How to Write a Decision Journal

Why you need a record

Human memory of past decisions is not a reliable archive. It is a reconstruction, and that reconstruction is heavily influenced by what we now know about how the decision turned out. If the outcome was good, we remember our reasoning as having been sound. If the outcome was bad, we remember external factors having intervened, bad luck, circumstances we could not have foreseen. This pattern, known as hindsight bias, makes it extremely difficult to learn from experience without an external record.

Investor Annie Duke documented this problem in detail in "Thinking in Bets" (2018). She observed that most people, including professionals who make high-stakes decisions regularly, evaluate decisions by their outcomes rather than their reasoning. She calls this "resulting": working backward from a good or bad result to a judgement about whether the decision was good or bad. The problem is that good decisions sometimes produce bad outcomes, and bad decisions sometimes produce good outcomes. If you judge only by results, you will reinforce bad reasoning that happened to work out and abandon good reasoning that happened to fail.

The decision journal is the antidote to resulting. It captures your reasoning at the moment of decision, before you know the outcome. When you review the entry later, you have a record of what you actually thought at the time, which means you can evaluate the quality of your reasoning independently of whether you got lucky.

This is not a new idea. Charles Darwin famously kept detailed notes on observations and arguments, specifically because he distrusted his own memory and knew that he would be tempted to forget inconvenient data. Richard Feynman kept problem journals that preserved his reasoning process, not just his conclusions. The journal as a tool for intellectual accountability has a long history; its application to personal decision-making is underused.

What to capture in each entry

A decision journal entry has two parts: the entry made at the time of the decision, and the review entries made at 30, 90, and 180 days.

The initial entry should capture: the decision in one sentence; the date and the context (why this decision is arising now); the alternatives you considered and why you rejected each one; the mental model or framework you used to analyse the decision; the two or three biases you identified as most likely to be active, and what you did to account for them; the expected outcome stated specifically (not "it will probably go well" but "I expect to reach X milestone by Y date"); and your confidence level as a percentage.

The confidence percentage is important. Over time, it allows you to track your calibration: are your 80% confident predictions right 80% of the time? Most people find their high-confidence predictions are right less often than they expect, which is direct evidence of overconfidence bias in their own reasoning. The journal makes this visible in a way that memory never can.

The review entries, written at the scheduled intervals, should capture: what has actually happened so far relative to the prediction; what the current confidence level is in the original expected outcome; what you would do differently if you were making the decision again with what you now know; and any patterns you notice connecting this decision to previous ones in your journal.

What format to use

The minimum viable version is a dedicated note in any app that timestamps entries: Apple Notes, Notion, Bear, Obsidian, or a physical notebook. What matters is that the entries are date-stamped, preserved without editing, and have a scheduled review reminder attached to them. The format is less important than the discipline of making the initial entry before you know the outcome and the discipline of returning to review it.

If you use DecisionsMatter.ai to run the 6-step analysis, the tool generates a structured record that serves as the journal entry. You can export it, save it, and return to it at the 90-day mark with the review prompts built into the platform.

Avoid the temptation to write long, reflective entries. The value of the journal is not in the quality of the prose but in the accuracy of the prediction and the honesty of the review. A two-paragraph entry that captures the decision clearly, the expected outcome specifically, and the reasoning transparently is more useful than a five-page reflection that arrives at the same decision through beautiful but untrackable logic.