Competency-based coach training and assessment implies that coaching skills and effectiveness are closely related. But who is best placed to determine ‘effectiveness’? While there are some studies comparing coach and coachee evaluations of coaching, none compare a coachee’s evaluation with a coach trainer-assessor’s rating of the coach’s competency in the same encounter. Neither are there studies using coach, coachee and assessor triads. This paper reports on research that examined how closely the evaluations of coachees, expert-assessors and coaches correspond. The research used a novel multi-method approach to triangulation including Clean Language interviewing (CLI) to explore coachees’ experience and evaluation of coaching. Assessor and coachee evaluations of the same coaching session were often at variance, both in terms of descriptive evaluations and numerical ratings. This suggests that compliance – or not – to a coaching methodology does not necessarily guarantee coachee satisfaction. While coach and coachee ratings showed no clear differences, in every triad coaches rated their own coaching considerably better than did the assessor. Practical implications include the need for multiple sources of evidence to establish coach effectiveness and certification standards, the need for coaches to develop calibration skills so they can be more responsive to the coachees’ in-session evaluations, and the usefulness of CLI together with established tools in evaluation research.