end, intentions matter far less than system design. Organizations rarely get the leaders they hope for. They get the leaders their systems are built to produce.
