How to keep a decision journal and measure your decision quality
The essentials in 30 seconds
- A decision journal records the question, options, estimates, and reasons at the time of the choice.
- Decision quality is scored on a process checklist of 5 items, separate from the outcome.
- In the illustrative log of 10 decisions, 7 show a good process and 6 a good outcome.
- Eight of 10 entries pass the regret check: the same choice again with the same information.
- The EQOS Asia decision engine returns a score that can be recomputed identically (EQOS Asia method page, as of October 2026).
Most executives remember outcomes and forget reasoning. A decision journal reverses that: it records the question, the options, the estimates, and the reasons at the time of the choice, so the choice can be judged later on its own merits. This guide shows what to record, how to score decision quality apart from results, and how a journal fits among the decision making frameworks.
On this page
Compare the EQOS Asia offers and request a 30-minute pre-diagnostic.
What is a decision journal and why keep one?
A decision journal is a dated record of important decisions, written when they are made. Each entry captures the question, the options considered, the estimates, the reasons, and the expected result. The entry is short, often one page, and it stays unchanged after the outcome is known. A paper notebook or a shared document serves equally well.
The journal solves a memory problem. After an outcome arrives, people recall their reasoning as more consistent with the result than it was. A written entry preserves the original reasoning, so a later review judges the process that was used and not the story told afterwards.
Which fields should a decision journal entry contain?
Seven fields cover most decisions: the question in one sentence, the options, the main assumptions, the estimates as ranges with a confidence level, the expected outcome with a date, the people consulted, and the state of the decider, such as tired or rushed. The last field sounds minor and often explains a miss.
Add a review date at the time of writing. A date in the calendar turns the journal from a diary into a feedback loop. The figure shows the cycle: record, wait, review, score, and adjust the next decision. Each stage takes a few minutes.
A fixed form makes the habit easier to keep. The page on a decision journal template lists 8 fields to fill in before the decision and 5 at the review, and completes them on an illustrative hiring entry, so that a first entry takes about 10 minutes.
How do you score the quality of a decision, apart from its outcome?
Score the process with five yes or no items, each worth one point: at least three options were considered, the key evidence has a source, the main risks were rated, a dissenting view was heard, and a review date was set. A total of 4 or 5 marks a good process, and 0 to 3 marks a weak one. The checklist takes two minutes to complete at the review.
The score stays independent of how things turned out. A decision that scores 5 and ends badly reflects a risk that was priced and then materialised. A decision that scores 2 and ends well reflects luck. Keeping the two apart keeps the team learning from the right signal.
Beyond a single entry, a team can track the numbers over a quarter. The page on decision quality metrics builds a scorecard of 6 measures, from the average process score to the reversal rate, with targets that keep the numbers honest and a light audit of the entries.
How do process and outcome combine in four decision quality cases?
Cross the two judgments and four cases appear. Good process with a good outcome earns repetition. Good process with a bad outcome calls for patience, since the reasoning may have been sound. Bad process with a good outcome calls for caution, since luck rarely repeats. Bad process with a bad outcome calls for a change of routine.
Take an illustrative log of 10 decisions: 5 in the first case, 2 in the second, 1 in the third, and 2 in the fourth. Seven decisions (70 percent) show a good process, and 6 (60 percent) show a good outcome. The two measures differ, and the gap between them is the value of keeping them apart.
How does regret fit into a decision journal?
Regret is the feeling attached to an outcome, and it tends to follow results and not reasoning. A regret check at the review asks two questions. Would I take the same decision again, knowing only what I knew then? And how large is my regret about the outcome, from 0 to 10? Both answers go into the entry beside the original reasoning.
In the illustrative log, 8 of 10 entries (80 percent) pass the first question, so the team would repeat the choice. Two of the decisions with a bad outcome still pass, and the regret rating of those entries is high while the process stays sound. Reading both columns separates a sting from a lesson.
How often should a decision journal be reviewed?
Review each entry on its own date, and review the whole journal every quarter. The entry review asks what happened and which assumption moved. The quarterly review looks for patterns: which entry field is skipped most often, which kind of estimate runs optimistic, and which person or setting precedes weak decisions. These patterns point to the one habit worth changing first.
Ten entries show tendencies and leave room for chance, so treat early conclusions as hypotheses. After a year the log holds enough cases to change one habit at a time. A single change, such as always hearing a dissenting view, is easier to maintain and to measure than a package of resolutions.
How does the EQOS Asia decision engine support a decision journal?
A journal works best when a decision leaves a record that can be replayed. The Codex Universalis core is deterministic, so every verdict can be recomputed identically from the same inputs (EQOS Asia method page, as of October 2026). The inputs and the score together form a ready-made entry.
The Sentinel service recomputes the model monthly, which provides a natural review date, and the Lambda Comex dashboard is shared across the team (EQOS Asia method page, as of October 2026). A 30-minute pre-diagnostic, scheduled in Singapore time (SGT), applies the equation to a real case, and the simulator offers a first score to log.
See also in this guide
Frequently asked questions
What should a decision journal entry contain?
Seven fields cover most decisions: the question, the options, the main assumptions, estimates as ranges with a confidence level, the expected outcome with a date, the people consulted, and the state of the decider. Add a review date at the time of writing so the entry becomes a feedback loop.
How can I judge a decision apart from its outcome?
Score the process with a five item checklist: options considered, evidence sourced, risks rated, a dissenting view heard, and a review date set. A score of 4 or 5 marks a good process. The outcome is recorded separately, so luck and reasoning stay distinct.
How long should a decision journal entry take?
Ten to fifteen minutes is enough for most decisions. A short entry written on the day is worth more than a long one reconstructed later. Reserve a longer entry for decisions that are costly to reverse, and use the same template each time so entries can be compared.
Can a decision engine feed a decision journal?
Yes. A deterministic engine returns the same score for the same inputs, so the inputs and the score form a record that can be replayed at review time. In the EQOS Asia method, the Sentinel service recomputes the model monthly (EQOS Asia method page, as of October 2026).
Test your knowledge
1. When is a decision journal entry written?
- After the outcome is known
- At the time of the decision
- At the quarterly review
Show the answer
At the time of the decision.
2. What score marks a good process in the five item checklist?
- 1 or 2
- 3 only
- 4 or 5
Show the answer
4 or 5.
3. In the illustrative log, how many of the 10 decisions show a good process?
- 7
- 6
- 8
Show the answer
7.
4. What does the first regret check question ask?
- Who is to blame
- Would I choose the same again with the same information
- How much did it cost
Show the answer
Would I choose the same again with the same information.
5. How often does the EQOS Asia Sentinel service recompute the model?
- Daily
- Yearly
- Monthly
Show the answer
Monthly.
About the author
Stéphane Jambu
President, EQOS Forecast Asia
Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile
Sources: EQOS Asia method page (eqos-asia.com/method.html), simulator page (eqos-asia.com/simulator.html), and contact page (eqos-asia.com/contact.html), consulted on 21 October 2026. Page updated on 21 October 2026.