A date is a point estimate pretending to be certainty. Test your 90% confidence, then move a cone across a project timeline and watch commitments age.
A roadmap that promises a feature "in Q3" reads as a commitment but hides three months of unknowns. Publish the confidence — "likely Q3, maybe Q4" — and stakeholders plan around the range instead of stacking launches on one optimistic pixel.
That single number quietly folds sourcing, interviews, offers, and notice periods into one point. The honest forecast is a range — say 6 to 14 weeks — that widens for senior or scarce roles. Backfill against the point and you inherit the gap nobody planned for.
"$1.2M" claims a precision the year doesn't have. A range — $1.0M to $1.6M at 90% confidence — tells the board what could actually happen and where the risk lives. A point gets treated as a ceiling, then blown; a range sets an expectation reality can fit inside.
A deal marked "closing March 31" is a guess wearing a fact's clothes — and a pipeline stacked from those points forecasts revenue that slips every quarter. Commit to a close window weighted by stage: harder to celebrate, far harder to miss four times running.
If your 90% ranges hit five out of ten, you are not unlucky. You are too narrow. Widen until reality fits the stated confidence.
Discovery, dependencies, and unknown unknowns dominate early. The cone should narrow as evidence arrives, not because a meeting needs a date.
At the wide end of the cone, commit to the next decision point. Commit to a delivery date only when the range has earned it.
The one move that survives every planning meeting: when someone asks for a date, hand back a range — then refuse to narrow it until evidence, not the calendar, has earned the edges.