When the CEO Wants to Proceed and the Data Says Wait
The hardest governance moment in a clinical-stage company is not the one that makes headlines. It is not fraud, not a safety catastrophe, not a boardroom coup. It is a quiet meeting on an ordinary agenda, in which a capable and committed CEO recommends advancing a program — into the next dose, into the pivotal trial, into a larger and more expensive commitment — and the data on the table are ambiguous enough that a reasonable person could read them as a reason to wait.
This is the situation this article is about. It is common, it is genuinely hard, and it exposes the deepest tension in the board's role: the board hired this CEO precisely because of their conviction and their willingness to push, and now that same conviction is pointing at a decision the data may not support. Governing this moment well requires the board to do something subtle — to take the CEO's judgment seriously without being captured by it, and to slow a decision without undermining the person the company depends on to make hard calls.
Get it wrong in one direction and the company burns its remaining capital advancing a program that the data were quietly warning against. Get it wrong in the other and the board becomes a brake on every ambitious decision, second-guessing a CEO into paralysis. Neither failure is hypothetical, and both are avoidable.
Why This Happens to Good CEOs, Not Just Bad Ones
It is tempting to frame this as a problem of management integrity — the CEO who spins bad data to protect their position or their story. That version exists, and it is a different problem with a different solution. But it is not the common case, and treating every proceed-versus-wait tension as a question of honesty will poison the relationship the board most needs to protect.
The common case is the good CEO, acting in good faith, subject to pressures that genuinely distort judgment. It is worth naming those pressures precisely, because the board's job is to compensate for them, not to condemn them.
The first is commitment. A founder or CEO who has spent years on a program has internalized a belief in it that is close to identity. That belief is an asset — it is what got the company this far — but it makes disconfirming data psychologically expensive to absorb. Ambiguous results get read charitably. The optimistic interpretation feels like realism from the inside.
The second is momentum and cost. Stopping to wait, or to run a confirmatory step, costs time and money the company may not feel it has. There is a powerful pull toward advancing simply because advancing is what the plan called for, the team is mobilized, and the runway assumes it. The sunk cost of everything already invested makes the forward path feel cheaper than it is.
The third is the narrative burden the CEO carries with investors. The CEO has told a story — to the last round, to the board, to prospective partners — and a decision to pause reads externally as a stumble. The CEO is not being cynical in feeling this pressure; they are responding to a real feature of their job. But it means the person recommending "proceed" is also the person for whom "wait" is most costly to admit.
None of these pressures make the CEO wrong. Sometimes the CEO's conviction is exactly right and the cautious reading of the data is the timid one. The point is only this: the CEO is the single least neutral party in the room on this specific question, through no fault of character. That is precisely why the decision cannot rest on the CEO's judgment alone — and why the board exists.
The Board's Job Is Not to Have a Better Opinion
Here is the trap boards fall into. Confronted with proceed-versus-wait, directors reach for their own read of the data and try to out-analyze the CEO. On a clinical-stage board, this is usually a losing game. The CEO and the management team know the program more deeply than any director does. A board that tries to win the argument on the scientific merits will lose it, or worse, will win it for the wrong reasons and take ownership of a call it was not equipped to make.
The board's job is not to have a better opinion than the CEO. It is to ensure the decision process is sound — that the question has been framed honestly, the disconfirming evidence has been given real weight, the alternatives have been genuinely considered, and the decision is being made against pre-committed standards rather than in-the-moment rationalization.
This reframing matters enormously in practice. It moves the board off the terrain where it is weakest (independent scientific judgment) and onto the terrain where it is strongest and most legitimate (the integrity of how consequential decisions get made). It also depersonalizes the conflict. The board is not saying "we think you're wrong about the science." It is saying "before we commit, we need to be sure we've tested this decision the way a decision this consequential deserves to be tested."
The Questions That Do the Work
A board does not need to match the CEO's scientific depth to govern this moment. It needs to ask a disciplined set of questions and insist on honest answers.
What would we have needed to see to say wait — and did we define it in advance? The most powerful single discipline is the pre-committed decision criterion. If, before the data arrived, the board and management agreed on what a "go" result looks like and what a "no-go" or "pause" result looks like, then the current data can be measured against a standard set when no one had a stake in the answer. If those criteria were never set, that absence is itself the finding: the company is about to make a pivotal decision with no benchmark except the CEO's interpretation. Where pre-committed criteria exist and the data fall short of them, the board should be extremely reluctant to accept a fresh interpretation invented to clear the lower bar.
What is the strongest version of the case for waiting — and who in the room is making it? If the only person articulating the cautious reading is a director doing it half-heartedly, the case for waiting has not actually been heard. The board should insist that someone credible — a management scientist, an independent advisor, a data monitoring perspective — lay out the best case for caution in full, on its own terms, not as a foil for the CEO to rebut. A decision that has only heard one side argued seriously has not been tested.
What does proceeding foreclose, and what does waiting cost? These are asymmetric and should be examined separately. Proceeding on ambiguous data often commits a large fraction of remaining capital and, if the program then fails, may leave nothing for the confirmatory work that would have resolved the ambiguity. Waiting costs time and money but frequently preserves optionality. The board should force an explicit accounting of both branches rather than accepting the implicit framing that forward motion is the safe default.
Is there a cheaper way to resolve the ambiguity? Frequently the choice is presented as binary — commit now or lose momentum — when a third path exists: a small confirmatory step, an additional cohort, a focused analysis, an agency conversation that would materially reduce the uncertainty before the large commitment. The board asking "is there a way to buy more certainty for less than the cost of being wrong?" often surfaces an option management, in its forward lean, had stopped considering.
If we proceed and it fails, will this decision look defensible in hindsight? Not as a matter of liability, but as a matter of honesty. Asking the room to imagine the post-mortem — a program that failed after the board waved through ambiguous data — is a clarifying discipline. If the honest answer is "we'd struggle to explain why we didn't pause," that is the board learning something in time to act on it.
Slowing a Decision Without Breaking the CEO
The relational dimension of this is as important as the analytical one, and it is where boards most often mishandle an otherwise correct instinct.
A board can be right to want caution and still do lasting damage in how it imposes it. If the intervention reads as "we don't trust your judgment," the board has spent relationship capital it will need later and has taught the CEO to bring the board cleaner, less candid presentations in future — exactly the opposite of what good governance requires. The CEO who feels second-guessed learns to manage the board rather than inform it.
The way through is to locate the disagreement in the process, not the person. The board is not overruling the CEO's scientific judgment; it is exercising its own legitimate responsibility to ensure a consequential and hard-to-reverse decision has been tested against a standard. A CEO who is treated as a partner in that testing — "help us get comfortable that we've stressed this properly" — responds very differently from one who is treated as a suspect.
It also helps to separate the timing question from the merits question. Often the board does not actually need to say "no." It needs to say "not yet, and here is the specific thing that would let us say yes." A concrete, bounded ask — see this analysis, get this agency input, add this cohort — gives the CEO a path forward and gives the board the assurance it needs, without either side having to be declared wrong. The unresolvable version of this conflict is rarer than it feels in the room.
And when, occasionally, the disagreement is genuine and unbridgeable — the board is not persuaded and the CEO will not wait — that is no longer a data question. It is a governance question about whether the board has confidence in the CEO's judgment on the decisions that matter most. That is a real question, and a board should be willing to face it. But it should arrive there deliberately, having done the process work first, not as the reflexive endpoint of every hard conversation.
The Underlying Principle
The proceed-versus-wait moment is where the board's role is most easily misunderstood. It is not the board's job to be smarter about the science than the people running the company. It is the board's job to ensure that the company's most consequential and least reversible decisions are made against honest standards, with the disconfirming case fully heard, by a process that compensates for the entirely human pressures acting on the person who has to recommend the call.
A CEO's conviction is not the problem to be solved; it is the quality the company was built on. The board's contribution is to make sure that conviction is tested where testing is cheap — in the boardroom, before the commitment — rather than in the market, after it. Done well, this does not weaken the CEO. It gives the CEO something valuable: a decision they can stand behind because it survived real scrutiny, rather than one they will spend the next two years hoping was right.
Lawrence Fine is CEO of AGCP Farmacêuticos and has advised on licensing, regulatory, and partnership strategy across the pharmaceutical and advanced materials sectors.