It is 4:07 on a Friday and your strongest IC Slacks "got a sec?" Six years on the team. Built the system the rest of the org runs on. They walk in, sit down, and say they are giving notice. You did not see it coming. The conversation will be the most expensive single one you have this year — and the moment it was lost was three quarters ago, in a coaching cycle that never matched what they actually needed.
Most coaching content is built for the median case. The frameworks the better managers know — Schein's Humble Inquiry, Bungay Stanier's question discipline, the BID feedback model — are correct, important, and calibrated to lower coachee defenses, increase psychological safety, and get the middle 60% to engage. They are also not what your top performer needs. They have not had defenses up in years. They are sitting across from you because they want one specific thing: someone willing to push them into the version of themselves they cannot yet see. Most managers default to the median motion anyway. The high performer politely keeps showing up. Three quarters later, they Slack "got a sec?" — and it is too late.
The pattern almost no one names out loud
High-performer attrition is the most expensive single category of voluntary turnover. The replacement search, the lost institutional knowledge, the cascading effect on the people who watched a top performer leave and quietly updated their own resumes — the cost dwarfs almost any other talent line item. And it is almost always traceable to the same root cause: coaching cycles calibrated to the median case being applied to someone who needs a structurally different motion.
The strongest performer on your team does not need encouragement, normalization, or an open-ended question about how things are going. They have asked themselves that question several times this year and quietly answered it: fine, but unchallenged. What they need is the conversation almost no one is willing to have with them — the one that names the version of the role they have been avoiding, the strength that is becoming a blind spot, and the place where they are about to plateau. And the part the data backs up most decisively: they will not ask for that conversation. They will simply leave to find a place that has it.
The empirical case for asymmetric coaching
Two pieces of research, taken together, change the math. Aguinis and O'Boyle's 2012 meta-analysis in Personnel Psychology — re-tested across 198 datasets — found that individual performance in knowledge work is power-law distributed, not normally distributed. Roughly 20% of contributors produce something on the order of 80% of the output. Allocating coaching capacity proportionally to headcount understates the leverage available from the top tier by an order of magnitude. The second is from Google's internal People Analytics work — Project Oxygen — which spent more than a decade ranking the manager behaviors most strongly correlated with team performance and retention. "Be a good coach" sits at the top of that list. Not just for the median performer. Especially for the top one, who has the option to leave.
Why most coaching content fails the top quartile
The coaching canon is largely correct — for the median coachee. Edgar Schein's Humble Inquiry is built around lowering the asker's status to invite genuine answers from someone who has not yet trusted that the conversation is safe. Michael Bungay Stanier's seven essential questions are calibrated to slow a busy manager down enough to actually hear someone before responding. The BID feedback model is built to land hard truths without breaking the relationship that delivered them. All three are good. All three assume a coachee whose default is some version of guarded.
The high performer's default is not guarded. It is bored. They have done the work to be heard, they have built the trust capital with you, and they have already lowered their defenses several times this quarter without it leading anywhere they did not already know. A "tell me more about what's on your mind" question reads to them as a manager going through the coaching motions. They will give you the answer they think you are equipped to receive, the meeting will end on schedule, and nothing will happen. Lowering defenses is the wrong objective. The right one is to put a constraint on the table that they have not yet considered, and let them sit with it.
The Expert Trap fires here too — but for a different reason
The Expert Trap — the manager's reflex to tell instead of ask — is the most common failure mode across all coaching tiers. With a high performer it fires for a particular set of reasons, and the diagnosis is uncomfortable.
"They should already know." The manager assumes the high performer has internalized the lesson and skips the conversation that would have made it explicit. The lesson stays implicit. The high performer infers, correctly, that their growth is no longer being thought about by the person whose job it is to think about it.
The manager is intimidated. The high performer has gone deeper into their domain than the manager has, and the manager — not wanting to be exposed — defaults to "you're handling it well." Marshall Goldsmith calls this the inability to provide a successful person with the feedback that the next move requires; the senior coach who was useful to them at level three has nothing left to offer at level five.
"Don't fix what isn't broken." The single most damaging move at the top tier — and the one that masquerades as good management. The high performer is delivering. There are five other things the manager could be working on. They quietly de-prioritize the one conversation that would have changed the high performer's trajectory, and pour their attention into the people who are visibly struggling. By the time the high performer hands in notice, no one is surprised they leave; everyone is surprised it happened "now."
"The higher you go in an organization, the more your problems are behavioral. They are not technical."
Marshall Goldsmith · What Got You Here Won't Get You There (2007)
What top performers actually need: three structurally different moves
The coaching motion that retains the top quartile is harder, not nicer. It rests on three moves that are absent from almost every coaching conversation calibrated to the median.
1. Calibrate to the next edge. Anders Ericsson's research on expert performance, distilled in Peak, returned again and again to a single idea: deliberate practice happens at the edge of competence — the place where the practitioner cannot yet do the thing reliably. For a high performer, that edge has moved past where it was last year, and almost no one in their orbit has bothered to update their model of where it now is. The coach's job is to find it. Not the next OKR. Not the next promotion. The next version of the role they have not yet attempted, and the move within it they cannot yet make. Naming it out loud is most of the work.
2. Provoke, not probe. Liz Wiseman's Multipliers identifies the Challenger discipline — the leader who introduces a problem the team has not yet decided is theirs to solve, and lets that decision happen in the room. Daniel Goleman's HBR work on the Coaching style describes the same move from the other side: not surveying the coachee's current frame, but introducing a constraint that forces them past it. With a high performer, a probing question ("how are you feeling about the new scope?") reads as a diversion; a provocation ("the version of this role that ships in eighteen months will not include the work you are best at — what does that mean for you?") reads as a coach who is still in the conversation. The first invites a polite answer. The second forces a real one.
3. Reflect their pattern back. The pattern that drove the high performer's rise is the same pattern that, untouched, becomes their derailer. CCL's multi-decade research on executive derailment makes the case clearly: a strength taken to excess becomes the visible reason for the senior plateau. The high performer cannot see it from inside the pattern. They will keep doing the thing that worked, harder, until the diminishing returns become the dominant signal — usually too late. Goldsmith's entire executive-coaching practice was built on a single move: holding up the mirror. Almost no internal manager is willing to. The coach who can — by naming the trait, the cost, and the version of the leader on the other side of letting it go — earns a different relationship from any other manager the high performer will ever have.
In Parlare's coaching framework, every conversation has a Basement — the floor where the coachee names the cost, the fear, or the version of themselves they have been avoiding. With a high performer, the Basement is not a stuck problem. It is the version of the role they have not yet attempted, the strength that is becoming a blind spot, or the regret they will have if they stay where they are.
Most managers skip it because the high performer is delivering. The session ends on Floor 3 — action and advice on the work in front of them. Three quarters later, the high performer Slacks "got a sec?"
Score consequence: the Expert Trap penalty fires hardest in this tier when "ask another question" is used as a reflex to avoid provoking. The session score reflects which floor the coach actually visited.
Five questions to ask the top quartile
You do not need a separate certification to break the median-coaching default with your strongest reports. Five questions, deployed in any of the next three one-on-ones, will visit the Basement on most high-performer conversations.
1. "What would 'great' look like that you have not yet asked yourself how to do?" — Most high performers have a private answer to this one. They have not said it out loud because they are not yet sure they can do it. The question grants permission to name it.
2. "Where do you feel yourself about to plateau?" — High performers see plateaus before their managers do. They will not volunteer the diagnosis. Asking directly converts a quiet worry into a coachable conversation.
3. "What is the thing you are best at that is becoming your blind spot?" — The Goldsmith question. Strength-becomes-weakness is invisible from inside the pattern. The coach who is willing to ask it earns a different kind of trust.
4. "If you stayed in this role for another two years exactly as it is, what would you regret?" — The retention question, asked correctly. Most "stay" conversations are calibrated for someone with one foot out the door; this one is calibrated for someone whose feet have not moved yet, but whose attention has. The answer is the work.
5. "What would I be missing if I told you to just keep doing what you are doing?" — The mirror. It is the explicit invitation to provoke the manager rather than be coached by them, and it gives a high performer permission to name the version of the conversation no one is having with them. It also operationalizes the trusted-advisor relationship: the manager is now in the room as a thinking partner, not a status-checker.
After each, hold the silence. Seven seconds, minimum. The follow-up question you have queued is almost always weaker than the answer the high performer is about to land on their own. Their answers will arrive late, in fragments, and frequently in a tone you have not heard from them before. That is the signal that the Basement is open.
The deeper move: harder coaching, not nicer coaching
Coaching the top quartile is not nicer coaching. It is harder coaching. It is being willing to say the thing the high performer cannot yet say to themselves — about the version of the role they have been avoiding, the strength that is becoming a blind spot, the place they are about to plateau, the regret they will have if they stay where they are. Almost no manager defaults to that conversation. The high performer notices its absence and quietly starts looking elsewhere.
The trusted-advisor relationship that retains a top performer through the next five years of their career is built on a single move, repeated: the coach is the one person in their orbit willing to provoke them at the edge of their domain. Not survey their current frame. Not normalize the discomfort. Provoke. Hold the silence. Reflect the pattern back. Watch them name the thing, three turns later, that no one — including them — has said out loud yet.
The math works in one direction. A top quartile that feels actively coached compounds for years. A top quartile that feels unattended to is one Slack message away from leaving. The expensive part was never the conversation. It was the conversation that did not happen.