Ask the room: “How many of your ten ranges contained the true answer?” Tap the buttons to count hands for each score. A well-calibrated room averages nine. Watch what it actually averages.
No hands counted yet.
Ten questions, none of which anyone knows exactly — that is the point. Each person gives a range they are 90% sure contains the truth. Ninety percent sure means nine of ten ranges should contain it. Across thousands of workshop runs of this classic exercise, most professionals score between three and six. The gap between nine and their number is a measurement of their own overconfidence — taken live, on themselves, with no way to argue.
Everyone opens this page on their own phone and answers alone. Stress the instruction: ranges so wide you'd bet money the truth is inside — nine times out of ten. Wide ranges are allowed. That's the trap: people can't bring themselves to use them.
Switch to the Group Tally tab on your screen. Count hands for each score, 0 to 10. The average appears live. Then say one sentence: “If this room were calibrated, that number would be nine.” Silence follows. Use it.
Why didn't we widen the ranges? Because narrow ranges feel competent and wide ones feel evasive — in front of colleagues, we perform confidence we don't have. Then bridge to work: budgets, timelines, sales forecasts.
Decision-making, risk, project estimation and forecasting programmes. Also devastating (kindly) with senior leadership teams — seniority does not improve calibration, it usually worsens it.