What Supervision Actually Achieves – A Look at the Research
(Reading time: 3 minutes)
Supervision is taken as read in both coaching and psychotherapy. As a supervisor myself, I have long wondered how much of that assumption the evidence will actually bear. In coaching, the research base is still thin. In psychotherapy, sturdier data have begun to accumulate – and they tell a more interesting story than everyday consensus would suggest.
The First Meta-Analysis, Published in 2025
Until recently, the effectiveness of supervision had only ever been reviewed qualitatively. That changed in November 2025, when a research group led by Schreyer published the first quantitative synthesis: 20,727 records screened, 32 intervention studies included, 16 of them randomised controlled trials (Frontiers in Psychiatry). Three findings stand out for practice.
Supervision works – and it works markedly well against no supervision at all.
Compared with control groups who received none, the effects on supervisees’ competence (Hedges’ g = 0.95) and on the quality of their working alliance with clients (g = 1.12) were large and statistically significant. Compare one supervision format with another, though, and the differences melt away. So whether supervision happens at all makes a considerable difference; which format you choose is still an open question scientifically.
Dosage matters
Strip out interventions of fewer than three sessions and the effect on competence climbs from g = 0.47 to g = 0.74, while the variation between studies disappears altogether. On that reading, the occasional one-off session is not really supervision in any meaningful sense.
Working from the original beats working from the account
Studies by Florian Weck’s group consistently find that supervision drawing on recordings or direct observation produces stronger gains in competence than supervision built on the supervisee’s verbal report (Journal of Consulting and Clinical Psychology). That matches my own experience. An account of a situation is always an edited version of it – and what has been edited out is very often the point.
What the Research Does Not Show
An effect on client outcomes has yet to be demonstrated convincingly. The meta-analytic estimate points the right way but falls just short of significance. That has less to do with an absence of effect than with the difficulty of detecting one: it has to travel two stages, from supervisor to coach and from coach to client. Having surveyed twenty-five years of research reviews, C. Edward Watkins puts it plainly – what we have amounts to “proof by association” rather than genuine evidence of effect (Counselling and Psychotherapy Research).
The distinction matters, not least in how we describe what we offer. There is a solid case to be made on professional development, relationship quality, satisfaction, lower drop-out rates, burnout prevention and staying power in the profession. Promise improved client outcomes, and you have gone further than the evidence will bear.
The Uncomfortable Finding
Supervision is not automatically a good thing. In a study of 363 supervisees, 93 per cent were receiving supervision that fell short of minimum professional standards, and for 35 per cent it was, on the criteria applied, harmful (The Counseling Psychologist). The striking part is not the size of those figures but the fact that most of those affected did not recognise the problem themselves. In a field as lightly regulated as coaching supervision, that is worth pausing over – and reason enough not to leave quality to chance.
What This Means for Coaching Supervision
The evidence specific to coaching supervision remains slender. One systematic review turned up just thirteen empirical publications and a single randomised field study (International Coaching Psychology Review). That study did find, though, that the after-effects of the work weighed more heavily on coaches who went without supervision – a clear pointer to how much protection the format affords.
Four principles carry over directly from the psychotherapy research, and they shape how I work as a supervisor:
- Regularity, not firefighting – three sessions at the very least, seven to ten if you can
- Work from the original – recordings or direct observation rather than retrospective narrative, wherever the setting and consent allow
- Mind the working alliance within supervision – it has the strongest evidence of any single ingredient
- Establish the need before settling on a format – active supervision strategies pay off where there is a genuine gap to close
Supervision is not a ritual of professionalism. It is an intervention with demonstrable effects, and one with design conditions worth knowing about. If you are thinking about supervision for your own practice, do get in touch.
Peter Ammann (27.08.2026)
