Round numbers get approved

I presented monthly reviews to executives for a couple of years, through a run of acquisitions, so the pack changed shape often and I got to watch the same audience react to a lot of different numbers.

The pattern showed up early enough that I started predicting it before the meeting. The figures that had been derived, traced, argued over and checked drew questions. The figures that were round drew none. It was not correlated with size, or with how much money was at stake, or with whose area it sat in. It tracked the digits.

What roundness signals

A round number reads as settled. It looks like somebody already did the work, formed a view and compressed it, and what you are seeing is the conclusion. It has the texture of a decision.

A precise number reads as in progress. All those digits look like the residue of a calculation still running. There is more surface on it, so there is more to ask about, and every extra digit is another place a reader can put a finger and say, where does that come from.

So the two signals invert the actual effort. The number I could defend for twenty minutes attracted scrutiny because it looked unfinished, and the number I had put in as a stand in sailed through because it looked resolved. Readers are not lazy here. They are reading confidence off the shape of the thing, which works fine in most of life and is exactly wrong in a management pack.

Then I stopped calling it laziness

My first conclusion was that this was a failure of attention in the room, and I held that view longer than I should have.

It is not, mostly. A round number is a legitimate signal in its own right. It often does mean approximate, and a reader who declines to interrogate the second decimal of an approximation is behaving sensibly. And precise numbers genuinely do deserve more questions, because there is more of them to be wrong about. Both halves of the reaction are defensible.

The problem is that nothing in the document distinguishes round because approximate from round because tidy. Those are opposite states and they have the same appearance. One means the producer is telling you the range is wide. The other means the producer had an exact figure and smoothed it for the slide, which discards information without saying so.

Nobody has ever introduced a figure to me as approximate out loud in that setting. It would be the most useful sentence available and I have not heard it said, and I have not said it either.

The incentive that follows

Here is the part I dislike, because I did it.

Once you know that round passes and precise gets questioned, you learn to round. Not to deceive. To reduce the number of questions in a meeting you have thirty minutes for and four other items in. The derived figure and the guess get presented in the same register, and the fact that one of them cost three days disappears from the record entirely.

Which means the pack stops carrying any information about how much is known, and the only place that information survives is in the head of whoever built it. That person leaves eventually.

I have no fix for it. I have a habit, which is that when I round something I try to leave one sentence in the file saying what the unrounded figure was and why I did it. It is for whoever opens it in eighteen months, possibly me, and it takes about as long to write as this paragraph did.