Retrieval Practice · Testing Effect
Retrieval Practice
What is retrieval practice?
- Pulling information out beats rereading
- Quizzes and recall strengthen memory
- Also called the "testing effect"
Adding retrieval practice means changing how you divide up the training schedule. In a 90-minute course, the instructor’s talking time drops to 75 minutes, and the remaining 15 minutes go to learners pulling answers from memory. You cover less material, but more of it is still there a month later. What designers need to change is not the amount of content but the ratio of input to output. The visible change is the point where the last page of the training materials switches from a “summary slide” to a “blank-page recall”.
Retrieval varies in intensity. Recognition (multiple choice), where options are given, is the weakest. Cued recall, where you get only a one-line keyword, is in the middle. Free recall, writing on a blank page with nothing in front of you, is the strongest. The principle is that the harder you work to retrieve something, the longer it lasts. If you only offer easy multiple-choice questions, retrieval is practice in form only, and the effect is little different from rereading. But questions that are too hard make learners give up, so in practice aim for a difficulty that puts the correct-answer rate at around 70 to 80%, and give the answers a few minutes later rather than immediately so the effort of recalling counts.
It is easy to confuse with spaced repetition (the spacing effect), but the two work at different levels. Spacing is a scheduling question: "when should learners encounter the material again?" Retrieval is an activity question: "what should learners do when they encounter it?" A design that sends five questions each at 1, 7, and 30 days after the course ends layers the two principles together. It should also be distinguished from formative assessment. Formative assessment is the instructor measuring learners in order to adjust the pace, while retrieval practice has learners recall in order to build memory, so collecting scores is not the goal.
The point where this breaks down most often in practice is the word 'test.' The moment quiz scores are recorded per person and passed to HR, learners focus on avoiding mistakes rather than recalling, and some sneak a look at the materials to fill in answers. It is safer to assume that retrieval practice dies the moment it is used for assessment. Another issue is satisfaction. Blank-page recall often draws complaints like 'unhelpful, no summary given,' and satisfaction scores commonly drop by 0.2 to 0.4 points. If you panic and go back to summary slides, the training people actually remember goes with them. Another frequent mistake is collecting only the wrong answers and never returning the explanations.
Jihye ChoiExpert in design thinking, new-hire onboarding, and train-the-trainer programs
John KimAuthor of B2B Marketing Design and CEO of Nanum ConsultingIt is backed by research on the "testing effect" in cognitive psychology and is widely known through experiments by Roediger and colleagues.
Consider a hypothetical case. Company A, an industrial parts distributor with 210 employees, ran a two-day new-product catalog training every quarter. But three weeks after each session, client inquiries were being answered with the wrong specs 14 times a month, and salespeople kept saying “I definitely heard that in training.” Instead of adding training days, the sales support lead decided to rebuild the format.
- Lee, a training designer, cut a two-day course from 180 slides to 120 and put the 60 minutes saved toward recall time. The first line of the proposal read “Teach less, make them retrieve more.”
- At the end of each session, participants closed their materials for five minutes and wrote down 'the three key specs of this product' on a blank sheet. Answers were checked after the break instead of right away.
- When Manager Kim, an eight-year sales veteran, protested, 'Did we come here to take a test?', the team lead made it clear that scores would not be recorded and only each person would see their own. In fact, not a single scoring sheet left the training room.
- Five questions were sent by mobile at 1, 7, and 30 days after the course ended. Instead of multiple choice, half were open-ended questions such as "How would you answer if a client asked about the pressure rating?" The response rate was 78%.
- Three months later, wrong-spec cases fell from 14 a month to 4, and day-30 recall accuracy rose from 52% in the first round to 81% in the third. Training satisfaction, however, dropped from 4.6 to 4.3, and the team lead reported that figure unchanged.