Human beings are obsessed with change . As large scale scientific studies show , most of us would like to change at least some aspects of our personality , defined as consistent patterns of behaviors or habits that make us who we are and different from others. For example, I’d love to be less impulsive, excitable, or cynical, which is why I have been working on not being myself for many years. Likewise, most individuals want to grow, to develop new adaptations that make them a better version of themselves, and organizations are equally interested in transforming, which typically depends on their people’s ability to grow and evolve. At its core, change is not about becoming someone else. Rather, it is about becoming a less exaggerated or extreme version of yourself. Most of our strengths, when overused, become weaknesses . Confidence becomes arrogance, attention to detail turns into obsessional perfectionism, and resilience mutates into stubbornness or “false hopes” in the face of problems and challenges that undermine our potential. Real development is not reinvention but calibration. Think of it as the ability to regulate your natural tendencies so they fit the demands of the situation, or optimize your behavior to develop better adaptations and become a more versatile and effective version of yourself, keeping your limitations and flaws in check, especially in high-stakes or critical situations. Sadly, left to our own devices, we rarely change. Human behavior is remarkably stable. Personality traits show high levels of consistency over time, and even when people receive feedback, they tend to interpret it in ways that protect their self-image. We are biased toward seeing ourselves as better than we are , and we systematically underestimate the gap between how we see ourselves and how others see us. This is where coaching comes in… The evidence is clear: coaching works, but not always, and not equally for all. A landmark meta-analysis by Tim Theeboom and colleagues found that coaching has significant positive effects on performance, well-being, coping, and goal attainment, with effect sizes ranging from moderate to large. More recent meta-analyses confirm that workplace coaching consistently leads to positive organizational outcomes, particularly when it focuses on behavior change. The key insight is that coaching is not a placebo. It is a scientifically validated intervention. But like most interventions, its effectiveness depends on how it is designed and delivered, who it’s delivered by, and of course who the recipient of coaching is. Some coaching relationships are transformative. Others are pleasant but inconsequential, like chatting to a friend of hairdresser. The difference is rarely about whether coaching “works” in principle. It is about whether the right coach is matched to the right person, for the right goal, in the right way. Four factors Choosing a coach, therefore, should be treated as a high-stakes decision. Yet most people approach it casually, relying on reputation, referrals, or vague impressions of “chemistry.” That is not enough. There are four factors that matter most. 1)    First, personality chemistry and style fit. Coaching is fundamentally a relationship, and like any relationship, it depends on trust. But chemistry is not just about liking someone. It is about alignment between the coach’s style and the coachee’s needs. Some coaches are direct and confrontational, excelling at telling people what they need to hear , even when they really don’t want to hear it .  They challenge assumptions, provide blunt feedback, and push for rapid change. Others are more facilitative and supportive, helping individuals reflect and arrive at their own conclusions. Neither approach is inherently superior. The question is which one works for you. If you are defensive, overconfident, or prone to dismiss feedback, you may need a coach who is willing to confront you. If you are already self-critical or risk-averse, a more supportive approach may be more effective. The goal is not comfort, but progress. There is also a personality dimension. Research suggests that similarity can build rapport, but difference can drive growth. A coach who mirrors your worldview may feel comfortable but may not stretch you. A coach who is too different may create friction without insight. The optimal point is somewhere in between: enough overlap to build trust, enough difference to challenge your thinking. 2)    Second, method fit to goal. Not all coaching is the same, and not all goals require the same approach. If your objective is to improve a specific skill, such as communication or decision-making, a structured, behavioral approach with clear feedback loops may be most effective. If your challenge is more psychological, such as managing derailers, improving self-awareness, or navigating interpersonal dynamics, a deeper, more reflective approach may be required. Some coaches draw on cognitive-behavioral techniques. Others rely on psychodynamic frameworks, systems thinking, or data-driven assessments. Increasingly, coaching is also augmented by analytics and AI . None of these approaches is universally better. What matters is alignment with the outcome you seek. Too often, organizations adopt a one-size-fits-all approach to coaching, offering the same intervention regardless of the problem. This is equivalent to prescribing the same medication for every illness. It is convenient, but ineffective. 3)    Third, expertise and proficiency. The coaching industry is largely unregulated, which means quality varies significantly. Anyone can call themselves a coach. Not everyone should, and not everyone is qualified. It is not enough for a coach to have a preferred method. They need to be good at it. This includes formal training, relevant experience, and, critically, a track record of impact. Have they worked with people at your level? Do they understand your context? Can they translate insight into action? In an era where AI can generate generic advice instantly, the value of a coach lies not in providing information, but in interpreting it, contextualizing it, and applying it to your specific situation. As I have argued in I, Human , the differentiator in the AI age is not access to answers, but the quality of judgment applied to those answers. A good coach enhances your judgment. A mediocre one simply adds noise. 4)    Fourth, and avoid stagnation through measurement and iteration. Coaching should not be an open-ended, indefinite process. It should be structured, with clear goals, regular checkpoints, and measurable outcomes. This is where many coaching engagements fall short. They focus on conversations rather than outcomes. They create insight, but not change. Effective coaching requires experimentation. Try new behaviors, gather feedback, adjust, and repeat. It also requires measurement. Progress should be assessed not only through self-reports, but through observable indicators. These may include improvements in team engagement, changes in leadership behavior as captured by 360 feedback, enhanced performance metrics, or even hard business outcomes. Research suggests that coaching has its strongest impact on behavioral change, which is precisely what should be measured . Indeed, without measurement, coaching risks becoming what much of corporate life already is: well-intentioned but performative. The AI role In many ways, selecting a coach is similar to selecting a leader. You are making a bet on someone’s ability to influence behavior and drive outcomes. You would not make that decision lightly in a business context. You should not make it lightly here. In the age of AI, the stakes are even higher. As machines take over more of the cognitive heavy lifting, the human edge will depend on adaptability, self-awareness, and the ability to evolve. Coaching, when done right, can enhance all three. The right coach will not change who you are. They will help you become a more effective version of yourself. They will challenge your assumptions, expose your blind spots, and help you build the adaptations you need to succeed so that the best version of yourself shows up more frequently, and the worst version is contained or silenced, at least during critical work interactions.