IX Coach

Decision Journaling: Learning to Decide Better Over Time

How to build a personal feedback loop for the decisions that actually matter

What is decision journaling and why does it work?

A decision journal is a written record of significant decisions — what you decided, why, what you expected to happen, and what actually happened. It counteracts hindsight bias and outcome bias by preserving your actual reasoning before the outcome arrives, and it creates the feedback loop that produces real calibration: comparing your predictions to results across many decisions over time. The reason a journal is needed at all is that memory will not do this job — once you know how something turned out, recall of what you expected quietly reshapes itself to match, so reviewing a decision from memory mostly reviews a story written after the fact. A written entry made before the outcome is the one record that revision cannot reach. The practice itself is a practitioner method rather than a trialed intervention: the bias it targets is heavily replicated, but journaling as a treatment for it has not been tested in controlled studies.

Most people never find out if they’re good at deciding. A single decision rarely produces clean feedback: outcomes are noisy, attribution is murky, and hindsight rewrites what you thought you knew at the time. A decision journal addresses this by creating an external record — before the outcome — that you can score honestly later. Two distinct biases are in play, and the journal answers both. Hindsight bias corrupts the input: after you learn the result, your memory of what you expected drifts toward what happened, so the reasoning you review is not the reasoning you used. Outcome bias corrupts the verdict: a good decision that got unlucky reads as a mistake, and a reckless one that got lucky reads as brilliance. Annie Duke calls the second of these "resulting" — judging the quality of a decision only by whether it worked out. Keeping the record separates the two questions so you can ask "was that a sound call given what I knew?" independently of how it landed. Worth being straight about the evidence: the biases are among the most replicated findings in judgment research, while the journal itself is a practitioner technique — sensible, widely used, and not something controlled trials have tested. The practices below build the habit of logging, reviewing, and calibrating from your own decision history.

Practices

How these practices connect

Every one of these practices exists to separate a decision's reasoning from its outcome, because outcomes in uncertain domains are noisy and hindsight quietly rewrites memory of what you actually thought at the time. Logging before the outcome preserves the raw reasoning, scoring quality separately from outcome keeps a lucky or unlucky result from corrupting the lesson you draw, and the review, classification, calibration, and information-gap steps all determine what signal gets extracted once reasoning and outcome are safely kept apart.

Do I need to do all of these, or is logging decisions enough on its own?

Logging the decision before you know the outcome is the foundational move, because it is the only step that creates an immutable record hindsight bias cannot rewrite. But a log nobody reviews just accumulates unread entries, which is why a weekly decision review session exists to actually close the loop and extract a lesson. Without that review, you have preserved the reasoning but never turned it into feedback.

Why can't I just judge my decisions by whether they worked out?

Judging decisions by outcomes is what Annie Duke calls 'resulting,' and it's the most common way people learn the wrong lesson: a bad process can produce a good outcome through luck, and a good process can still produce a bad outcome through bad luck. Scoring decision quality separately from outcome quality asks a different question — was the reasoning sound given what you knew at the time — which protects the actual signal about your judgment from the noise of chance. The catch is that this only works if you're honest about it; giving high scores to decisions that happened to work is resulting wearing a different label.

Do I need to write down a probability for every decision?

Only if you want to know whether you're systematically overconfident or underconfident, which a single decision can never tell you — outcomes are too noisy to judge one at a time. Calibration is the match between your stated confidence and how often you're actually right, and it only becomes visible across dozens of predictions with assigned probabilities. Skip the probability and you're left with a narrative record of what happened, not a statistical one that can show you the pattern.

Should I spend the same amount of time deciding everything?

No — most people already apply roughly equal deliberation everywhere, which produces analysis paralysis on trivial choices and under-deliberation on the ones that actually matter. Classifying a decision by its stakes and reversibility before you decide tells you how much deliberation it deserves: Jeff Bezos's two-door framing treats reversible decisions as fast and correctable, and irreversible ones as worth real deliberation because they can't easily be undone. The honest part is hard — it's tempting to label a decision reversible specifically because you want an excuse to decide it quickly.

What's the difference between reviewing a decision and naming what information I was missing?

The weekly review is about extracting a lesson from a decision now that its outcome is known, scored on process rather than result. Naming the information you wished you had is a narrower, specific entry you write at the time of the decision itself, describing exactly what you didn't know. The two connect during review: you go back and ask whether that missing information would actually have changed the decision, which is a direct test of whether the gap mattered or was a false lead.

Is there real research behind decision journaling, or is this just a personal productivity habit?

The pieces have different kinds of support, and it's worth being specific rather than treating the whole practice as equally proven. Logging before the outcome and naming missing information are grounded in mechanism — how hindsight bias and information gaps work — rather than in trials of the journaling habit itself. Calibration tracking draws on real superforecasting research from geopolitical prediction tournaments, but that evidence hasn't been directly established for everyday personal and professional decisions, so the transfer is plausible rather than proven.

How do you choose between these?

Start with logging before the outcome and a weekly review — those two alone create the basic feedback loop and are worth doing before anything else. Add process-versus-outcome scoring as soon as you notice yourself praising decisions that happened to work out; that's the signal that resulting has crept in. Reach for the reversibility classification specifically at the moment of deciding something big and hard to undo, not as a habit applied to every choice, since its whole point is to concentrate deliberation where stakes are high. Calibration tracking and naming information gaps both need volume and patience — they're suited to someone logging many decisions over months, not a single high-stakes call, because the pattern they reveal only shows up across repetition.

What these approaches do not settle

The practices with mechanistic support — logging before the outcome, classifying by reversibility, and naming information gaps — are principled derivations from how bias and uncertainty work, not interventions that have been tested against real personal-decision outcomes. The ones with observational support — process scoring, weekly review cadence, and calibration tracking — lean on judgment research and forecasting-tournament data whose effect sizes in ordinary life decisions haven't been directly measured. None of these practices have been tested together as a combined system, and even the calibration research shows that overconfidence persists even in experienced practitioners, so no version of this promises the bias goes away — only that it becomes visible.

Practice this with IX Coach

Practice this with IX Coach

Frequently asked questions

What is decision journaling and why does it work?

A decision journal is a written record of significant decisions — what you decided, why, what you expected to happen, and what actually happened. It counteracts hindsight bias and outcome bias by preserving your actual reasoning before the outcome arrives, and it creates the feedback loop that produces real calibration: comparing your predictions to results across many decisions over time. The reason a journal is needed at all is that memory will not do this job — once you know how something turned out, recall of what you expected quietly reshapes itself to match, so reviewing a decision from memory mostly reviews a story written after the fact. A written entry made before the outcome is the one record that revision cannot reach. The practice itself is a practitioner method rather than a trialed intervention: the bias it targets is heavily replicated, but journaling as a treatment for it has not been tested in controlled studies.

How do you log the decision before the outcome?

Hindsight bias overwrites the memory of your pre-decision reasoning with knowledge of what actually happened, making bad decisions look smart in retrospect (if lucky) and good decisions look naive (if unlucky). Writing the reasoning down before the outcome creates an immutable external record that the bias cannot reconstruct. This is the foundational move of decision journaling: without it, you have no ground truth to review.

How do you score decision quality separately from outcome quality?

"Resulting" — Annie Duke’s term for judging decisions by outcomes — is the most common way people learn the wrong lesson. In uncertain domains, a bad process can produce a good outcome by luck, and a good process can produce a bad outcome by bad luck. Scoring the process separately ("was the reasoning sound given what I knew?") breaks this conflation and protects the actual signal from noise.

How do you run a weekly decision review session?

Decisions and their outcomes are separated in time, which means feedback is only useful if you create a deliberate moment to receive it. Without a regular review ritual, entries in a decision journal never get closed, outcomes never get compared to predictions, and the feedback loop — the only reason to keep the journal — never closes. A weekly cadence is frequent enough to keep the reasoning fresh and infrequent enough not to feel onerous.

How do you classify decisions by reversibility and stakes?

Most people apply roughly the same amount of deliberation to all decisions, which means they over-deliberate on trivial ones (analysis paralysis) and under-deliberate on high-stakes irreversible ones (defaulting to intuition). Jeff Bezos’s "two-door" framing captures the principle: reversible decisions (door one) should be made quickly and corrected if needed; irreversible ones (door two) deserve much more deliberation because they can’t easily be undone.

How do you track your calibration across many predictions?

A single decision tells you almost nothing about your decision-making quality — outcomes are too noisy. Calibration — the match between stated confidence and actual accuracy — only becomes visible across many predictions. If you say "80% confident" across 20 decisions and are right only 50% of the time, you are systematically overconfident at that level, and the journal gives you the data to see it.

How do you log the information you wished you had?

Decisions made under uncertainty are limited by information gaps. Most people experience these gaps vaguely ("I just wasn’t sure") without articulating what specifically was missing. Naming the exact information gap does two things: it clarifies how the gap could have been filled before deciding, and it reveals, after the outcome, whether filling it would have changed the decision — which is a direct test of whether the gap actually mattered.

Do I need to do all of these, or is logging decisions enough on its own?

Logging the decision before you know the outcome is the foundational move, because it is the only step that creates an immutable record hindsight bias cannot rewrite. But a log nobody reviews just accumulates unread entries, which is why a weekly decision review session exists to actually close the loop and extract a lesson. Without that review, you have preserved the reasoning but never turned it into feedback.

Why can't I just judge my decisions by whether they worked out?

Judging decisions by outcomes is what Annie Duke calls 'resulting,' and it's the most common way people learn the wrong lesson: a bad process can produce a good outcome through luck, and a good process can still produce a bad outcome through bad luck. Scoring decision quality separately from outcome quality asks a different question — was the reasoning sound given what you knew at the time — which protects the actual signal about your judgment from the noise of chance. The catch is that this only works if you're honest about it; giving high scores to decisions that happened to work is resulting wearing a different label.

Do I need to write down a probability for every decision?

Only if you want to know whether you're systematically overconfident or underconfident, which a single decision can never tell you — outcomes are too noisy to judge one at a time. Calibration is the match between your stated confidence and how often you're actually right, and it only becomes visible across dozens of predictions with assigned probabilities. Skip the probability and you're left with a narrative record of what happened, not a statistical one that can show you the pattern.

Should I spend the same amount of time deciding everything?

No — most people already apply roughly equal deliberation everywhere, which produces analysis paralysis on trivial choices and under-deliberation on the ones that actually matter. Classifying a decision by its stakes and reversibility before you decide tells you how much deliberation it deserves: Jeff Bezos's two-door framing treats reversible decisions as fast and correctable, and irreversible ones as worth real deliberation because they can't easily be undone. The honest part is hard — it's tempting to label a decision reversible specifically because you want an excuse to decide it quickly.

What's the difference between reviewing a decision and naming what information I was missing?

The weekly review is about extracting a lesson from a decision now that its outcome is known, scored on process rather than result. Naming the information you wished you had is a narrower, specific entry you write at the time of the decision itself, describing exactly what you didn't know. The two connect during review: you go back and ask whether that missing information would actually have changed the decision, which is a direct test of whether the gap mattered or was a false lead.

Is there real research behind decision journaling, or is this just a personal productivity habit?

The pieces have different kinds of support, and it's worth being specific rather than treating the whole practice as equally proven. Logging before the outcome and naming missing information are grounded in mechanism — how hindsight bias and information gaps work — rather than in trials of the journaling habit itself. Calibration tracking draws on real superforecasting research from geopolitical prediction tournaments, but that evidence hasn't been directly established for everyday personal and professional decisions, so the transfer is plausible rather than proven.

Research

Related concepts

Articles that cover this