There is a seductive logic to the idea that the most important data in an executive coaching engagement is the data about the coachee. Their psychometric profile. Their 360-degree feedback. Their self-assessment of strengths and development areas. Their history of performance ratings and career progression. All of this is real data and some of it is genuinely useful. The problem is the assumption embedded in the framing: that the coaching engagement is primarily about understanding and developing the individual, and that the individual is therefore the primary unit of analysis.
Marshall Goldsmith’s foundational insight in stakeholder-centred coaching turns this assumption on its head. The primary data in an executive coaching engagement is not the coachee’s self-understanding. It is the stakeholder’s experience of the coachee. Not because the stakeholder is always right or because the coachee’s perspective is unimportant. Because the leadership behaviour that the coaching is designed to develop exists, by definition, in the relationship between the leader and those they lead. It cannot be assessed in isolation, and it cannot be changed in isolation. The behaviour that matters is behaviour as experienced by others, and the evidence that change has occurred is evidence provided by the people who have observed and experienced something different.
This is more than a methodological point. It is a fundamental reorientation of what coaching is for. Coaching oriented primarily toward the leader’s self-understanding produces leaders who understand themselves better. Coaching oriented primarily toward the stakeholder’s experience produces leaders who lead better. These are related but not identical outcomes, and the distinction between them determines the design of the entire engagement.
Why self-report is insufficient and sometimes actively misleading
The human capacity for self-assessment is both more and less reliable than most coaching frameworks acknowledge. People are generally accurate about their intentions, fairly accurate about their broad preferences and characteristic styles, and significantly inaccurate about their actual impact on others in specific situations under specific conditions. This last category, impact on others, is precisely the category most relevant to executive leadership development and precisely the category where self-report is least reliable.
The research on this specific asymmetry is consistent across decades and methodologies. Leaders consistently overestimate how much they listen, how open they are to challenge, how equitable they are in their treatment of different team members, and how clearly they communicate their expectations and reasoning. They are not lying when they report these self-assessments. They are providing honest accounts of their experience from inside the interaction, which is substantially different from the experience of those on the outside of the same interaction. The gap between these two perspectives is not pathological. It is the normal product of the fact that we have direct access to our intentions and only inferential access to our impact.
This gap has specific implications for how coaching conversations are most usefully structured. Coaching conversations that begin with the coachee’s self-assessment and work outward from there are building their diagnostic foundation on the data source most likely to contain systematic error in precisely the dimensions that matter most. Coaching conversations that begin with structured, honest stakeholder input and work inward from there are building their diagnostic foundation on the data source most likely to reveal the specific gap between intention and impact that is most productive to address.
The stakeholder interview process that Marshall Goldsmith and his colleagues developed as the foundation of stakeholder-centred coaching is designed to generate this specific kind of data with sufficient specificity and enough representative coverage that it cannot be dismissed as the perspective of a single disgruntled colleague or a single difficult period. When twelve stakeholders across different functions, levels, and relationships describe the same specific pattern of behaviour in independent interviews, the leader is confronted with evidence that has a different epistemic character than a single coach’s interpretation of their self-report. It is evidence about what actually happens in the relationship, seen from outside the relationship, and it is difficult to attribute to projection or misunderstanding when its consistency across independent sources has been established.
The specific design of the stakeholder interview that makes it useful
The quality of the stakeholder interview process is the single most important determinant of whether the coaching engagement will produce genuine behavioural change or merely increased self-awareness. This is worth dwelling on because most organisations that use 360-degree feedback processes, which is a majority of large organisations, are using processes that are less effective than they could be for the specific purpose of generating coaching-relevant data, for reasons that are identifiable and addressable.
The most common limitation of standard 360-degree feedback processes is that they generate aggregate data about perceptions of the leader across a set of competency dimensions. This aggregate data is useful for some purposes. For the specific purpose of generating the foundation for genuine behavioural change in coaching, it has two significant limitations. The first is that it describes what people think about the leader rather than what they observe the leader doing. The second is that the aggregation process removes exactly the specific behavioural instances that would make the feedback actionable rather than merely evaluative.
The stakeholder interview process that is most effective for coaching purposes generates different data: it asks stakeholders specifically what behaviours they observe, in what contexts, with what effects on their own ability to work effectively. It asks for specific examples rather than general assessments. It asks what the leader does that helps the stakeholder do their best work, and what the leader does that makes it harder. It asks what one or two changes in the leader’s behaviour would have the most significant positive impact on the stakeholder’s experience of working with them. This specific, behavioural, impact-oriented data is the raw material from which the coaching can generate the kind of focused development goals that actually change behaviour rather than simply informing self-understanding.
The process of feeding back and the specific skill it requires
The skill with which the stakeholder data is fed back to the coachee is as important as the quality of the data itself. Stakeholder feedback that is delivered in a way that produces defensiveness, shame, or the premature resolution of ambiguity into a narrative that the coachee controls will not produce the genuine engagement with the data that behavioural change requires. The coachee who leaves the feedback conversation having found an explanation for why each piece of challenging feedback is either inaccurate or attributable to factors outside their control has protected themselves from the development that the feedback was designed to produce.
The specific facilitation skill required is the ability to hold the leader in contact with the data long enough for genuine engagement to develop, without allowing the contact to become so distressing that it activates the defensive routines that close down genuine reflection. This requires the coach to resist several specific pressures that the feedback conversation generates: the pressure to move quickly to solutions before the diagnosis has been genuinely absorbed, the pressure to qualify the challenging data in ways that reduce its impact in the service of managing the coachee’s discomfort, and the pressure to accept the coachee’s initial framing of the data as definitive rather than as a first response to be explored.
The most productive moment in a feedback conversation is often the moment of genuine surprise, when a piece of data lands in a way that is both unexpected and, on reflection, clearly recognisable. This moment of recognition, when the gap between the leader’s self-understanding and the stakeholder’s experience becomes concretely visible, is the developmental pivot of the engagement. It is the moment from which genuine change can develop, because it establishes, beyond the capacity for rational dismissal, that there is something real to address. The coach who can recognise this moment and work with it carefully, holding it with appropriate weight without turning it into an indictment, is doing the most important facilitative work of the entire engagement.
What this means for how progress is measured
If the stakeholder is the data at the beginning of the coaching engagement, they are also the data at the end of it. The measure of whether genuine behavioural change has occurred is not the coachee’s self-assessment of their development. It is the stakeholder’s assessment of whether specific, named behaviours have changed in ways that are observable and that have had the impact the stakeholder described needing.
Goldsmith’s feedforward process, the structured re-engagement with stakeholders at the midpoint and conclusion of the coaching engagement, is designed to generate this specific evidence. Stakeholders are asked whether they have observed change in the specific behaviours that were identified as most important, and their assessment is the primary measure of the engagement’s effectiveness. This process has two functions that are both important. The first is measurement: it generates real evidence about whether behaviour change has occurred at the level where it matters, in the actual working relationships of the leader. The second is accountability: the knowledge that stakeholders will be asked at the end of the engagement whether they have observed change creates a specific and powerful motivation for the coachee to make the behavioural changes visible in those specific relationships rather than treating them as internal development work that stakeholders may or may not notice.
The broader lesson of this approach is about the location of accountability in development processes. Development that is primarily accountable to the individual who is developing, measured by their self-report of growth and assessed by their chosen indicators of progress, is less likely to produce genuine behavioural change than development that is accountable to the people whose experience of the leader is what the development is trying to improve. The stakeholder-centred model places accountability where the outcomes live, which is in the relationships, and measures development by the evidence available in those relationships. This is both more demanding and more meaningful than the alternative.
The coach who asks what the coachee thinks about themselves is gathering one kind of data. The coach who asks what the stakeholder experiences when working with the coachee is gathering the kind that changes things.