Number, arrow, line

Write the number 34 on a whiteboard and ask a room whether it is good. Nobody can answer. Add the word days and ask again. Still nobody can answer, although everyone now has a private guess. Then draw two more things beside it: an arrow pointing down, meaning lower is better, and a horizontal line across the board at 30. Ask a third time. Everyone answers at once, they all say the same thing, and nobody needed a discussion first.

A number, an arrow and a line. Only the first of the three is the figure your accounting software gives you. The arrow is the direction that counts as improvement, so nobody reads the number backwards. The line is the level at which you would do something about it, and it settles in advance the question of whether there is a problem. The most useful thing I know about measuring a business is that the two you have to write down yourself are the two doing the work.

Founders are rarely short of the first one. The accountant sends a profit and loss, the platform has its own dashboard, the bank app shows the balance. Every one of those figures is true. Between them they still do not answer the question you actually have, which is whether the business is alright. So you look at revenue, because revenue is the one you can always find. Revenue was up. It has been up for four months. And the quarter closed with less in the bank than it opened with, and you could name ten reasons for that without being able to say which one it was.

Missing the arrow is not a small omission. Debtor days rising reads like more sales unless somebody has written down that lower is better. Stock cover rising reads like being ready for the season rather than money standing still. In a business that holds inventory, a good half of the numbers worth watching are better when they fall, and one or two sit inside a band with a ceiling as well as a floor, where pushing further makes things worse rather than better.

Missing the line costs something different. With no threshold agreed in advance, every review opens with a negotiation about whether the number is a problem at all. Whoever sits closest to it explains why the month was unusual, the explanation is usually reasonable, and the meeting moves on. Do that four times and the drift is invisible, because it was never held against anything. A line written down before you needed it takes the argument away. The figure is over it or it is not, and the conversation can start where it should, at what to do.

The practice itself is small, and it has to happen before the month you need it rather than during it. Pick the handful of numbers you will look at, and keep the handful small, because a metric nobody acts on teaches everyone to stop reading the page. Agree with whoever produces each figure how it is worked out and where it comes from, and write both on the same row as the metric. Two people calculating the same number differently is how most dashboards stop being believed. Then set the direction, the target and the point at which it turns into a problem on every row, and put a name against each one. An unowned metric goes stale, and stale still looks like information.

One last thing, which is the part that stops people abandoning it in week three. Backfill three months from the accounts before your first review. Without that, the first page you produce is the least useful page you will ever look at, and that is not a fault in the method. One reading is a fact with no context. The second gives you a comparison. By the third there is a direction, and the direction was what you were after. Knowing it in month three rather than month nine is the return.