Insights · Learning science
The science of memory: why microlearning beats marathon courses
A two-day workshop feels substantial. That is exactly the problem. Memory is built by repeated effortful recall spread over time, not by a single dense exposure, so the format that feels most serious is often the one that transfers least.
What actually happens after training
Ebbinghaus described the forgetting curve in the 1880s and the shape has held up ever since: without reinforcement, most of what is learned in a single session decays within days. The curve is not a flaw in your people. It is how memory prunes information the brain has no evidence of needing again. The design question is therefore not "how much can we cover" but "how do we give the brain evidence this matters".
Three effects that do the heavy lifting
1. The spacing effect
The same total study time produces far stronger retention when it is distributed across several sessions instead of massed into one. Practically: four 15-minute modules across four weeks beats one 60-minute module, at identical cost. Each return visit re-signals importance and flattens the decay curve.
2. Retrieval practice, the testing effect
Pulling information out of memory strengthens it far more than putting it in again. A learner who answers a scenario question remembers more than one who rereads the same content twice, even though rereading feels easier and more productive. That mismatch, fluency mistaken for learning, is why learners rate passive courses highly and then fail to apply them.
3. Desirable difficulty
Learning that feels slightly hard, interleaved topics, varied contexts, delayed feedback, produces more durable and more transferable skill. Smooth, linear, perfectly scaffolded content produces confident learners who cannot perform under real conditions.
What this means for programme design
- Cap modules at 5–10 minutes with one objective each, so they can be scheduled and repeated.
- Open with a question, not an agenda. A failed attempt before instruction improves later retention.
- Space reinforcement at roughly day 1, day 3, day 10 and day 30 after the initial exposure.
- Use scenarios, not recall quizzes. Ask the learner to decide, then explain the consequence.
- Interleave related topics rather than teaching each to completion in isolation.
- Measure delayed performance, not end-of-module scores. A test 30 days later tells you something real.
The objection: "our people do not have time"
Microlearning does not add time, it redistributes it. The organisation already pays for the two-day workshop and then pays again when the behaviour does not change. Moving the same hours into short spaced sessions costs nothing extra and removes the need to pull whole teams offline.
Where AI changes the economics
Spaced, scenario-based content used to be expensive because it means producing many small assets and refreshing them constantly. AI-assisted production collapses that cost. The design judgment, what to teach, how to sequence it, what a good scenario looks like, stays human. The drafting, variation and localisation do not.
A simple redesign test
Take your most important existing course. Ask three questions. Could a learner meet each objective in under ten minutes? Is there a point where they have to retrieve rather than receive? Does anything reinforce it after week one? If the answer is no to any of them, you have a content library, not a learning system.