Outcomes rewrite memory. A decision log records what you knew and expected when you chose, so a later review can judge the reasoning instead of the luck.
· Note-Taking Methods · 8 min read
You made a call six months ago. Hire the contractor or wait for a full-time hire; ship the feature behind a flag or go all in; take the job or stay. Today you know how it turned out, and you are fairly sure you knew it would. You remember your reasons clearly, and — this is the part that should worry you — they are the reasons that turned out to be right.
Almost none of that memory is reliable. The moment an outcome arrives, it rewrites the story of the decision that led to it. Psychologists call this hindsight bias, and it is not a flaw you can think your way out of; knowing about it barely reduces it. The only defense that works is documentary: a record of what you actually believed at the moment of choosing, written before you could possibly know how things would go. That record is a decision log, and it is one of the highest-return notes you can keep, because it is the only one that lets you learn from your own judgment instead of from your luck.
A decision log is a dated entry, written at the time of a non-trivial decision, that captures three things: what you decided, why, and what you expected to happen. That is the whole method. The term "decision journal" was popularized by Shane Parrish at Farnam Street, drawing on Daniel Kahneman's advice to keep one, but the practice is older than the name — investors have kept versions of it for decades, for the plain reason that markets deliver verdicts and memory does not.
It is not a diary of everything you did. It is not a pros-and-cons list to help you decide — the log starts once you have decided. And it is not a place to record outcomes as they happen. The outcome goes in later, deliberately, in a separate pass, and keeping the two apart is the entire point.
The value of an entry comes almost entirely from when it was written. A note about a decision written after the outcome is known is worth roughly nothing, however honest you try to be, because the outcome has already edited your memory of the inputs. A note written the day you decided is a fossil: it preserves your actual state of knowledge, including the things you did not know, the risks you dismissed, and the confidence you felt.
That fossil is what makes review possible. When you come back to it, you can ask the only question that improves judgment: given what I knew, was this a good decision? — as distinct from did it work out? Good decisions go badly and bad decisions go well all the time. Without the record, you can only grade outcomes, and grading outcomes teaches you to be lucky, which is not a skill.
Keep it short enough that you will actually write it. A workable entry has five parts, each one line or a few bullets:
Then add one more line: a review date. Pick when you will realistically know something — three months for a hire, two weeks for a launch, a year for a career move. Write it down as a date, not as "later."
Five minutes. If it takes twenty, you are re-deciding, not logging.
Not the daily ones. Logging what to have for lunch produces noise and kills the habit by Thursday. The threshold that works: would I want to know, a year from now, what I was thinking? Hiring and firing. Anything involving money you would notice losing. Committing a team to a direction. Turning something down. Choosing between two credible plans where reasonable people disagreed.
A useful side effect of that threshold is that it forces you to notice when you are making a decision at all. A lot of consequential choices happen by drift — nobody decided, it just went that way — and the act of asking "is this log-worthy?" catches some of them while they can still be made on purpose.
Expect five to fifteen entries a month for most people. Fewer is fine. A log with six honest entries beats one with sixty perfunctory ones.
Writing entries is the cheap half. The return comes when the review date arrives and you reread what you wrote — ideally before you have consciously updated your memory of it.
Do it in two passes. First, record the outcome under the entry, factually: what happened, when, what surprised you. Second, grade the reasoning, not the result. Was the situation as you described it? Which expectation was right, which was wrong, and by how much? Were the alternatives you rejected actually worse, or did you dismiss one for a reason that now looks thin? Did the "how I feel" line predict anything?
After ten or so of these, patterns appear that no single review would show. You are overconfident about timelines and well-calibrated about people, or the reverse. Decisions made when the feeling line says "rushed" go badly at twice the rate. You consistently underrate the option you were biased against. These are the findings that change how you decide next time, and they are invisible without the record — memory smooths them all into "I usually have good instincts."
Fold the reviews into whatever weekly review you already run: a glance at the log for entries whose date has come up, then the two passes. It rarely takes more than a few minutes a week, because most weeks nothing is due.
One note, chronological, newest at the top or the bottom — pick one and stop thinking about it. Each decision is a headline with its five parts nested underneath, and the outcome and review nested under those when they arrive. An outliner suits this well: fold everything and the note reads as a list of the decisions you have made; open one and the reasoning is there; open the review and the grade is there.
Two mechanics matter. The first is that the review date has to come back to you without your remembering it. In Indenta, the review date on an entry is a !! date chip, and every line that mentions a date appears in that date's Day View alongside the day's journal and tasks, so the entry resurfaces on the morning it is due without a separate reminder system. The User Guide covers the details. In any other tool, a due date on a task line or a calendar reminder pointing at the note does the same job.
The second is that decisions rarely arrive in the log note. They happen in a meeting note, a daily note, a thread you are summarizing. The habit that survives is writing the entry where you are, then linking it — a wiki-link from the log to the meeting note, or the reverse — rather than opening a separate note and reconstructing the context. The log is an index of your decisions; it does not need to be the place they were made.
Logging after the fact. An entry written a week later, once the early signals are in, is contaminated. If you missed the moment, write the entry anyway, but date it honestly and note what you already know. It is still worth something as a record; just do not grade yourself on it.
Expectations that cannot be wrong. "I expect this to go well" is not a forecast. If a friend could not check it in six months and say yes or no, rewrite it.
Skipping the feeling. People skip it because it seems soft. It is the highest-signal line in the entry.
Never reviewing. A log without reviews is a diary with extra steps. If review dates keep passing unnoticed, the fix is mechanical — make the date surface itself — not motivational.
Grading the outcome. The hardest one. A good decision that went badly should be marked good decision, bad outcome and left alone. The urge to rewrite it as a mistake is hindsight bias doing exactly what the log exists to stop.
There is a decision somewhere in your next two weeks that clears the threshold. When you make it, before you tell anyone, open a note and write the five parts and a review date. It will feel slightly ceremonial and slightly pointless, which is how every entry feels on the day it is written. The return is deferred by design: nothing in the log is worth much until the review date arrives and you read a paragraph by someone who genuinely did not know how it would turn out — and discover, precisely, what they got right.
Try it in practice: Indenta is a free, offline-first outliner — nested notes, backlinks, tags, and peer-to-peer sync, with no account required. Start writing in your browser, or read the User Guide first.