Knowledge base › Direction & values

A leader makes hundreds of decisions a year and could, in principle, learn from every one. In practice they learn from almost none — not through carelessness, but because memory lies about our own decisions, reliably and predictably. When you look back at a call six months later, you no longer see what you saw then: you see the outcome, and you build the story backwards to fit it. A decision journal fixes exactly that failure. This article covers why memory doesn't work here, what to record, when to review, and which tool to use.

Why don't you learn from your own decisions?

Two well-documented biases work against you, and both switch on after the fact:

1. Outcome bias. We judge a decision by its result. It worked out, so it was a good call; it didn't, so it was a bad one. But those aren't the same thing: a carefully reasoned decision can end badly (the world is uncertain) and a reckless one can end well (you got lucky). Learn only from outcomes and eventually you'll learn luck as a method.
2. Hindsight bias. Once we know how it ended, we remember that "you could see it coming". Memory retroactively rewrites how uncertain we were. That makes every past decision look more logical than it was — and leaves nothing to learn from.

At leadership level the consequence is expensive: you can repeat the same mistake for years while feeling, the whole time, that you're learning from it. The fix isn't more reflection — it's a written trace from the moment of the decision, something memory can't rewrite afterwards.

What to record — the six fields

A good entry takes two or three minutes. This is what it needs:

1. The decision, in one sentence. Concretely. Not "we're getting into marketing" but "from October we put a fixed monthly budget into search ads with the current agency".
2. What information it rests on. What you know, what you're assuming, and what you simply don't know. The last one matters most — and is the most uncomfortable to write down.
3. What you expect. Measurably. "More leads" isn't enough; "25 qualified leads a month within three months" is.
4. How confident you are. As a percentage. It feels odd at first, but over time this is the most instructive field: you find out whether your 80% calls really land seven times out of eight, or closer to five.
5. What would disprove it. Which concrete signal would show you were wrong — and by when it should appear. This field is what stops you defending a bad direction for months.
6. When you'll review it. A specific date. Without it the review doesn't happen.

Not every decision belongs here. Only those that are hard to reverse or have consequences running for months: hiring, firing, pricing, larger spend, a change of direction, a partnership. For most leaders that's one or two a week.

The review — where the learning happens

Recording alone is worth nothing. The value is in the review, and only if you ask the right question.

The wrong question: "was the outcome good?" That often comes down to luck, and teaches nothing.

The right question: "was the reasoning good with what I knew then?" That yields four distinct cases, and each teaches something different:

1. Good decision, good outcome. Repeatable — worth looking at what made the process good.
2. Good decision, bad outcome. Bad luck. The hardest case, because organisations tend to punish it, when this is exactly what should be protected.
3. Bad decision, good outcome. The most dangerous case. Here you learn a bad habit and it feels like it works.
4. Bad decision, bad outcome. The most tangible lessons live here — provided you have in writing what you thought at the time.

You can only fill in those four boxes if the reasoning was recorded at the moment of decision. Without it, every case becomes "bad decision, bad outcome" or "good decision, good outcome" — because that's what memory finds most comfortable.

Which tool? — the field, honestly

1. Paper, a notebook. The simplest, and for many the best, because writing slows you down. Where it stops: it doesn't remind you to review, and six months later you can't search for similar decisions.
2. A spreadsheet (Excel, Sheets). Searchable, filterable, and you can actually evaluate those confidence percentages. Where it stops: nobody opens a spreadsheet because a decision came to mind — recording lapses fast.
3. A Notion template. There are good free decision journal templates, and Notion can handle the review date. Where it stops: the system needs maintaining, and your decisions still can't see the rest of your data.
4. A dedicated decision journal app. Surprisingly few exist, and most are experiments rather than finished products. The category is essentially empty.
5. Mirrify. Here the decision journal isn't a separate tool but part of the system: in the same account as your goals, your daily priorities, your reflections and your habits. That's what makes it more than an archive — the AI mentor reads your decisions alongside the rest of your data.

What the AI adds — concretely

A static note preserves the reasoning. A connected system goes further, because it can show a pattern:

1. Calibration. Record confidence percentages for months and it becomes visible how well-tuned you are. For many leaders the 90% calls land around 60% — and that's only noticeable from a written record.
2. Situational pattern. It may turn out your rushed decisions reliably follow the same kind of week — because your journal and mood live in the same system.
3. Business focus. Mirrify has a separate business and career coach mode aimed specifically at your work: it works from your goals, your daily three priorities and your reflections, quotes your own notes verbatim, and gives a concrete step you can take this week — not "be more confident" advice.
4. Reminding. The review date doesn't get lost between the pages of a notebook.

An honest limit: this is not business consulting and it doesn't replace your expert partners. What it gives you is visibility into your own decision patterns — the judgement is still yours.

How to start this week

1. Pick your next hard-to-reverse decision. Don't start retroactively — that record is already distorted.
2. Fill the six fields in two or three minutes. Don't flatter the confidence percentage.
3. Put the review date where you'll actually see it. In your calendar or in the system, not in your head.
4. After four or five entries, look back at the first. This is where the first real surprise arrives — usually that you were far less certain than you remembered.

Frequently asked questions

What is a decision journal?

A short note written when you make the decision: what it is, what information you have, what you expect, how confident you are, what would disprove it, and when you'll review. The timing is the point — at the moment of deciding your reasoning is still known.

Why don't leaders learn from their decisions?

Outcome bias makes us judge by the result; hindsight bias makes us remember we "knew it all along". Together they erase the lesson — unless there's a written trace from the moment of decision.

What should I record?

Six things: the decision in one sentence, the information, the measurable expectation, the confidence percentage, the disproving signal and the review date. The fifth is the most valuable and most often skipped.

When should I review?

Set the date at recording — one, three or six months. At review the question isn't whether the outcome was good but whether the reasoning was, given what you knew then.

Is there a dedicated decision journal app?

Barely — most people use paper, a spreadsheet or a Notion template. Those don't remind you to review and can't see your decisions alongside your other data. In Mirrify the decision journal is part of the system.

What does an AI mentor add?

The connection: calibration, situational patterns and reminders. The business and career coach mode works from your goals, daily priorities and reflections, and gives a concrete step for this week.

How much time does it take?

Two or three minutes to record, five to review. Only decisions that are hard to reverse — for most leaders one or two a week.