Exams and deadlines: a plan that holds up

It's early October. A student writes in their planner: “starting tomorrow, I study an hour every day.” The exam is in January, there's plenty of time, the plan sounds reasonable. On day nine, illness hits — the flu, a family event, or just an evening that didn't work out — and the chain breaks. By day ten there's nothing left to continue. Then January arrives, along with a week of four hours of sleep a night, cramming material into your head in whatever order you happen to find it in your notes.

This story isn't about weak willpower. It's about a plan that was never actually a plan. “An hour every day” says nothing about how much material needs to be covered, in what order, when it should be reviewed, or how you'll know you're on track. It's a resolution about input, not a plan for output. A plan that holds up doesn't get written forward from today — it gets written backward from the deadline.

That applies to final school exams, entrance exams, oral finals, and to adults sitting a certification, a driving test, or submitting a tender proposal. The deadline is a fixed point you can't negotiate — everything else is a variable. Plan from today, and you're planning against an unknown target. Plan from the deadline, and within the first fifteen minutes you know whether your intention is even physically possible. That's information you want in October, not in January.

Working backward from the deadline: why it works better

Planning forward has one sneaky property: it never hits a wall. You can schedule an hour a day and feel content, because nothing in that plan tells you it isn't enough. Planning backward hits the wall immediately. If you write down that, the day before the exam, every topic has to be covered and one full review pass completed, and you start subtracting from there, the calendar either works out or it doesn't. If it doesn't, you have three options: start earlier, cut somewhere else, or lower your ambition. All three are legitimate. The worst option is the fourth — not knowing.

A backward plan has a second advantage too. It forces you to define what “being ready” actually means. That sounds like a philosophical question, but it's a very practical one: does it mean being able to list the topics? Solving a typical problem in twenty minutes? Answering the follow-up question the examiner likes to ask? Without that definition, you study endlessly and still feel it isn't enough, because you have no yardstick. With it, you know when to stop.

Milestones in the plan have to be externally checkable. “Study chapter 4” is a milestone you can check off even after skimming it. “Solve five problems from chapter 4 without looking at the solutions” is a milestone you either hit or you don't. The difference between these two phrasings decides whether your plan measures preparation, or just time spent sitting at a desk.

And one last thing: a buffer isn't hope, it's a line item in the calendar. In a backward plan, it gets written in as a block with a name (“catch up on whatever slipped”) and a date. A plan with no buffer is a plan that assumes you'll never get sick, never lose a day, and never discover that one topic is three times harder than you expected. The tip Final exams with AI: a year-long plan works through the concrete shape of a year-long schedule for a big deadline; the tip Deadline minus two days covers the version for two working days before a submission.

Breaking material into topics, and an estimate that actually holds

Before you can plan anything, it has to be clear what needs to be learned in total. That sounds trivial, and yet it's the most commonly skipped step. Most people start their prep by opening the first material they have on hand — and plan around wherever they currently are, not around what's still left.

An inventory should look like this: a list of topics from an official source (a syllabus, the official list of final-exam topics, the exam brief, a certification outline) — not from your own notes. Notes are always incomplete, because they capture what you managed to cover, not what will actually be tested. Then, next to each item on that list, write one of three labels: I know it (I can explain it right now), I know where it is (I understand it, but I need to review), no idea (I've never really covered it). This sorting is uncomfortable, and that's exactly why it's valuable — it usually reveals that the critical part of your prep sits in the last two categories, and that it concerns four or five topics, not thirty.

You estimate per topic, not per subject. “Learn biology” is an estimate that can never hold up, because there's nothing to derive it from. “Cover photosynthesis to the point where I can draw the diagram from memory and explain it out loud” is a task you can put two hours against — and above all, one where you'll find out you were wrong the same day.

Why we keep underestimating time

The tendency to underestimate how long something will take is one of the most consistent phenomena the psychology of planning describes — it's known as the planning fallacy, and research repeatedly finds it in individuals and in large projects alike. What's interesting is that it doesn't apply to estimates in general: people estimate other people's tasks fairly soberly, and they know from their own past that things ran long. The error mainly shows up in your own future plans, where you picture the ideal run and forget everything that delayed you in the past.

There are three practical consequences. First, estimate from reality, not from intention: how long did a similar topic actually take you last time? Second, measure. It's enough to log the estimate and the actual time for each block; after two weeks you'll see your personal coefficient — some people systematically run a third over, others half over — and you multiply future estimates by that coefficient. Third, budget for maintenance. A plan that only accounts for new topics ignores the time needed to review earlier ones, and it's exactly that hidden cost that wrecks most schedules in their final third.

But watch out for the opposite extreme too. Padded estimates “just to be safe” produce a plan you don't believe yourself, and a plan you don't trust gets abandoned exactly as fast as one you can't fulfill. The goal isn't an estimate you'll never exceed. The goal is an estimate you can correct with actual data after two weeks.

The error log: the most valuable document in your whole prep

If you could keep only one file from your entire prep, make it the error log. Not your notes, not your summaries — a list of the mistakes you made, and why.

Most people study by reviewing the material they already know. It feels good, it goes fast, and it creates a sense of progress. An error log reverses that mechanism: it aims your attention exactly where performance is weak, and it's the one place in your whole prep where improvement can actually be measured. It's kept simply — one line per mistake: what the question was, what you answered, what's actually correct, and above all why you got it wrong.

That last column is the whole point. A mistake isn't one kind of event — it's at least four different things, and each gets fixed completely differently.

The difference matters, because each category needs different treatment. Studying helps with not knowing something. Studying doesn't help carelessness at all — a routine does (read the prompt twice, underline the key word, check the units at the end), and anyone who adds another hour of review instead of a routine is solving the wrong problem. Mistakes from time pressure aren't about knowledge, they're about strategy: going through the whole test and picking off the easy parts first is a trainable skill. And a misread question is the most treacherous of all, because it looks like not knowing the material when it's actually a reading error.

After two or three weeks, the error log will tell you something you'd never have guessed on your own: usually one category turns out to dominate heavily. When three-quarters of the mistakes are carelessness, adding more material is pointless — the rest of your prep should go into checking habits. When not-knowing dominates and it's concentrated in two topics, you have a precisely defined target for next week.

The error log gets worked with actively: read it once a week, retest mistakes older than two weeks (and cross them off once they pass), and copy recurring mistakes onto flashcards for spaced repetition. For long-form written work, the same principle works at a different level — instead of test mistakes, you collect your advisor's comments and the recurring weak spots in your text; see A thesis with AI: an honest workflow.

Mock exams: they hurt, but they work

There's one training exercise almost everyone avoids, and yet it has the best ratio of time to payoff of anything in the whole prep: testing yourself under exam conditions, before the exam happens.

We avoid it for an understandable reason — it's unpleasant. Reading your notes creates a pleasant feeling of fluency: the text is familiar, it makes sense, your head nods along. But familiarity isn't knowledge. Recognizing the right answer is something different from generating it out of an empty head under time pressure, and the only way to find out what you actually know is to try it with nothing to lean on. That testing teaches more than re-reading is one of the best-documented findings in memory research; retrieval strengthens the memory trace, while reading merely refreshes it.

A mock exam should resemble the real thing as closely as possible: the same time limit, no materials, the phone somewhere else, pen and paper if you'll be writing by hand. Out loud, if it'll be an oral exam — and ideally recorded, because a recording reveals how much you actually can't say coherently, even when you have the right content in your head. Once it's over, the more important half comes: mark it against the answer key, sort the mistakes by cause, and log them in the error log.

Timing is key. The first mock exam should come early — at a point when you still don't know a lot of things and the result will be unpleasant. That's on purpose: the result is a diagnosis, not a grade. A mock exam three days before the real thing won't change anything anymore, because there's no time left to react. The tip Entrance exams with AI describes an ongoing version for entrance-exam prep.

There's a side effect nobody talks about much: repeated exposure to exam conditions lowers stress on the actual day. Most exam anxiety doesn't come from the material — it comes from an unfamiliar situation. Once you've been through that situation five times at home, the sixth time isn't a discovery anymore.

The last week: what to stop chasing

The last week runs on a different rule than all the ones before it. Up to that point, the rule is “learn more.” From this point on, the rule is “don't lose what I already know.”

Concretely, that means three bans. Don't introduce new material you haven't opened at all yet — there's no time left to master it, and its presence in your head will disrupt what's already solid. Don't introduce a new system, a new tool, a new flashcard app, or a new method; the last week is the worst possible time for experiments. And don't compare yourself to others, because in the last week, messages like “I've got everything done” are either untrue or irrelevant.

What does make sense in the last week: go through the whole error log, do one or two complete passes over the material at the outline level (not the detail level), review a handful of typical problems, and sort out logistics. Logistics get underrated: exactly where the exam is held, how long the trip takes, what you need to bring, when to get up. Every one of these questions left unresolved on the morning of the exam eats up a chunk of attention you wanted to use elsewhere.

And then there's the decision nobody likes to make: what to let go of. With five days left and three uncovered topics, it's more rational to get two of them to a decent level than all three to a poor one. The criterion is probability times cost — a topic that comes up almost every time takes priority over one the examiner mentioned once every three years. A deliberate drop is a strategy. An unplanned miss is an accident.

Stress and sleep aren't weakness, they're performance factors

The most widespread misconception about exam prep is that sleep is a luxury you allow yourself once you're done. But it's exactly during sleep that the memory traces from the day get consolidated and stored — so a night spent over your notes doesn't mean “more learned,” it means trading consolidation for a few extra pages read. On top of that comes worse attention, slower recall, and a reduced ability to handle an unexpected question. This isn't a moral argument, it's arithmetic: an hour of sleep before an exam usually has a higher return than an extra hour of reading.

Stress is similarly misunderstood. Mild tension actually improves performance — it raises alertness and focus, which is why people often recall things better in an exam than they expected. The problem starts when tension crosses the line into blocking working memory; that's when someone “knows that they know it” but can't get to it. That state can be influenced, mainly through the body: a slow exhale longer than the inhale, relaxing your shoulders, a few minutes of walking before you go into the room. You can't talk your head into it, but you can talk your body into it.

Long-term, the same things work here as for any other demanding performance — regular exercise, food that doesn't cause energy crashes, and breaks that are actually breaks, not a switch to a different screen. The chapters The psychology of productivity and Working in tune with your body go into this in more detail.

It's also worth mentioning what the plan itself does for stress. Most exam anxiety isn't about the material — it's about the uncertainty of whether you'll get there in time. A visible backward plan significantly reduces that uncertainty — a second, rarely mentioned benefit of having one.

After the exam: what to keep for next time

The moment an exam ends, it's natural to want to forget about it. But that's exactly when your head is holding information that will vanish irretrievably within a week — information that could save you dozens of hours next time.

Twenty minutes right after the exam is enough for four notes. What questions actually came up, and how they differed from what you prepared for. What was missing from your prep — was it, by any chance, exactly the part you kept postponing? How your estimate held up: how much you planned, how much you actually spent, how many times you had to replan. And what worked among your methods — what you'd repeat, what you'd never do again.

This note belongs in one permanent place, not in a notebook that disappears along with the subject. Over the course of long-term study, after a few exams it turns into a fairly accurate personal model: you know a topic actually takes you a third more time than you estimate, that your weak spot is careless mistakes in the last quarter of a test, and that a morning block with problems plus an evening error-log review works best for you. None of this can be derived theoretically — it only comes from recording your own experience, which the chapter Studying productively also covers.

And one more thing worth keeping: the error log. Don't throw it out after the exam. When you return to the same material a year later, for a follow-up exam, an oral final, or a job interview, it's the fastest way to get back into form — because your old mistakes are probably still your mistakes.

Key takeaways

• Plan backward from the deadline, not forward from today. A backward plan immediately tells you whether your intention is physically possible, and it includes a buffer as a line item, not a hope.
• Do an inventory of topics from an official source first, then study. Estimate per topic, measure the estimate against reality, and after two weeks calculate your own personal coefficient.
• The error log is the most valuable document in your prep. Sort mistakes by cause — not knowing, carelessness, time, misreading the question — because each one gets treated differently.
• Do mock exams early and under real conditions. An unpleasant result three weeks out is a diagnosis; a pleasant feeling from reading your notes is an illusion.
• In the last week, don't introduce new material or a new system. Deliberately drop low-probability topics and sort out your logistics.
• Sleep and managed stress aren't comfort, they're performance factors. And twenty minutes of reflection after an exam saves you dozens of hours before the next one.