First: stop doing these four things
Each is effortful, each feels like studying, and none of them require you to retrieve anything. In the final month they are pure displacement.
- Rewriting notes. Copying is a transcription exercise. If your notes are not usable now, you do not have time to fix them — work from the ones you have.
- Making new summaries. The one exception: a summary written from a blank page, without looking, which is retrieval wearing a summary costume. If the textbook is open, it is copying.
- Highlighting. Fluency theatre. It marks what looks important and teaches you nothing about whether you can produce it.
- Building the timetable. Two hours in a spreadsheet with colour-coded blocks, abandoned by Thursday. A list of what to do next is enough.
Choosing what to study when you cannot cover everything
You cannot cover everything, and pretending otherwise is how people end up uniformly mediocre across a whole syllabus. Triage, in this order:
- Things you have already lost marks on. Documented weaknesses beat suspected ones. If you keep an error log, this is simply the list, sorted by how often each topic appears.
- High-frequency, high-mark topics. Flick through five past papers and count what appears every single year, weighted by marks. Some topics carry 15 marks a paper and some carry 3 in one year out of four.
- Topics you avoid. The ones you have quietly never practised. Avoidance is a reliable signal, and these are usually cheaper to fix than you assume — you have been dreading them rather than attempting them.
What comes off the list: material you can already do reliably, and material that is both rare and enormous. Deciding in week one that you are not going to master one obscure topic is a legitimate strategic choice. Discovering it accidentally in week four is not.
The four weeks
Diagnose, and close real content gaps
The only week where learning new content is a good use of time. Sit one full paper untimed and open-ended early in the week, purely as a diagnostic — you want to find out what you cannot do, not what you can do in three hours.
- Mark it brutally, then sort every lost mark into: didn't know / misread / slipped / ran out of time.
- The "didn't know" pile is this week's work. That is the last window where sitting down with a textbook pays.
- Build the shortlist from the triage above. Write it on one page. That page is your plan; you do not need a timetable.
Papers by topic, reviewed hard
Not full papers yet. Pull the questions on your shortlist topics from across every past paper you can find and do them in blocks — eight questions on the same skill, back to back, teaches that skill faster than eight scattered across four papers.
- Attempt cold, from blank paper, mark scheme closed.
- Every lost mark gets logged with a one-line cause. This is the week the log fills up and starts being useful.
- Re-attempt anything you got wrong three days later, cold — not by re-reading your corrections.
- Running out of past-paper questions on a topic is normal by now; generated exam-style questions keep the supply going without you re-doing papers you have half-memorised.
Full papers under exam conditions
Now timing matters. Two or three full papers this week, each sat properly — real time limit, phone in another room, no notes, one sitting, ideally at the time of day your real exam is.
- Review each one the way papers are meant to be reviewed: two to three hours per paper, cause per lost mark.
- Watch the mix of causes change. By now most losses should be misreads, slips and timing rather than content — if they are not, your shortlist was wrong and needs revisiting.
- Build the checking routine for your specific recurring slips and practise it in these papers, not for the first time in the exam hall.
- Fix pacing explicitly: decide in advance how many minutes a mark is worth and when you abandon a question.
Retrieval only, volume down, sleep up
No new content, no new topics, nothing you have never seen. This week is entirely about keeping what you have accessible and walking in steady.
- Short daily retrieval sessions on your logged mistakes — twenty minutes of questions you previously got wrong is the single highest-density revision that exists for you.
- One last timed paper early in the week, maximum. Not two days out.
- Formulae and definitions you have historically forgotten: blank-page them daily.
- Sleep on the real schedule. Waking up at 6am on exam day after a fortnight of 2am finishes costs more marks than the sessions gained.
The last 48 hours
| Do | Don't |
|---|---|
| Read your error log end to end — it is the concentrated version of the whole year | Start a topic you have never studied because a friend mentioned it |
| A few questions on your strongest topics, to walk in feeling capable | Sit a full paper the day before and spook yourself with a bad score |
| Check logistics — time, venue, calculator, equipment, spare batteries | Compare revision totals with other people in a group chat |
| Stop by early evening; sleep properly | Cram overnight and arrive with degraded working memory |
The evidence on the last one is unusually blunt: sleep deprivation degrades exactly the faculties an exam taxes — working memory, attention, error-checking. A night of cramming typically buys a small amount of fragile recall and costs a chunk of the machinery you need to use everything else you know.
If you are genuinely behind
Not "feel behind" — actually behind, with topics never covered. Then the plan compresses and the triage gets ruthless:
- Count marks per topic across five past papers. Rank. Work strictly down that ranking.
- For each topic, learn only enough to attempt the standard question — the one that appears in some form every year. Not the hardest extension part.
- Accept explicit write-offs. Deliberately dropping two rare topics to be solid on eight common ones scores better than being vague on all ten.
- Do not start with the subject you like. Start where the marks are recoverable per hour, which is usually a subject you dislike and are close to a boundary in.
Keeping the plan honest
Everything above depends on one thing: knowing what you actually get wrong, not what you feel shaky about. Self-assessment is unreliable in both directions — people avoid topics they are fine at and stay confident about topics quietly costing them marks every paper.
errorboard exists to make that record cost nothing. Photograph a question you got wrong and it reads the page, works out the cause, categorises it and files it by topic — so the shortlist in week 4 is a real ranked list rather than a guess, the review quiz brings each mistake back on schedule through weeks 3 and 2, and by the final week your error log is the only document you need to read.
You can also give it your exam date and it will work backwards, telling you what to prioritise as the weeks close in — including flagging topics that have gone quiet long enough to be worth re-testing.
The plan in five lines
- Week 4: diagnose with one untimed paper; close real content gaps.
- Week 3: topic blocks of past-paper questions, reviewed hard, everything logged.
- Week 2: full papers under real conditions; fix pacing and your slips.
- Week 1: retrieval on logged mistakes only; volume down, sleep up.
- Last 48 hours: read the log, light questions, logistics, sleep.
No colour-coding required. The plan is a one-page list of topics you got wrong, worked through in order of what they cost you.