What should a board do about a narcissistic CEO? Diagnose the phase before the person, gather stakeholder reality independently of the CEO’s narration, act while results are still good, and choose deliberately between three options: govern the pattern, develop it through structured coaching, or plan succession. The one guaranteed failure is the default option, which is waiting.
Supervisory boards rarely call me about a mystery. By the time a chair is searching for this article, the diagnosis is usually complete and has been for some time: the CEO who was magnificent in the crisis has not noticed the crisis ending, the executive committee has gone quiet in a way engagement surveys do not capture, two capable senior people have left for reasons that were explained a shade too smoothly, and board meetings have become performances with an audience of eight. What the chair actually wants to know is not “is this narcissism?” but “what do we do that is smarter than either capitulation or war?” This guide is that answer, drawn from Let’s Talk Leadership, from the research Martin Appelo and I published in Rescue the Alpha Wolf, and from years of sitting in exactly these conversations.
First, subtract the myth
Boards inherit the popular villain story, and it produces two equal and opposite governance failures. The first is the witch hunt: a checklist circulates, confidence is reclassified as pathology, and the board amputates its most capable change-maker to the applause of everyone who was tired of being pushed. The second is the romance: the board decides genius has its price, reclassifies the damage as intensity, and discovers years later what the filtered information was hiding. Both failures come from treating narcissism as a verdict rather than a profile.
The working frame from our research is more useful. Narcissistic dynamics at the top are a risk profile, not a diagnosis; they arrive bundled with genuine capability (we call the bundle talented narcissism); their value is phase-dependent; and their signature cost is a deficit in reciprocity that degrades exactly the systems a board depends on for truth. Your CEO is not a monster or a messiah. He is, most likely, an unpolished alpha wolf: superb in fire, expensive in peacetime, and running on fuel that has a history.
The phase question comes first
In Let’s Talk Leadership I argue that organisations must retire the fantasy of the universal CEO. Every organisation moves through phases (formation, growth, consolidation, crisis, renewal) and each phase calls on a different leadership script. In fire you need an alpha wolf: hypersensitive to threat, decisive under ambiguity, willing to own the downside. When the smoke clears, that same person will, by temperament, reach for the next storm. So before any conversation about the person, the board owes itself an honest answer about the phase. If the company is genuinely in crisis, the profile that worries you may be the only thing keeping it alive, and your task is governance, not replacement. If the company entered consolidation two years ago and the CEO is still governing by emergency, the mismatch is the problem, and it will not improve by itself, because storms can be manufactured and the pattern is good at manufacturing them.
What the board is not seeing, and how to see it
The structural effect of an unpolished narcissistic CEO is information distortion, and boards consistently underestimate how completely it applies to them. Bad news learns to travel slowly and dress carefully. Forecasts acquire systematic optimism. The executives one layer down, who know precisely what is happening, have run the cost-benefit analysis on telling you and concluded, rationally, that the messenger’s fate is not worth it. Meanwhile the CEO’s own narration is confident, fluent and, in the precise sense I described in Let’s Talk Leadership, a projection rather than a recording: his mind is a projector, not a camera, and the board is watching the film.
The countermeasures are unglamorous and non-negotiable. Structured, normalised contact between board members and the layer below the CEO, established as routine governance rather than as a crisis signal. Exit interviews conducted independently and actually read at board level, because the best people leave first and they leave informed. And when the situation warrants it, a formal stakeholder measurement: confidential interviews with fifteen to twenty-five people around the CEO, conducted by an outside professional, reported in aggregate. This is the instrument I use to begin every coaching engagement, and boards are routinely astonished by the gap between the picture it produces and the picture they held. Assessments entertain, references defend and interviews seduce, as I put it in the book, but nothing predicts behaviour like behaviour, and nothing measures it like the people who experience it daily.
The three real options
Govern the pattern. Appropriate when the phase genuinely needs the profile, or as an interim stance. Concretely: tighten the reciprocity the pattern will not supply voluntarily. Contract explicit decision rights and consultation duties. Put dissent physically into the room by protecting the executives who provide it, visibly, so the organisation updates its cost-benefit analysis. Insist on the mirror structures described above as a standing condition, not a punishment. Expect this to be resented; the resentment is diagnostic, and its intensity tells you how far along the pattern is.
Develop the pattern. Structured coaching can work, and the conditions under which it works are knowable in advance; I have written them up separately in Can Narcissistic Executives Be Coached? The short version: the CEO must be feeling the cost privately (coaching imposed as pure punishment fails; the novel Wahlberg opens with exactly that failure, a CEO sent to the internal coach by his supervisory board, and the engagement dies in weeks), the method must be stakeholder-measured rather than self-reported, and the board must fund a serious engagement rather than a gesture. When those conditions hold, the transformation is real and the organisation keeps a leader whose drive no competitor can hire. When they do not hold, coaching becomes an expensive way to postpone the third option, and postponement always bills at compound interest.
Plan succession. Appropriate when the phase mismatch is structural, the cost is compounding, and the felt urgency on the CEO’s side is absent. Do it while results are still good, which is precisely when it feels unjustifiable; the alternative is doing it after the information distortion has spent your option space. And design the succession with the pattern in mind: an unpolished alpha wolf does not hand over an apparatus he built to be necessary to. The psychology of that resistance is its own subject, which I have treated in Why Founders Cannot Let Go.
The conversation itself
However you proceed, one board conversation is unavoidable, and boards reliably conduct it badly, either as an ambush with a dossier or as a hint so diplomatic it is never received. From practice: hold it with facts, not adjectives; behaviour and consequence, not character. Anchor it in the stakeholder data, which is arguable with reality rather than with the chair. Frame development as investment in a valuable asset, which is both true and the only frame the pattern can accept without existential defence. And put a boundary in the room with a date on it, because the pattern tests boundaries as reliably as water finds cracks, and a boundary without consequence is, to this profile specifically, an invitation.
The quotable version: a board that waits for a narcissistic CEO to fail is not being patient, it is being billed monthly for its own postponement.
Frequently asked questions
How do we know if our CEO is a narcissist or just a strong leader? Watch four things over time: what happens to people who correct him publicly, where credit and blame travel, whether disconfirming information still reaches the top unfiltered, and the gap between stated values and behaviour. The full comparison is in Narcissism or Strong Leadership? How to Tell. Avoid diagnostic labels entirely; govern behaviour and consequences.
Should the board force the CEO into coaching? Forced coaching as punishment fails almost universally; the pattern converts it into persecution or theatre. What a board can legitimately do is set behavioural expectations with consequences, offer serious development as the supported path, and let the CEO choose it. Choice, even constrained choice, is a working condition of the method.
How long should we give a coaching engagement to show results? A serious engagement runs about twelve months, with stakeholder re-measurement toward the end providing the evidence. Meaningful directional signals (visible behavioural experiments, changed meeting dynamics) appear within a quarter. No signal by then is itself a result.
What if the narcissistic dynamic is in the chair, not the CEO? More common than the literature admits, and structurally harder, because the mirror structures point the wrong way. The same principles apply: phase diagnosis, independent stakeholder reality, behavioural contracting. External facilitation is close to mandatory, since nobody inside the system can hold the mirror safely.
Is it safe to keep a narcissistic CEO through a genuine crisis? Often it is the right call: crisis is this script’s native phase. Keep the governance tight, agree in advance what the end of the crisis looks like, and put the transition conversation in the calendar before the crisis ends, because afterwards the definition of crisis becomes negotiable and the CEO holds the pen.
Arvid Buit is a master executive coach (ICF, EMCC, APECS, Marshall Goldsmith SCC), founder of TRUE Leadership, and author of Let’s Talk Leadership, Red de Alfawolf (with clinical psychologist Martin Appelo) and Wahlberg. He works with CEOs, board members and owners of companies from €100 million revenue upward. This guide is organisational psychology for governance contexts, not clinical advice, and no content here constitutes a diagnosis of any individual. The full field guide · How an engagement works