Build from evidence, not instinct

Begin with the performance gap

A request for “training” is not yet a design problem. Ask what people currently do, what they should do, what evidence shows the difference and what conditions influence performance. A knowledge gap may call for instruction; a tool, incentive or workflow problem may not.

Interview representative learners and subject-matter experts. Review previous work, common errors and the environment in which the target behavior occurs. In a US higher education setting, also consider accessibility, time constraints, technology access and the varied experiences learners bring.

Write outcomes with consequences

A useful outcome names an observable performance, relevant conditions and a meaningful standard. “Understand assessment” is difficult to design for. “Evaluate a draft rubric against three stated quality criteria” offers a clearer target.

Design assessment before explanation

First decide what acceptable evidence looks like. Then build practice that prepares learners to produce it. This backward sequence reduces attractive but disconnected activities. Use low-stakes checks early, coached attempts in the middle and an authentic performance near the end.

Criteria should describe quality rather than merely count components. Offer examples at different levels, invite learners to apply the criteria, and discuss why judgments differ.

Make practice progressively authentic

Start with enough structure to prevent avoidable overload. Model a process, expose expert thinking, then remove supports as competence grows. Variation matters: if every example looks alike, learners may memorize surface cues rather than recognize the underlying principle.

Feedback should arrive while another attempt is possible. Focus it on the task, the strategy and the next decision—not on fixed judgments about the learner.

Revise the system, not just the slides

Compare outcomes, activity participation, learner work and recurring questions. A weak result may reveal an unclear criterion, insufficient practice, inaccessible material or an unrealistic outcome. Document the reason for each revision so later designers can understand the decision.

Evaluation need not become an enormous research project. One focused question—such as where learners misapply a concept—can generate a useful next iteration.