How to keep a decision journal: what to write, and when to look back

What each field is for, and why the order matters

The table follows the book's tool card for the decision journal. The first six fields happen before you choose. The last two happen on the review date. Keep that order. It is the whole point.

One rule matters more than the rest. On the review date, write what happened first. Only then reread the old entry. If you reread first, the old words start to bend toward the result.

The number in the forecast field matters most. "This hire will work out" cannot be checked. "Seventy percent chance she is still here and doing well in a year" can. Write enough of those and you can see if your 70 percent calls come true about 70 percent of the time.

This is not a diary. You do not write about your day. You write before choices that matter and repeat, the kind where you find out later how it went.

Why your memory cannot be the record

In 1975, a study on remembered guesses asked people to guess the odds of events around President Nixon's 1972 trips. Later they were asked to recall their guesses. They remembered giving higher odds to the things that had happened.

This is called hindsight bias. It is not lying. It is how memory works once you know the end. A 2004 review pooled 95 studies and found a moderate effect again and again.

And it hurts learning. To get better, you compare what you expected with what happened. Hindsight bias quietly edits the first half.

A year later, you open the notebook

Picture a hiring choice. You wrote that you were 80 percent sure. You listed her references and her sample project as the main reasons. In small letters at the bottom you also wrote: rushed, need someone by Monday.

She left after four months. Without the entry, you would remember having doubts all along. With it, you see the real lesson. The deadline made the call, not the candidate.

That small mood line is easy to skip. It can turn out to be the most useful line on the page.

Which choices to log, and which to skip

Most guides say to log only decisions that matter. The book is more specific. The journal pays off when you make the same kind of choice again and again and find out later how it went: hiring, projects, purchases, investments, bets on what a customer wants.

It does much less for one-off choices you will never repeat. And a quick call you can undo by tomorrow does not need ten minutes of writing. Keep the notebook for the decisions where a pattern could show up.

Reading many entries at once

One entry tells you about one choice. The real lesson shows up across many. The book's sister card for this is Keep Score, and it carries a SOLID badge.

Every 20 forecasts, group your entries by how sure you were. Then check. Of the things you called 80 percent likely, how many happened? If it was closer to half, you run too sure on that kind of call. Write one sentence about what you will change, such as "Too sure about clients you like," and keep going.

That is why the number in the forecast field matters. Words like "probably" cannot be scored. Numbers can.

Grade the choice apart from the result

The journal gives you an honest record. The book pairs it with a second card on how to read it.

Write the outcome in one line, then cover it. List what you knew back then. Grade the process from 1 to 5. Then uncover the outcome and grade it apart. Place the choice in one of four boxes: good choice and good result, good choice and bad result, bad choice and good result, or bad choice and bad result.

Of the four, we find the bad choice, good result box the most useful. It is where you got away with something. A 2023 study with 692 people found that people judge the same choice as better when it ends well, even when they say outcomes should not count.

The chapters behind the journal

This is the problem Mental Models That Actually Work was written for.

Decision Journals has both cards: The Decision Journal and Judge the Decision, Not the Outcome. It also sorts out the Kahneman story, which comes from a retelling, not his own writing.

Calibration: How Sure Should You Be? shows what to do with the numbers in your forecast field once you have a few dozen. It has a way to keep score on your own calls.

The Decision Selector and the 30-Day Plan has a blank Decision Log page and a calibration tracker you can copy, plus a month of short drills that build the habit.