How to Revise Effectively: What Actually Works

If you want revision that actually sticks, do two things: test yourself instead of re-reading, and spread that testing out over days and weeks instead of cramming it into one sitting. Researchers call these habits retrieval practice and spaced practice, and they remain, by a wide margin, the two most reliably effective study methods we know of. Re-reading notes and running a highlighter down the page feel productive, but the evidence says they do almost nothing for long-term memory. This guide is updated for the 2026 exam year — what works, why it works, where AI revision tools now genuinely help, and how to build all of it into a routine your child can actually keep to.

Why re-reading and highlighting quietly fail

Open most students' revision and you will see the same thing: the textbook read again, key lines highlighted, notes copied out neatly. It feels reassuring because the material starts to look familiar. That familiarity is the trap.

Psychologists call it the fluency illusion. When you read a page for the third time, it flows easily, and your brain reads that ease as "I know this." But recognising information on a page is not the same as being able to produce it in an exam when the page is gone. According to a 2013 study review published in the journal Psychological Science in the Public Interest, led by the psychologist John Dunlosky, highlighting, underlining and simple re-reading were rated among the least effective techniques students commonly use — low-utility habits that eat time without building durable memory. The same review rated two other methods as high-utility for almost every learner, subject and age group. Those two are the backbone of everything below.

There is a related idea worth knowing: the generation effect. Information you generate yourself — an answer you produce, a summary you write from memory, a worked solution you attempt cold — is remembered better than the same information simply read. Re-reading and highlighting never ask a student to generate anything; that is the root of why they underperform. The practical takeaway is blunt: if a revision session involves your child mostly looking at material they have already seen, it is probably the wrong session. Good revision should feel harder than re-reading, because the effort is the point.

What works, part one: retrieval practice (test yourself)

Retrieval practice means closing the book and pulling the answer out of your own head — through a quiz, a blank page, flashcards, or past-paper questions done from memory. Every time you successfully retrieve a fact, you make it easier to retrieve next time. Every time you struggle and then check, you find the exact gap you need to fix.

This is not a hunch. In a well-known study by Henry Roediger and Jeffrey Karpicke, published in the journal Psychological Science, students who revised by testing themselves remembered substantially more a week later than students who spent the same time simply re-reading — even though the re-readers felt more confident at the time. The effect is so consistent that researchers gave it a name: the testing effect. Testing is not just a way to measure learning; it causes learning.

How to actually do it:

• Turn notes into questions. Instead of a page of facts, write a list of questions the facts answer. Revise from the questions, not the answers.
• Use the blank-page method. Read a topic once, close everything, and write down everything you can remember. Then open the book and fill in what you missed in a different colour. The gaps are your next session.
• Do past-paper questions from memory first, then mark them honestly against the mark scheme. Struggling before you check the answer makes the answer stick far better than reading it cold.
• Flashcards done properly — see the question, say the answer out loud before you flip. Flipping first turns a flashcard back into re-reading.
• Explain it out loud, as if teaching someone else. Cognitive scientist Michelene Chi's research on self-explanation found that students who have to justify their own answer, step by step, catch gaps that silent revision never surfaces. It is one of the cheapest, most-overlooked retrieval techniques.

The discomfort of not immediately knowing an answer is a feature, not a fault. Cognitive scientists Robert and Elizabeth Bjork call these "desirable difficulties" — conditions that slow learning down in the moment but make it far more durable. If revision feels effortful, it is usually working. This is also why a student's own sense of how well revision is going can be misleading: fluent, comfortable re-reading feels like progress, while effortful retrieval, which is actually building the memory, often feels like struggling.

What works, part two: spaced and interleaved practice

The second habit is about when and how you order revision, not just how you test. Spaced practice means revisiting a topic several times with gaps in between — a little today, again in a few days, again next week — rather than blocking all of it into one long cramming session.

The reason is the way memory fades. Forgetting begins almost as soon as you learn something, and a single session, however long, gives memory only one chance to form. Reviewing a topic just as you are starting to forget it forces your brain to rebuild the memory, and each rebuild makes it stronger and slower to fade. A large 2006 research review by Nicholas Cepeda and colleagues, published in the journal Psychological Bulletin, found that spreading study over time reliably beat massing it together for long-term retention across a wide range of material.

A closely related habit is interleaving: mixing different topics or question types within a single session instead of drilling one type until it feels mastered. Doug Rohrer and Kelli Taylor's research on how students practise maths problems found that mixing problem types forces a harder, more useful decision every question — which method applies here? — which is exactly the decision a real exam demands and blocked practice never trains. Interleaving feels worse at the time, because performance dips while you are doing it. It produces stronger results by the time the exam arrives, which is the only measurement that counts.

Cramming still has one honest use: it can get you through a test tomorrow morning. What it cannot do is put knowledge somewhere you can reach it at the end of a two-year GCSE or A-level course — which is exactly what exams test. For anything that matters months from now, spacing and interleaving win.

How to actually do it:

• Start early and revisit, don't finish and move on. Cover a topic, then schedule two or three short return visits over the following weeks rather than one marathon.
• Interleave subjects and question types. Instead of a whole evening on one topic, mix two or three, or mix past-paper questions from different topics within one sitting. It trains the skill exams actually require: choosing the right method for an unfamiliar question.
• Keep sessions short and frequent. Three 25-to-40-minute retrieval sessions across a week beat one three-hour block, and they are far easier to keep to.

The UK's Education Endowment Foundation, a schools evidence body that reviews what actually moves outcomes for the cost, rates metacognition and self-regulation strategies — the umbrella that retrieval and spaced practice sit under — among its highest-impact, lowest-cost recommendations for schools. That is a second, independent line of evidence behind the same two habits, this time drawn straight from UK classroom research rather than the lab.

Where AI revision tools fit in 2026 — and where they don't

By 2026, an AI chatbot or an AI study tool has become a normal part of most students' revision, sitting alongside notes and flashcards rather than replacing them. Used well, that is a genuine gain: an AI tool can generate an unlimited supply of practice questions on demand, which is exactly the raw material retrieval practice needs, and it can do it instantly on whatever topic a student is stuck on. Used badly, it becomes a new version of re-reading — a student pastes in a question, reads the AI's worked answer, and feels like they have revised without ever producing an answer of their own.

The rule that separates the two uses is the same rule that governs flashcards: retrieve first, check second. Ask the tool to quiz you, not to explain the topic to you from scratch. Our updated look at AI study tools and what actually helps covers the current landscape in more depth, and our guide to whether an AI tutor can prepare a child for GCSEs looks at where an AI tutor is a genuine substitute and where it still isn't. Tutorwise's own AI tutor, Sage, is built around question-and-check rather than answer-and-read for exactly this reason — see Sage's free GCSE maths help for what that looks like in practice. AI tools are a supplement to the two habits above, not a replacement for them, and that has not changed even as the tools themselves have got better.

Putting it together: a revision routine that holds

The methods combine into a simple loop. First pass: learn a topic and immediately test yourself on it. Second pass, a few days later: test yourself again from memory, before looking anything up. Later passes, spaced further apart and increasingly mixed with other topics: keep testing the ones you find hardest, and let the easy ones fade further into the schedule.

A workable weekly shape for a student in an exam year:

1. Monday — cover a new topic, then do a blank-page recall of it.
2. Wednesday — a short quiz on Monday's topic, plus a new topic covered and recalled.
3. Friday — past-paper questions mixing both topics, marked against the mark scheme.
4. Next week — the hardest questions from the past two weeks come back around, interleaved with newer material.

Keep a single sheet listing every topic and how confident you feel about each, updated after each session. It turns a vague "I should revise" into a concrete "these four topics are red, start there." That list is the timetable — and if building one from scratch feels daunting, our guide on building a revision timetable that works walks through the process step by step.

Sleep matters more than a late night of cramming: memory is consolidated during sleep, so a rested exam morning usually beats one more hour of tired re-reading. And if revision itself is triggering serious stress rather than ordinary exam nerves, treat that as its own problem to solve first — our guide to GCSE exam anxiety is written for exactly that.

When a tutor is worth it

Most of this a motivated student can run alone. The point at which a tutor earns their place is narrower and worth naming: when a student cannot tell which topics they are genuinely weak on, when they have plateaued despite putting the hours in, or when exam technique — not knowledge — is losing the marks. A good tutor builds retrieval and spacing into every session by default, sets the right questions at the right difficulty, and marks honestly so the feedback is real.

If you do look for one, the hard part is knowing whether their credibility is real. On Tutorwise a tutor's standing is a computed score built from verifiable signals — a checked DBS and identity, confirmed qualifications, delivered outcomes and reviews — rather than a self-written bio. Picture two tutors who both claim years of GCSE experience: one's score is built mostly on verified subject outcomes reported by past students and a long-standing review history, the other's on strong credentials but a shorter track record on the platform. Comparing those two is comparing evidence you can inspect, not adjectives either of them chose about themselves — and you can see which trade-off matters more for your child before you ever book a session. Our explainer on what a tutoring credibility score actually measures breaks down exactly which signals feed the score and why each one is there.

The headline holds whether you revise alone, with an AI tool, or with a tutor: test yourself, space and mix it out, and treat effort as the signal that it is working.

Frequently asked questions

Is highlighting really useless? Not useless, but close to it on its own. Highlighting can help you find key points later, but the act of highlighting builds almost no lasting memory — the 2013 Dunlosky study review rated it low-utility. If you highlight, treat it as a first pass only, then revise from questions and recall, not from the highlighted page.

How far in advance should revision start? As early as is realistic, because the benefit of spaced practice comes from the gaps between sessions, and you cannot create gaps at the last minute. Starting weeks or months out lets each topic be revisited several times and mixed with others. A late start forces cramming, which helps tomorrow's test but fades before the real exams.

Isn't testing yourself just stressful? Low-stakes self-testing at home is very different from a real exam, and it is one of the best ways to reduce exam stress. Familiarity with retrieving answers under mild pressure makes the real thing feel less daunting. The discomfort of not knowing an answer straight away is the learning happening, not a sign of failure.

Is it safe to use ChatGPT or an AI tutor for revision? Yes, if it is used to generate practice questions and check answers you attempted yourself first — the same retrieve-then-check rule as flashcards. It stops helping the moment a student reads a generated answer instead of producing their own. Our AI study tools guide covers the specifics of what to look for.

Does this work for every subject? The two core methods apply everywhere, because they train memory and recall themselves. Retrieval practice suits fact-heavy subjects like science and history especially well; for maths and problem-solving subjects, the "test" is working problems from scratch, ideally interleaved across topics, and spacing that practice out. The principle is the same: produce the answer from memory, in a mixed order, and spread the practice over time.