How do you measure whether a journal review actually changed your decisions?

Most trade journals record what happened and why, but few people test whether reviewing the journal changed anything afterward. Without a measure, a review can feel productive while the same mistakes keep recurring. One approach from a quantitative angle is to tag each entry with a small set of error categories before the review, then count how often each category appears across a fixed window, such as the last 50 logged decisions. Comparing those counts before and after a review period gives a rough sense of whether anything shifted. The sample is small, so treat any change with caution and watch for regression to the mean. I'd be curious how others define the categories themselves. Do you tag by process step, by market condition, or by the emotion you noted at the time, and how many entries did you need before the pattern felt stable enough to act on?