A learner repeats a software procedure exactly as shown. The practice screen looks perfect, but when a new file behaves differently, they do not know how to recover. They learned the sequence; they have not yet practiced what to do when the sequence breaks.
Error management training is a learning approach that allows people to make mistakes they can safely recover from, then helps them notice what happened and correct it. The emphasis is on experimenting within a protected practice setting, not on making mistakes for their own sake.

Make the error useful, not costly
A practice error teaches something when the learner can see what happened, connect it to a decision, and attempt a recovery. A simple learning loop might look like this:
- Set a boundary. Use a training file, scrap material, rehearsal, or simulation where a mistake will not harm a client, a colleague, or an important result.
- Try before receiving the full answer. Let the learner notice the mismatch and test a reasonable next step.
- Name the cue and the correction. What showed that the approach was not working? What did you change?
- Try a changed version. A new example checks whether the learner understood the principle rather than memorizing one recovery.
For example, someone practicing a spreadsheet workflow can work on a copy, try to repair an unexpected formatting error, and compare the fix with a model answer afterward. The point is not to create needless difficulty. It is to build a little experience with noticing, recovering, and carrying the lesson into a new case.
Error-friendly practice is a poor fit when a mistake could cause injury, expose private information, create an irreversible financial action, or damage a live system. In those settings, use safeguards, simulation, checklists, and supervised instruction. The learning goal is competent performance, not error for its own sake.
This method works best alongside deliberate practice and specific feedback. Our guides to focused skill building and useful feedback cover those complementary parts.
What the evidence supports
Keith and Frese’s meta-analysis found a positive average effect across the studies they reviewed, with stronger transfer in structurally different tasks. Read the 2008 Journal of Applied Psychology paper for the full analysis and its moderators.
Selected for this topic
Useful next reads
Four editorially reviewed guides chosen for relevance, not at random.




