Tool/Mental Model/No. 0239

Decision Journal

A decision journal is a record of a choice, its reasons, assumptions and expected results made before the outcome is known. Later review compares reasoning with results, helping distinguish sound judgment from luck and limit hindsight bias and outcome bias.

a tool: pick it up

Evidence: Useful, modest evidence
Read: 6 min
Links: 19 connections
Useful when: Deciding under uncertainty · Forecasting · Learning and memory · Money and investing · Running projects

01You've seen this when…

in life
Before accepting a job with a longer commute, you record what you expect to gain and lose. Three months later, the note reminds you that working from home twice a week was an assumption you had made without a promise from your employer.
at work
Your team chooses a supplier and saves its forecast alongside the contract. When deliveries slip, the review starts with the risks you actually considered, not the risks everyone now remembers considering.
out in the world
A school board approves a tutoring pilot and publishes its expectations before it starts. At the review, members compare the results with those expectations, keeping the original definition of success visible and unchanged.

02The idea

After a choice works out, your reasons tend to look stronger. After it fails, the warning signs seem more obvious. Memory edits the decision to fit the ending.

A decision journal preserves the version that existed before you knew the result. It records the choice, the alternatives, your reasons, the assumptions those reasons depend on, and what you expect to happen. Later, you compare that record with events.

This helps separate two questions: was the decision reasonable with the information available, and did it turn out well? A careful choice can fail. A reckless choice can succeed. Treating the result as a verdict on the reasoning is outcome bias. Remembering the result as something you anticipated is hindsight bias.

A decision journal keeps your reasoning available for inspection. You can faithfully record terrible reasoning, so the quality of the decision still needs to be assessed. Across several decisions, you can start finding recurring mistakes: optimistic estimates, untested assumptions, or alternatives you dismiss too quickly.

03How to use it

Write before you commit. Use a dated note, form, or email to yourself. Preserve the original and save later updates separately. A useful entry usually takes ten to fifteen minutes.
State the choice and its purpose. Name what you’re deciding, what you want to achieve, and the available alternatives, including waiting or doing nothing. Otherwise, a later review may judge the choice against a different goal.
Separate facts from assumptions. Record the evidence you have and what you’re taking on trust. Include important missing information. Check your expectations against similar past cases using the outside view.
Make expectations checkable. Say what outcome you expect by a specific deadline and how uncertain you are. For a yes/no event, give a probability if you can do so meaningfully. For a cost or timeline, give a plausible range. Avoid fake precision and vague predictions that almost any result could satisfy.
Record what could change your mind. Identify the strongest reason against your choice, the most fragile assumption, and any condition that would make you stop or switch. Note pressure, fatigue, or excitement if they seem relevant.
Schedule the review now. Choose a date when useful evidence should exist. At that point, assess the reasoning and the outcome separately. What information was available? What was overlooked? Which surprises were hard to anticipate? Finish with one change to your future process.

For repeated yes/no forecasts, a Brier score can track forecast accuracy. But one failed prediction doesn’t establish poor calibration: events assigned an 80% chance should sometimes fail.

04A worked example

Consider a fictional retailer choosing between its established delivery supplier and a cheaper newcomer. The newcomer would save $4,000 a month. A trial went well, so the operations manager recommends switching.

Before signing, she records an 80% chance that at least 95% of orders will arrive on time over the next three months. She identifies handling a demand surge as the main uncertainty. She also records that the trial covered 200 orders during a quiet period and sets a review date.

Three months later, only 91% of orders have arrived on time. Most delays followed a surge in demand.

What it looks like A bad result that makes the switch seem obviously foolish. Colleagues remember having doubts, while the manager remembers being less confident than she was.

What’s actually going on The original note prevents both revisions. It also reveals a specific weakness: the manager recognized peak capacity as a risk and accepted a trial that left it untested. The review can investigate that untested assumption to assess the choice’s reasonableness, which the failure alone leaves unresolved.

What made it work The dated forecast, the recorded limits of the trial, and the scheduled review. The team adopts a new rule for similar switches: test surge capacity or retain a backup supplier. The lesson changes a procedure instead of becoming a general warning to be more careful.

05When to reach for it

The choice will teach you something reusable. Hiring, supplier selection, investing, and project estimates offer repeated opportunities to improve.
Feedback will arrive slowly. Without a record, you’ll forget which assumptions belonged to the original decision.
Reasonable people disagree. Record their different expectations, not just the final consensus.
The stakes invite self-defense. Money, reputation, and public commitments make an unchanged record especially valuable.

Start with a few consequential decisions. Logging every small choice turns a learning tool into paperwork.

06When it misleads

It becomes a defense brief. A polished justification can hide uncertainty as effectively as no record at all. Include evidence against your choice and alternatives you nearly selected.
You treat one result as a complete test. An unlikely event can happen. Look for patterns across comparable decisions before diagnosing a forecasting weakness.
You invent the missing alternative’s outcome. You observe what happened after your choice. What would have happened under another choice remains unknown. A successful switch doesn’t prove that staying would have been worse.
You punish honest forecasts. If a workplace uses journals to prosecute every miss, people will write safe, empty predictions. Reviews should reward clear reasoning and useful corrections while giving people room to express uncertainty.
You collect notes without changing anything. Learning from a journal requires using it to make changes. End reviews with an adjustment you can actually use, such as gathering a missing piece of evidence or changing a test.

New information should change your decision when appropriate. Bayesian updating means revising your beliefs. Keep an accurate account of what you believed earlier.

07Roots

Peter Drucker’s advice to managers was unusually low-tech: write down what you expect when making a key decision, then return to the note nine to twelve months later. In his 1999 essay Managing Oneself, he called this feedback analysis. The comparison was meant to reveal strengths and weaknesses that intuition alone could miss. The revealing detail was the wait: the note had to survive long enough for expectations to meet reality.

Psychology supplied a reason not to trust the retrospective version. Baruch Fischhoff’s 1975 experiments showed that learning an outcome changes people’s judgments of how foreseeable it was. Jonathan Baron and John Hershey later showed that outcomes affect evaluations of decisions even when the information available to the decision-maker is held constant. Together, these findings explain why an untouched record can be more useful than a confident recollection.

Credit for the modern decision journal’s invention remains unsettled. It combines feedback practices from management with those used in investing and forecasting. Writers such as Annie Duke have helped popularize the distinction between a good decision and a good result. Today’s journal format turns that distinction into a practical habit: preserve your thinking before the outcome gives you a better story.

08How solid is this?

Hindsight bias and outcome bias are well documented, giving a strong rationale for preserving expectations before results arrive. Direct evidence that decision journals themselves improve real-world decision performance is limited; their value depends on honest entries and thoughtful reviews.

10Origin and sources

A decision-practice tradition with no established sole inventor. Peter Drucker described a closely related method, feedback analysis, in Managing Oneself (1999); investing and forecasting writers later promoted decision journals.

[1]Drucker, P. F. (1999). Managing oneself. Harvard Business Review, 77(2), 64–74, 185.
[2]Fischhoff, B. (1975). Hindsight ≠ foresight: The effect of outcome knowledge on judgment under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1(3), 288–299.
[3]Baron, J., & Hershey, J. C. (1988). Outcome bias in decision evaluation. Journal of Personality and Social Psychology, 54(4), 569–579.
[4]Duke, A. (2018). Thinking in Bets: Making Smarter Decisions When You Don't Have All the Facts. Portfolio.

Suggest an edit· Updated 2026-10-02