Skip to main content
Making better decisions

Decision journal apps

5 min read

Search for a decision journal app and most of what comes back is a notes template with a nicer icon. That isn't a complaint about notes apps, it's that the two words hide a real difference.

A journal holds what you wrote. A decision journal has to hold what you predicted before you knew the outcome, a number for how sure you were, what happened, and the score, and it has to bring the decision back to you on a date you'd otherwise forget. Only some tools can do the second half, and the ones that can were built for slightly different things.

Here's what each kind does, where it stops, and what I'd check if you're choosing.

A notes app, or paper

Apple Notes, Obsidian, a paper notebook. Zero setup, and the template pastes straight in. This is the right place to find out whether you'll keep the habit at all. The reasoning, the part your future self can't reconstruct, is comfortable to write here.

Where it stops: nothing counts. After 20 entries you have 20 pages and no idea whether the calls you made at 80% confidence actually came in 8 times out of 10, unless you tally them by hand, and nothing brings a decision back on its review date. The second look, 30 to 60 days after the outcome, is the one nobody comes back for, because nothing asks.

Notion

Notion fixes the remembering half. Filtered views show you what's due, and a chart splits your resolved calls by confidence band into right and wrong. The setup is its own page, with a template to duplicate. Writing's comfortable there, which matters quite a bit more than it sounds.

Where it stops: the chart gives you counts to eyeball, one cut at a time, so calibration per domain means a view per domain, rebuilt by hand. And the second look still has no reminder, because you can't hang one on a formula.

A spreadsheet

A sheet is the opposite trade: the worst place to write, the best free place to count. Four formulas give you a hit rate per confidence band and a Brier score, and 2 conditional formats colour what's due. But if the measurement is what you're after, this is the fastest way to it.

Where it stops: the reasoning usually shrinks to 3 words, because a cell is a bad place for a paragraph, calibration per domain is copy-paste, and it only prompts you on the days you open the file.

Prediction trackers

Fatebook is the best of these, I think, and free: type a question, a probability and a resolve-by date, and it emails you when the date comes, then keeps your Brier score and a calibration chart as the resolutions accumulate. Private by default, with Slack and browser integrations if you want them. If your decisions are the kind that resolve cleanly, yes or no, on a date, it's excellent, and probably faster to log than anything else on this page.

Where it stops: it's built for forecasts, not decisions. A decision has reasoning, a state of mind, things that would change your mind, check-ins while it's open, and the delayed question of whether you'd make the same call again; a tracker just holds a question and an answer. And plenty of decisions don't resolve on a date, or to a yes or no. Was taking the contract the right call? That's felt for a year, and your sense of how it went keeps moving long after you supposedly know.

Metaculus and Manifold are forecasting platforms: world events, scored against a community, in public. Good calibration practice if world events are your material, and nothing to do with your own decisions.

What to check

Whatever you pick, a decision journal app has to do these, and the list is short enough that I'd run through it in a free trial:

  • Lock the prediction and the confidence number to the date you wrote them. If either can still be edited after the outcome, hindsight will get to it sooner or later, and the record stops meaning anything.
  • Remind you about the decision: on the review date, and again a month or two after it resolves, without you remembering to look.
  • Score by confidence, and by domain once you have enough, so it can tell you that the calls you made at 70% confidence came true 55% of the time.
  • Keep the reasoning comfortable to write, because 3 words of reasoning is a journal you'll stop keeping.
  • Keep the data yours: on your device or in your own account, private by default.

Notes apps fail the first 3, Notion and Sheets the second, and a tracker the fourth. Pick the failure you can live with, or the one you'd rather pay to remove.

More guides in Making better decisions.

Full disclosure: I make Reckon, and it's the kind this page hasn't listed yet: a decision journal app proper, for iPhone and iPad, built around the list above. The prediction and the number lock to their date, the review comes to you and the second look arrives on its own a month or two after resolution, and the calibration splits by band and by domain once you have around 15 resolved calls. One-time, no subscription, no account, syncs over iCloud. It's iPhone and iPad only for now, with a Mac version in progress, and if Fatebook's shape fits your decisions, use Fatebook; it's free and I recommend it either way.