How to run a weekly error review

Mistakes evaporate unless something catches them

Every mock produces twenty pieces of evidence about how you answer papers. By the following Sunday, most students can recall two of them: the score and the question they are still annoyed about.

That is the loss. Not the marks — the information. The score tells you what happened; the specific errors tell you what to change, and they are gone within a week unless they are written down.

A weekly error review is the smallest process that fixes this: thirty minutes, one page, every week.

It is a log of errors, not a notebook of theory

The review is not a revision notebook and not a summary of chapters. One line per error, written at the moment you find it, in your own words.

If an entry takes more than two minutes to write, it is too long. The value is in the specificity, not the prose.

The four columns

Keep one page per paper, with four columns:

• Where — question number, paper, and the sub-part.
• What happened — the error, stated as fact: missed the second half of the question, used the wrong rate, left the working out.
• Why — one of three: did not know, could not recall, or misread/mismanaged.
• The fix — one action, small enough to do: re-attempt this question timed, memorise this format, read the verbs before starting.

The "why" column is what makes the page useful. It decides whether the fix is content work, recall work, or a behaviour change — and those three are not interchangeable.

The thirty-minute Sunday process

Minutes 0-10: collect. Pull the week's marked mocks and practice sets. One line per error, no editing, no analysis yet.

Minutes 10-20: sort. Count the lines in each "why" bucket. The largest bucket is next week's priority, and it is usually misread or mismanaged — not content.

Minutes 20-28: choose three. Pick the three entries that cost the most marks, and write the one action for each on a separate strip of paper. These are next week's first work, before anything new.

Minutes 28-30: close the page. Date it and stop. A review that runs to ninety minutes is a revision session wearing a disguise, and it will not happen next week.

What to do with the output

Three actions, no more. If the review produces eleven actions it produces none, because none of them will be the first thing you do on Monday.

Put the three actions where you will physically start the week: top of the agenda, on the desk, or as the first block of your plan. The log is not the outcome; the three actions are.

The honesty problem

The most common way an error review becomes useless is writing the answer you wish you had written instead of the one you did.

If you lost marks for a wrong format, write "did not include the working". If you skipped a part, write "answered three of four". The suggested answer is a reference, not a draft of your log entry.

A useful test: would this line make sense to you in six weeks, with no memory of the mock? If not, it is too vague.

When to retire an entry

Keep an entry until the same error stops appearing. Once a specific mistake has not recurred across two mocks, cross it out — not delete it. The crossed-out lines are the only real progress record you will have, and they are more motivating than any score.

The one-line version

Collect every error the week it happens, sort by cause, act on three — and let the crossed-out lines do the encouraging.