Decision log — entry template

Most founders try to remember why they decided things. They are not very good at it. By month six, the team has reorganised once, the strategy has shifted twice, and three of the assumptions behind the original calls are no longer true. Without a log, the decisions look obvious in hindsight, the alternatives look weaker than they were, and the founder learns nothing.

A decision log is two minutes per entry. The compounding payoff arrives later, when you can read your own thinking from before the outcome was known.

When to log

• Any decision that takes more than thirty minutes of debate.
• Any decision a board member or investor would expect to be able to ask about later.
• Any decision where you can name two alternatives that lost.
• Any decision you have flipped on. (The flip is itself a logged decision.)

You are not logging every choice. You are logging the ones whose context will fade and whose outcome will shape the next one.

The six-field entry

One line each. The format is rigid on purpose — rigidity makes the log scannable.

• Date: ______
• Decision: a single sentence stating what was decided. Active voice.
• Alternatives: two or three options that lost, named, with one phrase each on why.
• Why this one: the deciding consideration. Not the full case. The consideration that closed it.
• Expected outcome: what we expect to see by [date], in concrete terms a third party could check.
• Review date: when we look at this decision again with the outcome in hand.

Worked example

A real-shape entry; the data is illustrative.

• Date: 2026-04-08
• Decision: We will price the standard tier at $79/month, not $49 or $99.
• Alternatives: $49 (loses to a near-competitor on tooling depth, cannot fund the support cost); $99 (clears the tooling-depth bar but our current case studies do not justify the premium).
• Why this one: $79 lets us hold the same tooling depth without forcing us to add the case studies we do not yet have.
• Expected outcome: by 2026-07-01, we expect $79 conversion to land within ten percent of the current $49 conversion at full quality cohort.
• Review date: 2026-07-15.

On review day

Two questions; ten minutes.

1. Did the expected outcome happen, with the assumptions still intact?
2. If not, was the call wrong or were the assumptions wrong? Log a follow-up entry with the new decision.

You will be wrong about a third of the time. The log makes that visible. The visibility is the engine of the calibration.

What not to do

• Do not write the case for the decision. Write the deciding consideration. The case lives in your head and in the meeting notes; the log is for the consideration that flipped it.
• Do not edit past entries when the outcome lands. The wrong-at-the-time entry is the most valuable one in the log.
• Do not store the log in a tool that requires logging in. Friction kills logs. A pinned doc, a Notion page, a markdown file in the repo. Anywhere you will actually open at 4pm on a Thursday.

Decisions are most useful as a record of how you thought before the outcome arrived. Without the log, you only get to learn from the outcome. With it, you get to learn from the call.

Template from Marga Haus · margahaus.com/resources · Adapt and use freely. Attribution appreciated, not required.