Most teams that keep a decision log keep it like a diary. Each entry says what was decided, and nobody opens it again. This decision log template is built around the opposite habit: every entry carries a review date, and on that date you grade the call.
The record is short, it fits on one page, and you can copy it into a doc or a spreadsheet today.
Why a log of what was decided is not enough
A list of decisions tells you what happened. It does not tell you whether the decision was any good.
Without a check-back, a log turns into an archive. People stop reading it because it never changes the next decision. The useful part is the loop: write down the bet, set a date, look at what happened, and say plainly whether the bet paid off. Over a few months that shows you which kinds of calls your team gets right and whose judgment holds up.
When to log a decision
Do not log everything. Log a decision when one of these is true:
- It binds other people. Someone else has to change what they do.
- It is costly to undo. It spends money, time, or trust you cannot easily get back.
Choosing a lunch spot is not a decision record. Changing your pricing, hiring a first salesperson, or dropping a product line is.
The fields of a decision log template
Each record has six parts. Write them in plain words.
- Decision. One sentence on what you chose, and the main option you turned down.
- Owner. One person, not a team. The owner answers for the result and for the follow-up.
- Dissent. Who disagreed and why. Write the objection down even if you overruled it. This is the field people skip, and it is the one you will be glad of later.
- Confidence. A percentage, or low, medium, or high. It forces the room to say how sure it really is.
- What would prove it wrong. Something you could see within weeks, stated before you start. A number crossing a line, a deadline missed, a customer reaction. Not a feeling.
- Review date. A real calendar date, close enough that you will still remember the context. Put it on a calendar.
The last two fields do the most work. "What would prove it wrong" stops you from explaining away a bad result afterward. The review date makes sure the question gets asked at all.
A labeled example record
This is an example, not a real company's record.
- Decision: Move our weekly sales follow-up emails from Friday afternoon to Tuesday morning. We considered hiring a part-time assistant to send them and decided against it for now.
- Owner: Dana, head of sales.
- Dissent: Priya thinks Friday still works because customers read email on the weekend. She would rather test a second send than move it.
- Confidence: Medium, about 60 percent.
- What would prove it wrong: Reply rate on the follow-ups is no better after four Tuesday sends than it was across the last four Friday sends.
- Review date: Four weeks from the decision.
Notice what is on the page. A stranger could read it and know what was bet, who disagreed, and exactly what result would end the experiment.
The review: worked, mixed, or failed
On the review date, open the record and look at what happened against what you wrote down. Then grade it with one of three words:
- Worked. The result matched or beat what you expected, and the cause was the decision rather than luck.
- Mixed. Part of it held up and part did not, or the result is real but smaller than you hoped.
- Failed. The thing that would prove it wrong happened, or the decision made things worse.
Then write one line on why. Not a post-mortem, one line. "Reply rate rose on Tuesday, but only for existing customers; new leads did not change." That line is what you will find useful in a year.
Two cautions. First, grade the decision, not the outcome alone. A good call can have a bad result, and a lucky guess can have a good one. Second, note how the dissent held up. If Priya was right, say so. That is how the record earns trust.
Keep it to one page and a few decisions a week
A log that takes an hour to maintain will be dropped. Keep each record to the six fields and cap yourself at a few decisions a week. If you are logging more, you are probably logging things that do not bind anyone or cost much to undo.
The best place to write the record is the weekly leadership meeting itself, in the last minutes, while the room still agrees on what was decided. If you do not have a meeting format yet, the Weekly Leadership Meeting Kit has an agenda, a scorecard sheet, and a one-page decision record you can copy.
How Cabinet keeps the record and runs the review
The hard part of this system is not the template. It is that someone has to write the record, remember the date, and ask the question. In a small company that someone is usually the owner, and the review is the first thing to slip.
Cabinet is agntwrk's AI leadership team, and it runs this loop for us. Each decision from the weekly meeting is recorded with an owner, dissent, confidence, what would prove it wrong, and a review date. On that date the team asks how it turned out and grades it worked, mixed, or failed. Cabinet is in an invite-only beta, new workspaces are approved by a person, and the owner approves what matters under rules the owner sets. You can see what each plan includes on the pricing page.
You do not need Cabinet to use the template. Copy the six fields, set the dates, and grade honestly. If you would rather have the record kept and the review run for you: