DoRevision
In-depth guide

Running retrieval practice in a GCSE lesson

Retrieval practice tends to go wrong in fairly predictable ways, and almost all of them are about the task rather than the idea. The commonest is that students are recognising rather than recalling, because the answer is somewhere on the page. The next commonest is that it has quietly turned into an assessment, so students stop attempting and start protecting themselves. This page is the practical half: what a twelve minute run of it looks like from the bell, the seven failure modes and the change each one needs, and how to keep it going for a whole term without making work for yourself. The evidence sits on its own page and isn't repeated here.

There's a page on this site about what retrieval practice is and what the research does and doesn't support. This isn't that page. This one assumes you're already doing it, or about to, and deals with the part that decides whether it works: the task itself.

The three studies it rests on, Roediger and Karpicke (2006), Karpicke and Blunt (2011), and Dunlosky and colleagues (2013), are set out with their limitations on the retrieval practice page. We haven't added further studies here, because we'd rather keep a short list than cite something we haven't checked ourselves.

Twelve minutes, from the bell

A worked example rather than a principle. Adjust the clock to your own lesson.

Minute 0. The questions are already on the board, or the link is already shared. Nothing about the topic gets said out loud, since a reminder is a free answer.

Minutes 1 to 8. Students answer, books closed. You don't circulate helping. You circulate reading, because eight minutes of watching what people get stuck on is the best diagnostic in the lesson.

Minutes 8 to 10. Answers go up. Students mark their own and write down the ones they got wrong rather than a total.

Minutes 10 to 12. You reteach one thing: the one the most people missed.

Those last two minutes are the ones that get dropped when time runs short, and they're the ones that turn a score into a change.

The seven ways it stops working

1. It's recognition, not retrieval. The answer is visible somewhere, on a display, in an open book or in the question itself. Students feel fluent and remember nothing. Change: close the books, cover the display, and check your own questions for the answer.

2. It became an assessment. Once a score is recorded, a student who's unsure stops attempting, and the attempt is the part that works. Change: stop recording it. If you need a grade, assess separately and be clear which is which.

3. The feedback arrived too late. Answers given the following lesson land on a student who's forgotten what they were thinking. Change: answers in the same lesson, ideally in the same ten minutes.

4. You only ever retrieve last lesson. It becomes a check on whether they were listening rather than on whether they still know it. Change: make half your questions about content from three weeks ago or more.

5. Everyone retrieves what they already know. Given a choice, a class tends to pick the comfortable topic, and the practice reinforces the strong bits. Change: you choose the topic, and choose it from what went wrong last time.

6. It replaced teaching. Retrieval practice strengthens what's already there. It doesn't build something that was never taught properly in the first place. Change: if the class can't retrieve it because they never really had it, that's a reteaching problem rather than a retrieval one.

7. It stopped by November. This is the most common one by some distance. A routine that needs preparing every time gets dropped the first week you're drowning. Change: get the preparation cost close to zero. A pre-made activity you share as a link survives a bad week in a way a hand-written quiz doesn't.

When it is not working, start here

What you are seeing Likely cause What to change
High scores, poor exam performance Recognition, not recall Close the books, remove visible prompts
Students not attempting It is being recorded Stop recording it
Same errors repeating for weeks Feedback too late, or no next step Answers in the same lesson, one thing retaught
Strong on new work, weak on old Only ever retrieving recent content Half the questions from three weeks back
Confident class, disappointing mocks Retrieving only comfortable topics You pick the topic, from last time's errors
Nobody can recall anything It was never securely taught Reteach, then retrieve
It has quietly stopped Preparation cost too high Use something pre-made and shareable

Running it all term without marking

The workload question decides whether this survives past half term, so it's worth being concrete.

Self-marking in the room. Students mark their own against answers on the board. You read the room rather than the books. This costs you nothing, and it's the version most teachers end up running.

A shared link that checks itself. The activities below are mapped to a specification, and each answer is checked as the student gives it, so the feedback arrives when it's most useful. Nothing comes back to you, and there's nothing to set up beyond sharing the link.

A record the student keeps. Since nothing is reported to you, the record lives with the student. The printable revision record is one week to a page, and reading four of them at the start of a lesson tells you more than a set of marked books.

What still needs you: choosing the topic, deciding what to reteach, and noticing the student whose sheet is empty. Those are the parts worth your time, and nothing automates them.

What this page is not claiming

It doesn't claim a size of effect. No percentages, and no "students who do this improve by", because the honest version of that number depends on which study you pick.

It doesn't claim these seven failure modes are exhaustive or drawn from research. They're the ones that come up, described plainly so you can recognise them.

It doesn't claim retrieval practice is enough on its own. It strengthens what teaching has already built.

Where to go next

For the evidence and its limits, read retrieval practice for GCSE. For specific tasks you can run tomorrow, on screen and off, see ready-to-run revision activities. To put a run of these in front of one class, build a sprint, or browse by subject on the revision pages.

Try it with your class

AQA Biology 7 minutes

Balance Basics

11 steps, 7 interactive, 7 activity types.

Best used for
Starter Independent study Mock preparation
  • Starter: Short enough to open a lesson and still leave time to teach.
  • Independent study: Explains before it asks, so a student can work alone.
  • Mock preparation: Asks students to build an answer, not just recognise one.

Try one question

Why must the body keep its internal conditions constant?

AQA Computer Science 8 minutes

Base Camp

13 steps, 8 interactive, 8 activity types.

Best used for
Starter Independent study
  • Starter: Short enough to open a lesson and still leave time to teach.
  • Independent study: Explains before it asks, so a student can work alone.

Try one question

Which digits can appear in a binary number?

Edexcel Physics 9 minutes

Physics Foundations

8 steps, 5 interactive, 6 activity types.

Best used for
Starter Tutor time Intervention
  • Starter: Short enough to open a lesson and still leave time to teach.
  • Tutor time: Fits a short tutor slot without being cut in half.
  • Intervention: Mostly questions, so it finds the gap quickly.

Try one question

On the Moon your mass is unchanged but your weight is less. Which statement is correct?

AQA Chemistry 16 minutes

Mixture or Compound?

13 steps, 9 interactive, 9 activity types.

Best used for
Homework Independent study Intervention
  • Homework: Set it as a link. It marks itself.
  • Independent study: Explains before it asks, so a student can work alone.
  • Intervention: Mostly questions, so it finds the gap quickly.

Try one question

Water is made of hydrogen and oxygen chemically bonded together. What type of substance is it?

Common questions

How often should I run retrieval practice?

Often enough that it's a routine rather than an event, which usually means most lessons with a class you see three times a fortnight. What matters more than the frequency is that it keeps reaching back to older content instead of only checking last lesson.

Should I record the scores?

Recording changes what students do. Once a score counts, a student who's unsure stops attempting and starts protecting themselves, and the attempt is the part that does the work. Keep it unrecorded, and use a separate assessment when you need a grade.

Does it work for lower-attaining students?

It isn't a technique that only suits confident students, but the task has to be within reach. If a student can recall nothing at all, the question is too far from what they know, and the fix is a narrower topic rather than an easier question type.

Is this the same as testing students more?

No, and it stops working if it turns into that. The point is the act of recall rather than the measurement. A question nobody marks does the same memory work as one that counts.

What if it makes students anxious?

Anxiety usually comes from the stakes rather than from the recall. Nothing recorded, mistakes treated as information, and a short predictable format are what keep the pressure low. If a class is still anxious, it's worth asking what they think happens to the results.

Where is the evidence for all this?

On our retrieval practice page, with the three studies cited in full and an explicit account of what they don't show. This page is about running it.

Put one in front of a class

Tell us the subject and how long you have. You get a link to hand out, and it marks itself.

Create a free teacher account

Written by David Wright, Co-Founder, DoRevision.

David builds DoRevision with his son Noah (15). David has spent his career building EdTech products, and Noah has grown up watching lines of code being written. As a student at an important stage of his education, together they are building the revision platform that Noah is using to help revise for his GCSE exams and hopes his friends will find useful too.

Spotted an error on this page? [email protected]

How DoRevision writes its teacher material How DoRevision uses evidence and cites sources

Related for teachers

  • Ready-to-run GCSE revision activities A working list of GCSE revision activities, each with the thing it makes students recall, the moment they find out whether they were right, and what they do about it. Some run on a screen and some need nothing but a board.
  • Retrieval practice for GCSE, without the marking What retrieval practice is, what the research does and doesn't show about it, and how to run it in a GCSE class without making work for yourself. Every claim on this page names its source.
  • The GCSE revision homework toolkit What to actually set as revision homework at fifteen and thirty minutes, a weekly routine that lasts a full term, options for a class with a wide range in it, what to tell parents, and a printable sheet students keep themselves.