Decisions.

The framework

1. 01

   Write the entry before the result exists.

   The journal is worth keeping only because it is written in ignorance. Once you know how something turned out, your memory of what you expected quietly slides towards what happened: you recall doubts you never voiced, or a confidence you never held, and both feel like honest recollection. That is hindsight bias, and it is not a character flaw you can correct by trying harder to be fair — it is how memory works. So capture the entry on the day you commit, in ten minutes: the choice, the options you rejected and why, the two or three facts carrying the most weight, and the gap you are choosing to live with.

2. 02

   Make the expectation scoreable: a number, an observable, a date.

   Most decision notes collapse the moment you try to grade them, because they record intent rather than expectation. 'This will strengthen our position' cannot be wrong. A year later it still reads as true whatever happened, so it grades itself and you finish the review knowing nothing you did not know going in. Write instead what you expect to become true in a form someone else could check: the metric, the level, the date by which it should have moved, and the observation that would prove you wrong. Add your confidence as a percentage. It feels absurd the first few times, and the awkwardness is the point — a number you can be wrong about is the only kind you can learn from. Then put the review date in the calendar with the entry attached, or you are keeping a diary.

3. 03

   Score the decision and the outcome separately — four boxes, not one.

   At review, grade two things independently: was the reasoning sound given what you knew and could reasonably have found out, and did it work. Deserved wins are the least instructive box. The teaching happens on the diagonals. A sound decision with a bad outcome is variance, and the discipline is to hold the process rather than rewrite your whole approach in front of one result. A poor decision with a good outcome is the genuinely dangerous box, because it pays you for sloppiness, gets retold at dinner as a triumph, and quietly hardens into how you do things. Name that box in the entry when you land in it, and name it again to whoever has been retelling it as a triumph, or it bills you later at a worse moment.

4. 04

   One entry is an anecdote. Thirty are a diagnosis.

   The journal compounds only when you read it in a stack. Once a quarter, sit with every entry since the last read and hunt for the repeating flaw rather than the individual verdict: timelines you habitually halve, hiring calls you rush when a role is painful to cover, commitments made in the hour after a difficult meeting that you would not have made that morning. Check your calibration while you are there — of the calls you marked eighty per cent confident, how many landed. The trap is the private journal, kept in a personal file, read by nobody, changing no rule. A pattern you have named should leave the read as a written rule your team can hold you to.