Organizations invest heavily in seeing their numbers and almost nothing in recording what they decided because of them. The result is a familiar pattern: the same metric dips each quarter, the same meeting reaches a similar conclusion, and nobody can say whether the last intervention worked. The missing piece isn't a better chart. It's memory.
What a decision log is
A decision log is a short, structured record attached to the metrics you already track. Each entry captures what the data showed, what was decided, who owns the follow-up, what outcome was expected, and when it will be checked. It takes a minute to write and answers the questions that otherwise vanish: why did we do that, and did it work?
The five fields that matter
- Signal — the metric movement or alert that triggered the discussion, with a link to the data at that moment.
- Decision — what was chosen, including what was deliberately not done.
- Owner and date — one named person and a review date. Without both, the entry is a diary.
- Expected effect — a specific, checkable prediction, such as 'response time falls below one day within six weeks'.
- Result — filled in at the review date, including when the prediction was wrong. The misses are the valuable entries.
A log full of confirmed predictions suggests nobody is taking real risks. A log with honest misses is how a team learns what actually moves its numbers.
Where AI helps, and where it doesn't
Language models are good at the clerical parts: drafting the entry from a meeting summary, linking it to the relevant metric, and pulling up similar past decisions when a signal repeats. They should not decide the entry's content or grade the result — that judgment belongs to the owner. Used this way, the system makes the habit nearly free without replacing the thinking.
What it changes
After a few months the log answers questions no dashboard can: which kinds of intervention tend to work, which signals deserve attention, and where forecasts consistently mislead. That is the feedback loop that makes the next forecast, and the next decision, better. Analytics stops being a reporting function and starts compounding.