A decision you made eighteen months ago turned out badly, and you genuinely cannot remember what you were thinking at the time. Was the data wrong? Were the assumptions reasonable? Did you ignore a signal you'd noticed? With nothing written down, the post-mortem is mostly fiction — your present self reconstructing the past self's reasoning, which is the version most flattering to your present self.
A decision journal solves a small piece of this honestly. You write down the decision when you make it, in plain language, before you know how it turns out. Months later, when the outcome is in, you can compare the reasoning you actually used against what actually happened. The judgments don't get rosier with time because they're written down.
This isn't about being a more rational person. It's about giving your future self enough information to learn from your present self.
What a decision journal actually captures
The mistake most people make is over-engineering the format. A decision journal that requires fifteen fields and a forty-five-minute writeup gets used twice and abandoned. The lightweight version captures four things and lets the agent do the rest of the work later.
For each decision, write down:
- The decision and the date. Plain English, one sentence. "Took the senior engineer offer at [Company] over staying at [Other Company]."
- What you knew at the time. The facts available, the inputs you considered. Not what you wished you'd known — what you actually had.
- What you assumed. The things you weren't sure about but treated as probably true. "Assumed the team would actually ship the migration this quarter."
- What you expected to happen. A specific prediction. "Expected to hit director track in 18-24 months."
That's it. Five to ten minutes per entry. The discipline isn't in the writing — it's in writing it before the outcome is known. After-the-fact reasoning is biased even when you mean to be honest. The pattern shares a lot with the broader habit covered in The Capture Habit: Remembering the Things That Actually Matter — the value comes from getting it down before it fades.
One page per decision, in a single decisions folder
In Docapybara, each decision gets its own page under a parent Decisions page. Pages nest with no depth limit, so you can group by category if it helps — Career, Money, Relationships, Health — or just keep one flat list and let the agent search across.
The page title is the decision in one line plus the date. "2026-04-15 Bought the second car instead of getting it repaired." You'll thank yourself later for the date in the title; the agent can also pull by date but it makes scanning the list easier.
For decisions that span weeks of deliberation — a job change, a major purchase, a relationship — the page can hold the journal entry plus child pages for the research you did along the way. The actual decision-recording lives in one place; the supporting work has somewhere to go that doesn't clutter the main entry. For the broader workflow of weighing options before you commit, see Capture and Compare Options for Any Major Decision.
The follow-up entry — what actually happened
Six months later, twelve months later, three years later, the entry gets a follow-up. Same page, new section underneath. "What actually happened" with the date.
This is the part most journaling systems neglect. The original entry is easy to write because you're motivated; the follow-up requires going back to a page you'd rather not revisit. The fix is making it part of a regular review cadence — pick a frequency (quarterly or twice a year works for most people) and ask the agent to surface decisions whose outcomes you haven't recorded yet.
"Pull all decisions from more than six months ago that don't have a follow-up entry yet." You get the list. Open them one at a time, write what happened, move on. Twenty minutes for a quarterly review.
The follow-up should describe the actual outcome and — separately — your honest assessment of the original reasoning. The two aren't the same. A decision can turn out well from bad reasoning, and badly from good reasoning. The journal helps you tell the difference.
Asking the agent to find patterns
The reason a decision journal pays off over years rather than months is that the agent can read across all of it when you ask the right question. Single decisions are noisy. Patterns across thirty decisions are signal.
Useful questions to ask after you have a year of entries:
- "Look at every career decision I've journaled. What kinds of reasoning have tended to lead to outcomes I'm happy with vs. unhappy with?"
- "Which assumptions have I made repeatedly that turned out to be wrong?"
- "For decisions I expected to play out in a specific timeframe, how often did I get the timing right?"
- "Show me decisions where the original reasoning still looks sound to me even though the outcome was bad."
The agent reads across your actual entries and answers grounded in what you wrote. The patterns are about your own thinking, not generic advice. Some of these questions overlap with the deliberate self-knowledge work covered in How to Use AI Notes for Journaling and Daily Reflection — the decision journal is a more structured slice of the same broader habit.
Where voice helps and where it doesn't
For the decision entry itself, typing usually wins. The discipline of writing in clear sentences forces clarity that voice doesn't. A rambling voice note about a major decision feels productive at the time and reads as fog later.
For the follow-up entries, voice can work. "Twelve months later — the engineer job worked out fine. The migration didn't ship that quarter, did ship the next one. Director track turned out to be eighteen months exactly. The biggest assumption I got wrong was about the manager's tenure — they left after eight months." Voice catches the honest summary in a couple of minutes; the transcript drops on the page.
For decisions you're actively wrestling with, voice notes during a walk or drive can be useful raw material. They don't replace the written entry — they feed into it. The agent can take a voice transcript and help you draft the formal entry. "Turn the recording I just made into a decision journal entry following my standard format." You get a draft to refine.
A few decisions worth journaling that people often skip
Most people start a decision journal by capturing the big obvious calls — job changes, major purchases, relationship choices. Those are right to capture, but they're not where the highest learning rate lives. The pattern-rich material is in the medium-sized decisions you make often:
- Hiring or firing a service. Choosing between two contractors, switching tax preparers, hiring a coach.
- Recurring time commitments. Joining a board, taking on a side project, starting a class.
- Investments under a certain size. The ones that aren't existential but accumulate. "Bought $5k of [stock] because of [reason]."
- No-decisions. The deliberate choice not to do something. "Decided not to apply for [opportunity] because…" These are the hardest to remember and often the most informative in retrospect.
A vault of fifty or eighty entries across a couple of years gives the agent enough material to find real patterns. Three or four entries don't.
A starter format you can copy today
Here's the minimum that works. Five sections at the time of the decision:
- Decision — one sentence, with the date.
- What I knew at the time — bullet list of facts.
- What I assumed — bullet list of assumptions, marked as such.
- What I expected to happen — a specific prediction, with a timeframe if it applies.
- Why I'm choosing this — a short paragraph.
And later, when the outcome is in:
- What actually happened — the outcome, dated.
- Honest assessment of the original reasoning — was the reasoning sound? Did the outcome reflect the reasoning, or was it luck either way?
That's it. No tags, no template engine, no scoring rubric. The discipline is in the before, not the form.
The benefit isn't that you make better decisions because you wrote one down. It's that over a couple of years, you build a private record of how you actually think — not how you remember thinking — and the patterns you find about your own reasoning are more useful than any framework you'd download.
Try Docapybara free — start with one decision you're wrestling with this week, and a follow-up reminder for six months out. The vault holds the rest. You can read more about the agent that acts on your documents once your journal has a few entries.