Where Insight Runs Out
A leader can do everything right and still find themself stuck in the same place. They have read the books. They have sat through the trainings. They can describe, with real precision, exactly what they should be doing differently: staying calmer under pressure, delegating instead of absorbing every problem themselves, having the direct conversation instead of the comfortable one. None of that is the hard part. The hard part is that knowing what to do and doing it, consistently, under real pressure, are two different skills, and only one of them gets taught in a workshop.
This is where a lot of leadership development quietly stalls. A leader walks out of a program with a clear picture of who they want to become and a real intention to get there, and then Monday arrives, the same meetings fill the calendar, the same pressures return, and the old patterns are simply easier to fall back into than the new ones are to practice. Insight fades fast when nothing in the leader’s week is built to hold onto it.
What is usually missing is not more information. It is somewhere to practice the harder version of leadership out loud, in real time, with someone whose only job in the room is to notice what the leader cannot see about themselves in the moment. A leader can read about staying calm under pressure. Staying calm requires a live situation and someone paying close enough attention to name what just happened.
This is the role coaching is built to play, and it is worth being clear about what that role is not. Coaching is not a remedial step reserved for leaders who have fallen behind, and it is not a reward reserved for those who have already arrived. It is the structural piece that keeps everything else in a leader’s development from fading the moment the training ends and the real work resumes.
What Coaching Is (And Is Not)
What coaching looks like depends on who is giving it and who is receiving it, and a single style rarely works for everyone. One leader might need a coach who is direct and assertive, naming exactly what went wrong and why. Another might need time and space to talk through a situation before arriving at their own understanding of it. Both are legitimate coaching approaches. Applied to the wrong person, either one can do more harm than good.
Coaching, mentoring, and therapy are often used interchangeably, but they diverge in ways that matter. Coaching is a two-way, balanced relationship. The coach does not direct or manage the client but works collaboratively toward goals the client holds. Mentoring is not built on the same footing, since a mentor typically holds a position of authority (direct, indirect or implied) over the person being mentored, and the relationship leans more toward advice and sharing experiences than helped discovery. Therapy differs further still. It is prescriptive and diagnostic by design, built to treat clinical populations. A coach is neither a mentor nor a therapist, and the distinction is practical, not technical. A leader who walks into coaching expecting to be told what to do will misread what is happening in the room.
Coaching is also a widely researched and increasingly professionalized field, established enough that independent sources converge on nearly the same definition. De Haan and Nilsson (2023) frame coaching around what the client gains from it: a structured, ongoing process where reflection, inquiry, and honest challenge from another person help the client think more clearly and act more effectively over time. Graßmann, Schölmerich, and Schermuly (2019) describe the same essential relationship from another angle, built on equal footing between coach and client and organized around goals the client sets, not goals assigned from outside. Bozer and Jones (2018) add a practical anchor to both: whatever else varies across coaching styles and contexts, the outcomes being pursued must be the ones the client values, not results imposed by an employer or a program.
What falls outside that definition matters just as much. Health coaching and sports coaching are directive in ways coaching is not. Managerial coaching carries a different problem: it comes from someone who already holds power over the client’s job, which changes the goals being pursued, who sets them, and how freely the client can speak.
Yes, It Works. It Can Also Go Wrong.
Before asking what makes coaching effective, it is worth answering a more basic question directly: does it work at all? For a discipline this widely adopted, a real answer took a surprisingly long time to uncover. A meta-analysis restricted only to randomized controlled trials, the most methodologically demanding standard available, found a real, moderate effect of workplace coaching across leadership and personal outcomes (de Haan & Nilsson, 2023). Coaching is not a placebo dressed up as professional development. It produces measurable change.
What explains that effect is where the research gets more interesting than a simple yes. The same study proposes that coaching works through coregulation: an ongoing, mutual influence between coach and client that plays out moment to moment across a session and builds cumulatively across an engagement (de Haan & Nilsson, 2023). That framing matters because it locates the source of change in the relationship itself rather than in any single technique a coach applies. It also predicted that self-reported outcomes would show larger effects than outcomes an outside observer could verify independently, and that is exactly what turned up. That gap is not a flaw unique to coaching. It shows up across almost any intervention that relies on a person’s own account of their change, and it is a reason to treat client-reported growth as real but not as the final word.
None of this means coaching always goes the way a client hopes, and that deserves to be said plainly rather than downplayed. Coaching can also lead somewhere a client did not expect and would not have chosen. Across the research on this question, a large share of clients report at least one unintended negative effect from a coaching engagement, with estimates ranging from 57% when coaches report on their own clients to nearly 68% when clients report on themselves (Schermuly & Graßmann, 2018). That range on its own can sound alarming. The rest of the picture brings it back down to size.
Coaching can produce effects a client did not want, and that is not the same thing as coaching failing.
These effects turned up even in engagements clients rated as successful overall, and most of what gets reported has a small, short-lived impact. The comparison worth holding onto is to side effects in medicine. A treatment can work exactly as intended and still come with something uncomfortable attached, and that discomfort is not evidence the treatment failed.
Schermuly and Graßmann’s broader research program sorts these effects into six categories: personal well-being, social integration, performance, how the client evaluates their own work role, financial loss, and a catch-all for what does not fit elsewhere. The most frequently reported effect by far falls in the well-being category: a coaching conversation surfacing a problem more serious than anything coaching itself is equipped to resolve, reported by roughly a quarter of clients and coaches alike (23% of clients, 26% of coaches). That is a genuinely useful thing for a leader considering coaching to know going in. Coaching is built to develop capability, not to treat a clinical problem, and a good coaching relationship will recognize the difference and redirect rather than attempt to hold something it was never designed to carry.
There is one antecedent the research keeps returning to, and it is the same one that will anchor the rest of this paper: relationship quality was the most consistent predictor of whether negative effects occurred at all (Schermuly & Graßmann, 2018). Clients and coaches with a stronger working relationship reported fewer unintended effects, and this held up as close to the only variable that mattered once other factors were accounted for. Coaching works, on balance and with real effect. It can also occasionally cause harm. Both facts trace back to the same source.
What Makes Coaching Effective
Establishing that coaching works answers only part of the question most senior leaders have. The more useful question is why coaching changes behavior for some leaders and does nothing for others. The research points to a single relationship-level factor with more explanatory power than any other variable studied: the working alliance between coach and client.
Working alliance is not rapport in the casual sense. It refers to three specific things happening together: coach and client agree on what they are working toward, they agree on how they will get there, and they build a bond defined by trust and mutual respect. A meta-analysis synthesizing 27 samples and more than 3,500 coaching engagements found that the strength of this alliance was tied to every outcome category the researchers measured, and the relationship held regardless of whether the client or the coach was doing the rating, regardless of how many sessions the engagement ran, and regardless of whether the coach was a seasoned professional or new to the work (Graßmann et al., 2019). Coaching outcomes do not hinge on finding an elite coach or extending an engagement indefinitely. They hinge on whether coach and client can build a working relationship together.
Coaching outcomes do not hinge on finding an elite coach or extending an engagement indefinitely. They hinge on whether coach and client can build a working relationship together.
What that relationship predicts is worth sitting with. Working alliance was most strongly tied to how satisfied clients felt with coaching and how effective they judged it to be. It also showed a real, meaningful connection to whether clients reached their goals, though a more modest one (Graßmann et al., 2019). That gap between the two is informative. Alliance most powerfully shapes how a leader experiences the coaching relationship: whether it felt worthwhile, whether it felt like it worked. Its grip on the harder, more concrete outcome of goal attainment is real but looser, because goal attainment depends on more than the relationship alone. It depends on the goal itself, the client’s circumstances, and everything happening outside the coaching conversation. A strong alliance is not a guarantee of results. It is the mechanism most consistently associated with getting them.
Why would a relationship carry this much weight? The explanation researchers point to is an exchange, not a technique. When a client trusts a coach enough to be genuinely honest, including about the parts of a problem that are uncomfortable to admit, the coach gains the accurate picture needed to be useful. In return, the coach offers something the client could not generate alone: an outside perspective, a challenge to an assumption, a question that reframes the problem. Each side is giving something the other needs, and that exchange is what allows the work to move (Graßmann et al., 2019). Without trust, a client withholds the parts of a problem that actually matter, and a coach is left working with an incomplete picture no matter how skilled they are.
This same idea shows up under a different name in research on coaching relationships more broadly, described as a kind of coregulation between coach and client, a back-and-forth influence that plays out moment to moment inside a session and cumulatively across an engagement (de Haan & Nilsson, 2023). A meta-analysis of coaching outcomes measured against randomized control groups found a moderate, reliable effect for coaching overall, and the same coregulation model successfully predicted which conditions strengthened that effect: coaches with real credibility, clients bringing genuine motivation into the room, and trust built between them (de Haan & Nilsson, 2023). Different language, same conclusion as the alliance research: what happens between coach and client is doing more work than either person is doing alone.
Trust of this kind is not automatic, and it is not owed to a coach by virtue of title or credential. Research on trust in professional relationships distinguishes between trust based on someone’s track record and trust based on someone’s character, their integrity, their consistency, whether their actions match what they say (Mayer et al., 1995, as cited in Bozer & Jones, 2018). Coaching depends heavily on the second kind. A client tends to not want to talk about the real issue, the one tangled up with ego or fear or uncertainty about their own judgment, until they believe the person across from them can be trusted with it. Vulnerability of that kind is not a byproduct of good coaching. It is a prerequisite for it.
There is a further reason these findings matter, and it connects directly to a pattern raised earlier: coaching engagements do sometimes produce effects clients did not want, and working alliance is where the difference showed up. In the same meta-analysis, a stronger alliance was tied to fewer unintended negative effects, not more of them (Graßmann et al., 2019). That is not a coincidence. A client who trusts their coach is more likely to say when something feels off, more likely to name a problem the coaching relationship cannot hold, and more likely to be met with the judgment needed to redirect rather than push forward regardless. A weak alliance removes that safeguard. The same relationship that predicts whether coaching helps also appears to predict whether it avoids doing harm.
None of this suggests that the specific technique a coach uses does not matter, or that any two people who like each other will produce results equal to a well-designed engagement. It suggests something more precise: technique operates through the relationship, not around it. A well-timed question lands because the client trusts the person asking it. A challenge to a leader’s thinking is received as useful rather than as an attack because the relationship can hold that kind of friction. The mechanism connecting coaching activity to leadership change is not the intervention itself. It is the relationship carrying it.
Coaching as a Structural Support
A study of executives coached during a period of major organizational upheaval found that the outcomes coaching directly targeted moved the most. Goal attainment, the thing the coaching sessions were explicitly built around, showed by far the largest change, nearly four times larger than the next largest effect (Grant, 2013). Resilience moved too, along with change readiness, leadership self-efficacy, and a decrease in depression, though these effects were smaller. Stress, anxiety, and general workplace satisfaction, none of which the coaching sessions specifically addressed, did not move at all.
That pattern makes sense once the boundaries of coaching are clear. A coach is not a psychologist or a physician. The work is not diagnosis and it is not treatment. What a coach brings is awareness, structure, and a concrete plan for improving something specific, and that is a fundamentally different kind of intervention than one built to treat anxiety or depression directly. So it should not surprise anyone that coaching moved the goal it was built around far more than it moved outcomes it was never structured to touch. Coaching can still meaningfully affect things like stress or resilience along the way, as this study shows, but that impact rides alongside the work coaching is actually built to do. It is a byproduct, not the design.
Coaching moves what it is aimed at. The things left unaddressed stay where they were.
That has a direct implication for how coaching fits into the rest of a leader’s development. It works best as a support to something, not as the whole of something. A leader building new habits, learning to sit with the discomfort of accountability, or developing sharper self-awareness is doing work that coaching can accelerate and sharpen. Coaching alone was not built to carry the entire weight of that change, and the research on its targeted effects backs that up directly.
One model for how coaching fits into a leader’s broader development is Intentional Change Theory, which describes sustained development as a sequence: a leader gets clear on who they want to become, compares that against who they actually are right now, builds a concrete plan to close the distance, and then practices new behavior in the real world (Boyatzis et al., 2013). What holds that sequence together, in this model, is a trusting relationship that provides support at each step. Coaching is not positioned as a single conversation that produces insight and then ends. It is the relationship that keeps a leader moving through each stage of that process rather than stalling out after the first one.
A related model makes a similar case from a different angle, focused specifically on what makes a coach effective rather than on the change process itself. Drawing on self-determination theory, this framework proposes that a coach’s central job is helping a leader feel a sense of autonomy, relatedness, and competence throughout the change process, not just handing over a diagnosis and letting the leader take it from there (Taylor et al., 2020). Sustained change, in this view, depends less on any single coaching insight and more on a leader continuing to feel capable and supported long enough for a new behavior to stick.
Both models agree with what the study already showed: coaching’s power holds up when it stays connected to something ongoing, whether that is a goal, a plan, or a relationship a leader can keep returning to. A single strong session can open a leader’s eyes. Whether that insight survives a return to normal work depends on what happens around it afterward, and that is exactly where coaching earns its place as a structural support rather than a stand-alone fix.
From Coaching to Change
Coaching works. The research at this point is clear on that. What it actually takes to make it work, and where its limits sit, is the more useful thing to understand.
A few findings stand out as the ones most worth carrying forward. Coaching produces a real, measurable effect, though a targeted one rather than a sweeping one. It improves what it is aimed at, and a leader who wants to see change in a particular area is better served naming that area clearly than hoping coaching will improve everything at once. The working relationship between coach and client is not a soft or secondary detail. It is the single strongest predictor of whether coaching produces the outcomes a leader is looking for, and it is also the strongest protection against the kind of unintended effects that can come from opening up difficult territory without enough trust in place to hold it. Coaching can surface problems it was never built to solve, and a good coaching relationship recognizes that boundary rather than pushing past it. None of this makes coaching fragile. It makes it precise, and precision is what makes it useful.
Coaching does not fail by accident, and it does not succeed by accident either. Both outcomes trace back to the same relationship.
What emerges across this research is a picture of coaching as something that works best in relationship to everything else already happening around a leader. It sharpens self-awareness that has already begun to form. It gives structure to habits already in motion. It holds a leader accountable to a plan they have already started building. Coaching does not replace that surrounding work. It reinforces it, for as long as a leader keeps returning to it, and that ongoing return is where its value actually lives.
From Insight to Action: The Ascent Diagnostics Difference
Most organizations treat coaching as a remedy: something you bring in when a leader is struggling, or a perk you extend to the ones you’re betting on. Both framings miss what coaching actually does. Its strongest effects come from creating the one condition most development never does — sustained accountability for a change the leader has chosen to make.
A workshop ends. A framework gets filed. Coaching is the structure that continues beyond the workshop to ask the question, among others: did you do the thing, and if not, what got in the way? Behavioral change fails in the gap between intention and the next hard, uncomfortable action to take; that gap is where a coach earns their place.
But coaching without measurement is just conversation with good intentions and, at best, anecdotal evidence of progress. Ascent Diagnostics often begins any engagement with the ResilienceBuilder® assessment to establish where a leader’s ability to perform at a high level while managing stress is pressure tested and measured. Other engagements benefit from 360-degree feedback that gives the leader and the coach a shared, brutally honest starting point — not just the leader’s account of how they lead, but a read on how others “in their orbit” experience them.
Coaching is the fifth of the five support tools in the Ascent Behavior Change model, and its placement is deliberate. It is not a separate intervention layered on top of the others. It is the wrap around that holds the other four in place — accountability that turns a diagnosed pattern, a reframed thought, a practiced response, and a genuine intention into a change that survives real world practice and ultimately identity transformation.
The shift from “I wish I could be more/less…” to “I am now seen as…”
As comforting as it might be to think otherwise, there is no finish line. Leaders change because something…as well as someone…held them accountable long enough for the new response to become the new normal.
And the transformation is complete.


