Write Down Why
Six months after a decision, nobody remembers why you made it. They remember how it turned out. If it worked, the reasoning gets remembered as wise, even if it was a coin flip. If it failed, the reasoning gets remembered as obviously foolish, even if it was the best call anyone could have made with what was known at the time. Both memories are usually wrong, and a company that runs on them learns the wrong lessons with great confidence.
I came to this slowly, across a long run of companies in payments, crypto and now AI. It took me longer than I would like to admit to see the pattern, because the pattern hides inside outcomes.
Two ways to learn the wrong thing
The first is the good decision with bad luck. You weigh it carefully, you make a sound call, and something you could not have foreseen breaks it. The market turns, a partner changes strategy, a regulator reinterprets a rule. The outcome is bad. And because nobody wrote down what the reasoning was, the reasoning gets thrown out with the outcome. The team learns never to do that again, when the honest lesson was that they should do exactly that again and buy better insurance.
The second is worse: the bad decision with good luck. You cut a corner, skip a check, bet on something you did not understand, and it works. The outcome is good, so the reasoning gets promoted to doctrine. People repeat it. It becomes how things are done. And it keeps working until the day the luck runs out, usually at a much larger scale than the first time, when there is much more to lose.
In both cases the problem is the same. The only record of the decision is the outcome, and outcomes are a noisy measure of judgment. You cannot learn to decide better from a record that contains nothing about how you decided.
The fix is ten minutes
Every company I run now keeps a numbered decision log. It is not sophisticated. Each entry is short, and it is written at the time, by the people deciding, before anyone knows how it will turn out. It answers five questions.
What did we decide, and when? One sentence, with a date. Not "a while back." The actual day.
What did we know, and what did we not know? This is the part that protects you later. It is the difference between a decision that was wrong and a decision that was right on the information available.
What did we reject, and why? The options you turned down are half the decision. When someone proposes one of them again a year from now, you will want to know whether it was considered and what killed it.
What would change our mind? This is the most valuable line, and the one most often skipped. A decision with no stated reversal condition becomes permanent by default. A decision that says "if churn passes this level, or this partner leaves, we revisit" has a built-in alarm.
Who decided? Not to assign blame. To know whom to ask.
The entries are never edited afterward. If we change course, that is a new entry, pointing back to the old one and saying what changed. Over time the log becomes something more useful than a history. It becomes a record of how the company thinks.
What it buys
The first thing it buys is honesty about judgment. You can look back at a decision and grade it on what was knowable, not on what happened. That sounds abstract until you have watched a team punish a good call and reward a lucky one, and watched the best people quietly learn that being right does not matter as much as being fortunate.
The second is fewer reopened arguments. Every company has questions that get relitigated every quarter because nobody can remember how they were settled. With a log, the answer and its reasons are one search away. If the reasons still hold, the conversation is short. If they do not, you have a specific thing to discuss instead of a vague feeling.
The third is onboarding. A new person joining a company inherits its results but not its reasoning, and so they either defer to decisions they do not understand or challenge decisions that were already thought through. A decision log lets them read the reasoning directly. It is the closest thing a company has to letting a newcomer sit in the old meetings.
The fourth is that it changes the decisions themselves. Knowing you will write down what would change your mind forces you to think about it. Knowing you will list what you did not know forces you to notice it. The act of writing the why is half of thinking it.
There is a fifth, quieter benefit, and it may be the one I value most. A written why is kinder to the people who made the call. When a decision goes badly, the log shows that the team saw the risk, named it, and chose anyway for stated reasons. That turns a postmortem from a search for someone to blame into a conversation about what we would weigh differently next time. People who are not afraid of being judged by hindsight make braver and more honest decisions.
Keep it boring
The temptation is to make this elaborate: templates, approvals, a tool with workflows. Resist it. The best decision log I have seen is a folder of short numbered files in plain text, one decision per file, with a date at the top. It costs almost nothing to keep and almost nothing to read. The moment it becomes a process, people stop writing in it, and a decision log nobody writes in is worse than none, because it looks like a record and is not.
Write down why. Ten minutes, at the time, never edited after. It is the cheapest discipline I know in a company, and the one I most wish I had started earlier.