Before a strategy retreat or board strategy day most board chairs, directors, CEOs and CFOs want to know: how do you stress-test a strategic decision when the paper in front of you is thorough, the numbers hold up, and the people presenting it are the ones you trust most? The honest answer is that many reviews never actually test the decision. They test the discussion around it.

That difference matters more than it looks. Decision quality on a major call rarely comes down to how vigorous the debate felt in the room. It comes down to something far less visible: whether the evidence behind the recommendation was ever assembled to do anything other than confirm it.

I come at this from Execution Intelligence™: the gap between what organisations decide and what they actually do. That gap often opens right here, when a well-built case gets approved without anyone testing what it rests on.

Why Most Stress-Tests Examine the People Instead of the Evidence

Most of what passes for stress-testing in a boardroom is a check on the room, not on the recommendation. Was the debate robust? Did someone push back? Was dissent genuinely welcome? Those are fair questions. They’re also questions about people, not about the case itself.

Often the paper on the table is a product of what I call Recycled Certainty: confidence in a major decision that comes from familiar inputs (the same question, the same options, the same evidence and the same voices as the last big call) rather than tested ones. This piece is about the third of those, the same evidence: refreshed for this cycle, then assembled again towards the same yes. A confident, candid, well-run room can wave that paper through without anyone ever touching the evidence that built it. It happens to boards that do everything else right.

If you’re mainly worried about how to avoid groupthink in strategic planning, that’s a real risk with its own mechanics, and I’ve written separately about how to break groupthink in a leadership team. This piece is about something that survives even a perfectly candid room: evidence that was only ever gathered in one direction.

It’s worth being clear about why that’s so hard to see. Nobody sets out to build a one-sided case. A recommendation gets chosen early, often for good reasons, and from that moment every analyst, every model and every request for data is quietly working for it. The paper that results isn’t dishonest. It’s just finished before it was ever tested.

What Would Have to Be True for This Decision to Be Right?

Start here, before anyone argues for or against the recommendation. Ask the team to state, plainly, the two or three things that would have to be true for this call to be the right one. Not the conclusion. The load-bearing assumptions underneath it.

Be strict about “load-bearing”. A good paper will list plenty of assumptions, and most of them could shift without changing the answer. You’re looking for the few that would flip the recommendation if they turned out to be wrong. Usually that’s a belief about customers, a competitor or the organisation’s own capability, rather than a line in the financial model.

Then ask a second question, and don’t let it slide: is any of that new since the last time this call was considered, or is it the same belief the organisation has carried for years, restated with fresh numbers attached?

Daniel Kahneman, Dan Lovallo and Olivier Sibony made the case in Harvard Business Review, in “The Big Idea: Before You Make That Big Decision…” (June 2011), for treating a recommendation someone else has built as something to quality-control before you decide on it, checking it for the biases that may have shaped it rather than only checking the numbers. I agree, and I’d extend it with one question aimed squarely at the evidence: was any of it ever gathered with a real chance of saying no?

A team that can answer these quickly has probably done the work. A team that goes quiet has found, in real time, the gap between a recommendation and a decision.

What Would Show Us We’re Wrong, and When Would We See It?

This is the question that actually improves decision quality, rather than just prolonging the meeting. It doesn’t ask anyone to argue against the recommendation. It asks the team to specify, in advance, what disconfirming evidence would look like and roughly when it would show up.

Picture an executive team a week out from its strategy retreat. The recommendation is to bring an outsourced customer operations function back in-house. The paper is genuinely thorough: a cost comparison, a service-level analysis, a transition plan with dates against it. In review, one director asks a single question. What would we have to see, and by when, to know this was the wrong call?

Nobody in the room can answer. Not because the team is weak, and not because anyone was hiding something. Every piece of evidence in that paper was gathered to support the recommendation. Nobody had been asked to define what would disconfirm it. The team spends the following week building that answer, and the decision that eventually goes to the retreat is sharper for it.

So what does a good answer look like? It’s specific enough that someone could check it without arguing about it. It names a signal (a service measure moving the wrong way, say, or a cost line that doesn’t come down), a rough point by which it should be visible, and who will be watching for it. “We’ll monitor performance closely” isn’t an answer. “If this measure hasn’t moved by the second quarter after transition, it comes back to the board” is.

You’ll often catch this faster from an outside view than from inside the process that produced the paper, which is why it’s worth bringing in deliberately.

Were the Strategic Options Real, or Just Listed?

Almost every strategy paper includes alternatives. The evidence question is simple: were they tested with the same seriousness as the recommendation? Look at where the analysis went. If the recommended path has a full model, a risk assessment and a transition plan, and the alternatives have a paragraph each, the strategic options were listed, not developed. Nobody was ever asked to build the best possible case for anything else.

So ask two questions of every option on the page. What evidence would have to turn up for this option to win? And did anyone go looking for it? If the honest answer to the second is no, the option hasn’t lost the argument. It was never in it.

What Decision Quality Looks Like 48 Hours After You Commit

Here’s a test I use on every recommendation, however well argued: a decision you can’t describe at 48 hours can’t be tested.

Stress-testing needs two specifics that most papers never supply. First, the exact first visible move within 48 hours of committing, not a phase, not a quarter, but the actual action someone takes on the Monday after the yes. Second, the earliest piece of evidence that would tell you the call was wrong, and roughly when you’d expect to see it.

If the paper in front of the board can’t state either of those, it hasn’t been specified enough to stress-test yet, no matter how much analysis sits behind it. That gap is itself the finding. It tells a board more about decision quality than another round of debate ever will.

It sits alongside the Stop Test from my pillar piece, but where that test asks whether a decision is genuinely new, this one asks whether it’s specified well enough to be checked at all.

Why I Bring This Method Into the Room

I work as a speaker on decision-making for executive teams because the analysis is rarely the problem. The instinct to build a case towards yes is human. It happens to good teams and capable people, not careless ones. What changes the outcome is having someone in the room whose job is to ask the disconfirming question before the vote, not after the commitment.

In practice that’s a keynote before the retreat, a working session on the one recommendation heading for a decision, or both, so the room leaves with these questions answered rather than deferred.

If there’s a major recommendation heading towards your board or your next strategy retreat, that’s a conversation worth having before the paper is finalised, not after. Tell me about the decision and when it’s due to lock in.