A review year generates a lot of numbers: hours, pages, videos, streaks, mock scores. Almost all of them are activity metrics, and activity is what you do when you cannot measure results.
Four numbers do predict, and they take ten minutes a week to record.
The four
1. Timed answers written and marked. Count them. This is the single best predictor of September, because it is the only activity that rehearses the actual task.
2. Topics rebuilt from memory, clean. Not "revised": rebuilt, closed-book, against the outline, with the misses counted. A topic that fails is not a topic done.
3. Dashes remaining. From the syllabus audit — the topics you cannot place at all. This number should fall; if it is flat for two months, the study is not reaching the syllabus — see reading the syllabus.
4. Delayed retention. Ten doctrines from a subject finished three weeks ago, tested closed-book. Seven or better means the schedule is spaced; four means it is merely sequential.
That is it. Four numbers, once a week, in one line.
What to ignore
- Hours in the chair. An hour of retrieval and an hour of highlighting are both an hour.
- Pages read. Unmeasured, and trivially gamed.
- Streaks. A streak rewards showing up; the examination rewards producing. Broken streaks also produce a specific, demoralising despair that costs more days than the streak was worth.
- Mock scores, as a single number. A mock is a sample, marked by you; the trend across three mocks is data, one is noise.
- Mood. Useful for burnout, useless for progress — see handling review burnout.
The weekly line
One line per week, in one file:
Wk 14: timed 2 (both marked); topics rebuilt 3 (2 clean); dashes 41→38 (Civil); retention 7/10 (Labor). Next: 3 timed, Civil succession dashes, Labor retention 8.
The line is not a diary. It contains only numbers, direction, and the next week's target. Fourteen weeks of those lines is a picture of the year, and it takes ten minutes a week.
What the numbers do to the plan
- Timed answers flat at zero for three weeks → the plan has no application section; add one, cut coverage.
- Dashes flat → the study is not touching the syllabus; re-audit and re-prioritise.
- Rebuilds high but retention low → the schedule is blocked, not spaced; interleave — interleaving versus blocking.
- Everything falling → the plan is too big; compress per the last ninety days and protect retrieval and sleep.
Why so few numbers
Because the point of measurement in a review year is not self-knowledge; it is correction. Four numbers that change your next week are worth more than forty that describe your last one.
And because the log has to survive a bad week. A ten-minute weekly line survives; a daily dashboard does not.
The one question
If you record nothing else, record this each week: what can I now produce that I could not produce last week?
Three specific doctrines is a good week. "I read a lot" is a week that happened on paper.