A coach who is still necessary after two years has failed. I say that to every board that hires me, usually in the first conversation, and it is not modesty. It is the design specification. The deliverable of this work is not the sessions; it is the handover. The critical friend has to become internal — a voice the leader carries into the room alone — or nothing durable has happened, and what was bought instead was a very expensive, very pleasant relationship.
Buyers ask the right questions in the wrong order. They ask what the process looks like, what it costs, how success is measured. They rarely ask the question that actually decides whether the money was well spent: what has to be transferred out of my hands and into the building before I go, and how would we know it had been?
That is the honest test, and it is not administered in month six, when the leader is being observed and the calendar still holds a fortnightly appointment with someone who notices everything. Month eighteen. Under load. Alone.
The month-six illusion
Almost everyone performs well in month six. This is not cynicism either — it is how attention works.
Sustained, undivided, non-transactional attention is one of the rarest things at the top of an organisation. Nobody listens to a CEO the way a coach listens to a CEO. Everyone else in that week wants something: a decision, a budget, a signature, reassurance. The coaching hour is the only hour in which someone is fully present and wants nothing in return. That is not a small thing. For some leaders it is the first such hour in a decade.
And it produces change. Real change, often quickly. But you have to be careful about what you are looking at. A leader who is being attended to behaves differently. They are more reflective, slower to react, more curious about their own patterns — partly because they are doing the work, and partly because they know that on Thursday they will be describing the week to someone who will ask why. The observation itself is doing some of the lifting.
This is why I am suspicious of engagements that feel wonderful. The leaders who relapse hardest, in my experience, are the ones who experienced the coaching as being attended to rather than as being changed. They loved it. They said it was the best professional relationship of their career. And months after it ended the behaviour was back, because the behaviour had never been rebuilt. It had been borrowed against a relationship, and the relationship ended.
Dependency is the occupational hazard
The uncomfortable half of this is mine, not the client’s.
Being needed by a CEO is seductive. You are close to power, your calendar is full, your invoices are paid without argument, and a capable person tells you regularly that these conversations matter more than anything else in their month. There is a version of this profession that quietly optimises for that. Nobody plans it. It happens by drift — the engagement rolls over, the agenda softens into general reflection, and the sessions become a place to think out loud rather than a place where something is being changed and measured.
A coach has to stay permanently alert to becoming the very support the work is meant to make unnecessary. Practically, that means naming the endpoint at the start and defending it. When I set up an engagement I want the exit criteria written down before the first session, not discovered at the end — and I want them behavioural, not emotional. “The leader feels clearer” is not an exit criterion. “The stakeholders who named this behaviour at the outset now report it has changed, and the mechanism that told us so runs without me” is one. That distinction sits at the heart of the way I structure an engagement from intake to close, and it is the part most buyers skim.
What actually survives
So what carries the change after the last session? In my experience, three things, and only three reliably.
The mechanisms other people own
The most durable part of any engagement is the part that does not depend on the leader remembering to do it. This is the strength of the stakeholder-centred method I was trained in: the feedback loop is not a private habit, it is a social contract. The leader has told named colleagues what they are working on. They go back to those colleagues, at intervals, and ask how they are doing on that one thing. The colleagues expect the question. They notice when it stops.
That mechanism keeps running after I am gone because other people hold one end of it. A private intention dies quietly in a bad quarter. A commitment that four colleagues are actively monitoring is much harder to abandon, because abandoning it is visible. When a board asks me what they are actually buying, this is the honest answer, and it is also where the return on an executive coaching engagement sits — not in the sessions, in the machinery that outlives them.
The two or three questions
The second thing that survives is smaller and stranger: a handful of questions the leader has learned to ask themselves in the precise moments they used to react.
Not a framework. Not a model on a card. Two or three specific questions, tied to specific triggers, worn smooth by repetition. What am I protecting right now? — asked in the second before overruling a competent colleague. Whose problem am I solving, theirs or mine? — asked when the pull to rescue arrives. Is this urgent, or am I uncomfortable? — asked before the message goes out at eleven at night.
These land because they were built at the source. When a leader understands where a particular reflex comes from — the family system, the early role, the thing they learned to do to be safe or valued — the question stops being a technique and becomes recognition. They are not applying a tool. They are catching themselves, mid-move, in a pattern they can now see. That is the deeper layer of the work, and it is why behavioural change at this level cannot be trained in a workshop.
The named person with permission
The third thing is a person. One colleague — sometimes two, never five — who has been given explicit, specific, standing permission to tell the leader the truth.
Explicit matters. Every executive believes their door is open. Almost none of them have said, to a named individual, in words: when you see me do this thing, I want you to say so, in the moment, and I will not punish you for it. That sentence, actually spoken, is the transfer of the critical-friend function from outside the organisation to inside it. It is the whole logic of critical friendship — someone close enough to see clearly, safe enough to say it, loyal enough to bother.
I test this before I close an engagement. Who is it? What exactly did you tell them? When did they last use it? If the answer is vague, the handover has not happened, and the leader is about to walk out with a habit that only ever functioned in my presence.
Regression is normal, and it is not failure
Now the part engagements rarely prepare people for.
The change will slip. Under real load — an acquisition, a bad quarter, a resignation that hurts, an illness at home — the old behaviour returns, because it is the cheapest thing available. It costs nothing to fall back into a pattern you ran for thirty years. That is not backsliding as a character failure. That is what pressure does.
What separates the leaders who hold their gains from the ones who lose them is not willpower. It is recovery time. The first group notices within a week, names it out loud to the colleague who has permission, and comes back. The second group does not notice for a quarter, and by then the organisation has recalibrated around the old version of the leader and quietly stopped expecting anything else.
So I say it out loud before we finish: you will lose this at some point, probably during something difficult, and the measure of the work is not that it never happens but how fast you catch it. A leader who plans for regression treats it as information. A leader who was promised transformation treats it as proof that the whole thing was theatre, and gives up. This is also why I am blunt about who this work suits — it asks for a tolerance of being wrong in public that not everyone has, and no engagement can manufacture it.
Coming back is allowed. Needing to is the signal.
None of this means a leader should never work with a coach again. Roles change. A CFO who becomes a CEO is doing a different job with a different exposure, and new altitude surfaces new material. Coming back at a genuine threshold is sensible, and it is one of the better reasons a chief executive works with a coach at all.
The distinction is between returning for a new problem and returning because the last one never resolved. The first is a decision. The second is a symptom — and if it is the second, the honest conversation is not about scheduling more sessions. It is about what we both avoided the first time.
I keep a private measure of my own work, and it is not testimonials. It is whether, a year or two later, the mechanism is still running without me — whether someone in that building is still telling the leader the truth, and whether the leader is still asking for it.
If they are, the handover held. If they still need me for it, it did not.