Cancel the meeting

Ten minutes before a stakeholder readout, the numbers aren’t right. The query is still running, or it finished and the figure looks wrong in a way you can’t explain yet, or you’ve found a join that double-counts and you don’t know how far the damage goes. The room is booked. People are walking over.

Cancel it. Send the analysis once it’s solid.

This feels like the weak option and it’s the strong one, for reasons that have nothing to do with looking diligent and everything to do with how the people in that room end up forming beliefs.

Anchoring is the whole argument

The first number a stakeholder hears becomes the number they reason from afterwards. Not deliberately, which is the problem. Anchoring is one of the most robust findings in decision research and it doesn’t switch off because everyone in the room is senior and busy. You say “looks like roughly 40% retention”, you flag that it’s preliminary, and what lodges is 40%. An hour later the caveat has evaporated and the figure is still there.

Now say the number was wrong, which you already suspected. Tomorrow the real figure is 28%, so you send the correction. The 40% doesn’t get cleanly replaced by the 28%. The 28% arrives as a retreat from 40%. Their mental model is anchored on the high number, so the true figure reads as bad news, a disappointment, a walk-back. It’s just the answer. You end up spending the correction on fighting an impression you created instead of on the finding itself.

The correction never quite catches up with the first impression. That’s the mechanism, and it’s why the calendar slot is a trap. The slot doesn’t know whether your analysis is ready. It only knows that it’s 2pm. Let the slot set the timing and an arbitrary half hour plants a number in five people’s heads that takes you a week to remove.

The hidden cost

The instinct that keeps the meeting in the diary is that cancelling wastes the slot and makes you look unprepared. The slot is the cheap part. Thirty minutes of calendar you can get back. A stakeholder anchored on a wrong number you can’t, at least not for free. You’re choosing whether to pay now in schedule or later in credibility.

There’s a second cost that shows up later. Next time you present a solid number, people remember the figure that moved overnight. They discount you slightly, hedge against your numbers, ask for workings they didn’t used to ask for. Being known for numbers that hold is worth more than being known for never missing a meeting, and moments like this are exactly where those two trade against each other.

The objection, and the answer

The obvious objection is that cancelling looks like you didn’t deliver. You had one job and you bailed on it.

Answer that in the cancellation itself. Write “Holding this until the numbers are verified. I’ve found something in the data I’m not willing to put in front of you until I trust it. Full analysis with you by tomorrow morning” rather than “sorry, not ready”. Nobody reads that as a failure to deliver. It reads as someone who won’t put a figure in front of decision-makers before it’s true, and that was always the job. Filling the slot never was.

You’ll feel the pull to present anyway, caveat heavily, and fix it later. The caveat is the part that doesn’t work. Caveats don’t undo anchoring, because the caveat gets processed as language and the number gets processed as fact, and the fact wins. Saying “this is rough” doesn’t stop the rough number becoming the anchor. The only thing that reliably prevents the anchor is not saying the number out loud.

So don’t say it. Cancel the meeting, verify the figure, and send the one you’ll still stand behind next week.