The Case Against Confidence in Board Meetings

We sat in a board meeting last spring where two divisional presidents presented back-to-back. The first walked through her segment with the kind of conviction the room had been trained to reward: clear numbers, sharp narrative, no audible hedging. The second was quieter. She said, twice, "we don't yet know," and once, "this is the part of our plan I'd revise hard if the next two months go a certain way." She gave the room a structured picture of what she was sure of and what she wasn't.

After the meeting, the chair pulled us aside. He liked the first president. He worried about the second. We told him he had it exactly backwards, and that the board's instincts on this were probably costing the company eight figures a year.

Confidence is what boards can observe

The bias is structural, not personal. A board meets a senior executive for maybe sixty minutes a quarter. Inside that hour, the board cannot directly observe how well the executive runs an operating cadence, how she handles a 2 a.m. crisis, how her team feels about her on a Friday afternoon. What the board can directly observe is the executive's bearing in the room: how clearly she presents, how confidently she answers, how unshaken she looks when challenged.

So the board, doing what any reasonable observer would do, uses what it can see. It treats confidence as a signal of competence, because confidence is the signal available. Over many meetings, across many executives, this filter compounds. The leaders who get promoted into bigger roles are, on average, the leaders who present more confidently. The leaders who get pushed out, or who never get the big seat, are disproportionately the ones who hedged in the wrong meeting.

This filter has a name in the literature. Phil Tetlock spent two decades tracking expert political forecasters and found an unsettling inverse: the more confident the expert, the worse the accuracy. The hedgehogs — confident, theory-driven, one-big-idea thinkers — outperformed the foxes on television and underperformed them on actual prediction. The pattern shows up everywhere we have looked for it. Confidence and accuracy are, in moderately complex domains, negatively correlated.

A board that selects on confidence is therefore selecting, on average, against accuracy. Not always. Not in every case. But on average, and at the margin, and across enough decisions, the math becomes meaningful. The executive who presents at 90% conviction on a call that deserves 60% looks better in the meeting and is wrong more often in the world. The executive who frames the same call honestly — "60%, conditional on X, would revise to 40% if Y" — looks weaker in the meeting and is right more often. The board sees the meeting. The shareholders pay for the world.

The distortion is not hedging

Let us be careful here. We are not making a case for hedging. Hedging is what executives do when they want to keep their options open in case the decision goes badly. It is vague on purpose. "We need to be mindful of a number of factors." "There are headwinds and tailwinds." "It depends on execution." Hedging is rhetorical insurance. It is, if anything, worse than false confidence, because it commits to nothing and can be reread to mean anything.

What we are making a case for is structured uncertainty. The version that sounds like: "We believe X with about 70% confidence. The two things that would move us toward 90% are A and B. The two things that would move us below 50% are C and D. If we see C by July, we kill the program." This is the opposite of hedging. It commits, with precision, to a specific probabilistic claim and a specific update rule. It can be checked. It can be wrong. It is the version of executive communication that turns the board meeting from a performance into a working session.

Vague hedging protects the speaker. Structured uncertainty protects the company. Most boards have been trained to penalize the first, and they accidentally penalize the second along with it.

The asymmetry matters. The executive who delivers structured uncertainty is taking on more reputational risk, not less. She is going on the record with a specific number and a specific update rule. Six months from now the board can look at the meeting minutes and see whether she was right. The hedger has put nothing on the record that can be checked. The false confidence merchant has put a single number on the record and will spend the next six months reinterpreting it. Only the structured-uncertainty executive has actually made herself accountable.

This is why a board that learns to read the difference ends up with a different bench. It is not a small reform. It restructures who gets heard, who gets promoted, and how the company learns from its decisions.

How a board can change the norms

The good news is that the fix is largely within the board's control. The norms of board communication are set by the board, in the room, in real time. A few moves we have watched work.

Ask for ranges, by default. When an executive presents a forecast, the chair asks, every time, "what is your 80% range, and what would shift it?" Asked once, this looks like a one-off. Asked at every meeting, for two quarters, it becomes the format. Executives who come in with a single number learn to come in with a range. The point estimate stops being the language of the room.

Reward the update. When an executive comes back at the next meeting and says, "I told you 70% last quarter; I am now at 50% because of what happened with Z," the chair thanks her. Out loud. In front of the other directors. This is the cheapest possible signal that the board values updating over consistency. The executives in the room will learn it within one cycle.

Penalize the rewrite. When an executive comes back and quietly reframes what she said last quarter, the chair names it. Gently, but specifically: "Last quarter I had you at 70% on this; the minutes show that. We are now at 50%. What changed?" The reframe is the dark cousin of the update. Updating is the executive learning. Reframing is the executive protecting herself from having been wrong. The first deserves reward. The second deserves friction.

Read the minutes. This sounds trivial; in practice it is the largest single change a board can make. Most board minutes record decisions but not the confidence framing that produced them. The minutes that matter — the ones that make the company smarter year over year — record the confidence levels, the update rules, and what each executive said would change them. A board that takes minutes this way is a board that is, six months later, able to ask the questions that matter. A board that does not is a board re-litigating from memory.

Separate the meeting from the deck. Decks are the artifact of false confidence. They are produced to be persuasive in a room of partial attention. The board that asks executives to come in for thirty minutes without slides — to talk through what they are betting on, what would change their minds, where the pre-mortem landed — is a board that finds out, very quickly, who has been hiding behind the artwork.

Why firms that do this see it pay off

The pattern we have watched, in the firms that have adopted these norms, is consistent enough that we are willing to put it on the record. In the first two quarters, the meetings feel awkward. Executives who have spent careers performing confidence are uncomfortable with the new register. Some of them, frankly, do not make the transition.

By quarter four, the executive bench has tightened. The leaders who could speak honestly about their uncertainty turn out to be, on average, the same leaders whose forecasts were tightest and whose teams were running with the most honest internal information. The board starts to see them clearly for the first time. The leaders who could only operate on confidence either learn the new register or look, by comparison, increasingly thin.

By year two, the decisions are different. Bets are sized to confidence levels. Kill criteria are in the board pack. The annual plan stops missing by the usual margins. The decision tax that the company had been paying — quietly, every quarter, in the gap between the conviction the room rewarded and the calibration the decisions actually deserved — starts to shrink.

It does not shrink to zero. No board meeting is a calibration exercise; politics, narrative, and conviction will always have their place in the room. But the proportion shifts. The room rewards a different signal. The signal it rewards is the one correlated with being right.

We help boards install these norms before a large failure forces the conversation. The cost of the change is some short-term discomfort in the meeting cadence. The cost of not changing is the steady, compounding tax of decisions made by the room's most confident speaker rather than its most accurate one. Over a five-year stretch, in our experience, those are not close.