Most e-learning is information in a nicer wrapper. A tour of the policy, some tidy screens, a quiz at the end to prove you were awake. It teaches recall. The job needs judgement. Those are not the same thing, and the gap between them is where most training quietly fails.
The manual is only the surface
Your best people cannot easily tell you how they do the hard part. Ask them and you get the visible steps, the process anyone could read off a page. The valuable bit, the ‘I could tell this was about to go wrong’, sits underneath and is half-invisible even to them. There is a set of methods for digging it out, called cognitive task analysis, and it is the difference between teaching what the manual says and teaching how the expert actually thinks. It matters commercially too. When an expert retires, that hidden judgement walks out with them unless someone has captured it first.
Expertise is recognition, not a checklist
Watch someone good work under pressure and they do not weigh a tidy list of options. They recognise the situation as one they have met before, act, and run a quick mental simulation of whether it will hold. Real decisions get made fast, with missing information, while things move. If your training only tests whether people remember a rule, it is preparing them for a calm exam that never happens. Build for the recognition instead.
Reps, not packaging
Nobody got good at anything by being shown it once. Skill comes from trying, getting feedback, and trying again, enough times that the pattern sticks. Most e-learning skips this entirely and calls the end-of-module quiz ‘practice’. It is not. Real practice means repeated attempts at the actual judgement, with feedback that tells you what your choice caused. Reps matter, and the industry keeps quietly pretending they do not.
Which is why scenarios have to change
This is where the standard scenario falls down. Read a situation, pick A, B or C, one is right, the rest get a red cross and a ‘try again’. That is a knowledge check in a costume. A scenario earns its place when you change two things.
The first is to kill the single right answer. Offer options that all work, each carrying a different cost. Now the learner is not guessing what the designer wanted, they are weighing trade-offs, which is the actual job. The second is to let consequences compound. A choice that looked fine early quietly makes a later step harder, so the learner feels the chain of cause and effect rather than a verdict. Write the feedback as what happened next, not as the designer marking homework. And base the branches on how your real experts think, which is where the decision-mining from earlier earns its keep. A scenario is only ever as good as the judgement underneath it.
The course is not the finish line
The last mistake is treating the course as the place the work happens. It is not. The work happens at a desk, three weeks later, under pressure, in a workplace that either supports the trained behaviour or quietly punishes it. Design for that moment, the point of real use, and build the support that lives there, not just the module that comes before it.
None of this is new, and that is the point
Cognitive task analysis, recognition-primed decisions, practice with feedback, they have been sitting in the research for decades. The opportunity is that mainstream e-learning still ignores nearly all of it and keeps shipping slides with a quiz on the end. Doing the thing everyone has heard of and almost nobody does is not a niche flourish. It is the whole job, and it is how we build e-learning.