Your best deal may have been a terrible decision

21 September 2026, 13:00

A poker player goes all-in with a hand that has a twenty per cent chance of winning. He wins. Did he play well?

No. He won, which is a different thing. If he repeated that play a hundred times he would lose eighty of them, and anyone at the table who reasons that way always ends up leaving their money behind, even if tonight they go home happy.

Professional players have a name for this mistake: judging the quality of a decision by its outcome. It is the most expensive trap there is, and it is not only about poker.

In football they have stopped looking at the score

In recent years football analysts have introduced a measure called expected goals: for every shot they estimate the probability that it becomes a goal, based on distance, angle, pressure and the part of the body used. At the end of the match they don't tell you who scored more, they tell you who created the better chances.

It is useful because a team can win one-nil with a deflected shot and convince itself it played well, then lose the next six matches doing exactly the same things. The score of a single episode is noise. The quality of the process is what repeats itself.

In business the opposite happens, and nobody notices

In companies, decisions are judged almost solely by their results.

That job order taken on below cost to get a foot in the door with a new customer: it went well, the customer then ordered three more times, so it was the right decision. But was it? Or would the customer have ordered anyway, and you gave away margin to someone who had already made up their mind?

That investment in a machine made without doing the numbers, which paid for itself anyway because the market went up: was it a good investment, or a stroke of luck that will be repeated next time, when the market goes down?

And the opposite, more painful case: the decision made properly, with the numbers in hand, that went badly because a supplier went under. In the company it will be remembered as a mistake, and whoever proposed it will keep quiet next time.

This is the real cost: the organisation learns the wrong lesson. It repeats the processes that were lucky and abandons the good ones that were unlucky. And people quickly work out that it pays not to stick their necks out.

Separating the decision from the outcome can be done, and it takes five minutes

The remedy used by players and analysts is simple in form and hard in discipline: write it down beforehand.

Before deciding, three lines on a sheet of paper. What I expect, expressed as a number. By when. And what I would have to see to say I was wrong.

Then put it away and reread it after six months, or twelve.

At that point you are no longer arguing from memory about who was right, because memory, funnily enough, always rewrites itself in favour of whoever is remembering. You are comparing a written expectation with something that actually happened, and the difference between the two tells you something valuable: whether the reasoning held up and it went badly by chance, or whether the reasoning had a hole in it, in which case you find that hole and close it.

You already do this, but on one piece only

Think about it: this is exactly how the budget works. The budget is not a forecast: it is a position taken in writing beforehand, with a number and a date. And variance analysis is not there to pin the blame on someone, it is there to understand whether the assumption was wrong or the world changed.

The point is that in most companies this discipline is applied only to what goes through the budget: revenue, costs, margins. All the other decisions — a new customer, a job order below cost, a new hire, a machine, a foreign market — are taken verbally and judged from memory.

What to do tomorrow morning

Take the next decision worth more than €20,000, whatever it is.

Before signing, write three lines: what you expect, with a number; by when; what would make you say you got it wrong. Put them in a folder, on paper or on the computer, and stop there.

Reread them in six months. It will be the best-spent half hour of the year, and not because you will find out you were wrong: because you will find out which of your lines of reasoning really work, and those can be repeated. Luck cannot.