The question deserves a straight answer, and the honest one has three parts. The research evidence supports coaching more strongly than skeptics assume. It supports it less strongly than the industry's marketing implies. And what the evidence says most clearly is that the effect depends far more on conditions than on method.
I come to this literature as a physician, which colors how I read it. In medicine we are trained to be suspicious of self-reported outcomes, alert to publication bias, and unimpressed by studies without a control group. Applying that lens to the coaching literature is a useful exercise, because it produces a more accurate picture than either the advocates or the cynics offer.
What the Research Actually Shows
Several meta-analyses of workplace and executive coaching have been published over the past fifteen years, pooling results across dozens of studies. Their findings converge on a consistent picture: coaching produces positive effects across multiple outcome categories, and those effects are generally small to moderate in size.
The outcome domains where effects appear most reliably are these:
- Goal-directed self-regulation. The capacity to set, pursue, and adjust goals shows some of the most consistent improvement. This is unsurprising, since it is the mechanism coaching most directly trains.
- Individual performance and skills. Improvements are reported consistently, though performance is frequently measured by self-rating or by the rating of a manager who knows coaching occurred.
- Wellbeing and coping. Reductions in stress and improvements in resilience appear across studies, with effects that tend to be modest but durable.
- Work attitudes. Engagement, commitment, and job satisfaction improve, though these are the outcomes most vulnerable to expectation effects.
A small to moderate effect size is not a trivial finding. Many accepted interventions in medicine and organizational psychology operate in the same range. But it is also not the transformational claim that coaching marketing tends to make, and the gap between the two is where most disappointment lives.
Where the Evidence Is Weak
Intellectual honesty requires naming the limitations, and they are substantial.
Self-report dominates
A large share of coaching outcome data comes from the person who received the coaching, often collected shortly after an engagement they chose and valued. People who have invested time and money in something tend to report that it helped. This does not mean the reports are false. It means they cannot carry the full weight of the claim on their own.
Controlled trials are scarce
Randomizing executives to coaching or no coaching is expensive, organizationally awkward, and rarely done. Most studies compare before and after, which cannot separate the effect of coaching from the effect of time, of the promotion that prompted the coaching, or of the leader's own decision to take their development seriously.
Publication bias is likely
Studies showing no effect are less likely to be written up and less likely to be published. In a field where much of the research is conducted by people with a professional stake in the answer, this deserves more weight than it usually receives.
Heterogeneity is enormous
"Coaching" in these studies covers everything from six sessions of goal-setting with an internal HR practitioner to a year of intensive work with a highly trained external professional. Pooling these into a single effect size tells you about the average of a category so broad it barely coheres.
What Actually Predicts a Good Outcome
The more useful finding buried in this literature is that variance between engagements dwarfs the average effect. Coaching works very well for some people and not at all for others, and the factors separating them are reasonably well established.
Voluntary entry. Leaders who sought coaching themselves do substantially better than leaders who were sent. Being assigned a coach as a corrective measure creates a defensive posture that the work then has to overcome before it can begin. If you are considering coaching for someone else, this finding deserves real weight.
Goal specificity. Engagements organized around a defined objective outperform open-ended ones. "Become a better leader" is not evaluable. "Build the capacity to run a leadership team I did not hire" is.
The working relationship. Across the helping professions, the quality of the alliance between practitioner and client is one of the most reliable predictors of outcome, and coaching is no exception. This is not the same as rapport or comfort. It means shared understanding of the task, agreement on the goal, and enough trust to tolerate being challenged.
Correct identification of the problem. This one appears less often in the literature and matters more than any of the others in my clinical experience. An engagement aimed at the wrong problem does not produce a small effect. It produces none, and it consumes the months during which the real problem could have been addressed.
When Coaching Reliably Fails
Three failure modes recur, and each is predictable in advance.
The first is misidentified clinical material. A leader presenting with low motivation, difficulty deciding, and loss of interest may be describing a values misalignment. They may equally be describing a depressive episode, which coaching cannot treat. I have met executives who spent a year in developmental work on what was, on examination, a treatable illness. The coaching did not fail because coaching does not work. It failed because it was the wrong intervention, and nobody in the room was trained to notice. This is the practical argument for clinical depth in this work, and it is the premise of our approach to executive wellness coaching.
The second is a structural problem dressed as an individual one. An executive who cannot get results from a team with unclear authority, contradictory incentives, and an absent sponsor does not have a leadership deficit. Coaching them is an expensive way to avoid fixing the structure.
The third is coaching used as documentation. When an organization has already decided a leader will not survive and assigns coaching to establish a record, no outcome is achievable. Leaders can usually sense this, which is its own reason the work does not take.
The Reasonable Conclusion
Executive coaching works, in the same qualified sense that most interventions aimed at complex human behavior work. It produces real but moderate average effects, with wide variance driven by conditions that are largely knowable in advance.
The practical implication is not to be more or less skeptical of coaching as a category. It is to be considerably more careful about the specific engagement. Enter voluntarily. Define the goal so precisely that failure would be visible. Choose a practitioner whose preparation matches the actual depth of the material. And spend the first hour making sure the problem has been correctly identified, because every other variable is downstream of that one. If you would like that first hour to be a rigorous one, you can start here.