Simulation training

Almost every definition of simulation-based training leans on the word realistic, and that is where the trouble starts. Realism is expensive, and the realism organisations buy is usually the visible kind: the mock trading floor, the headset, the mannequin that blinks. None of it does anything if the participant is not genuinely on the hook for a decision.

Why the simple definition misleads

People who build simulations separate three kinds of fidelity. Physical fidelity is whether it looks like the real setting. Functional fidelity is whether it behaves like the real thing when you act on it. Psychological fidelity is whether it produces the same pressure, uncertainty and stakes. Transfer comes from the second and third. A cardboard cockpit whose instruments respond correctly, run for a crew who believe they are being judged, teaches more than a photoreal one that cannot be broken.

The second problem is the word training, which implies the simulation is the training. It is not. It is the raw material. The run produces evidence of how these particular people behave when it gets hard, and debriefing turns that evidence into something usable on Monday.

A simulation you do not debrief is not training. It is an anecdote with a budget.

What it looks like in practice

The discipline grew up in aviation, where crews fly full-mission scenarios and the interesting failures turn out not to be stick-and-rudder. Crew resource management training came out of accident investigations in the 1970s in which the aircraft was flyable and the coordination was not.

Healthcare took the idea into real corridors. An in-situ simulation drops a deteriorating patient into a working ward with the team on shift, and finds what no classroom would: the trolley missing an item, the registrar who noticed and hedged, the two people who each assumed the other had called.

The corporate version needs no equipment at all. A runnable one: teams of five, a written incident brief, and a facilitator playing the rest of the world. Every eight minutes an inject arrives, a customer email, a regulator's call, a colleague's rumour. Each team submits one written decision per inject and cannot take it back. Forty-five minutes of run, forty-five of debrief. It surfaces who escalates and who absorbs, who asks the outside world questions and who assumes, and how long a group defends its first theory. The debrief question that does the work: at what point did you stop asking questions, and what made you stop?

The versions we build for clients run in a browser on participants' own phones, with a facilitator driving the pace from one screen. HEARTWARE is five rounds and ninety minutes on judgment in the AI era: teams decide what to trust, verify and own, and interrogate the characters in free text rather than picking from a menu, which is where the revealing behaviour shows up. A spy-thriller induction we built for two hundred new joiners runs in two halves at once, ciphers inside the app and physical operations off-screen. How we build them covers the mechanics.

Simulation, role play, case study, drill

Role play is what people usually mean when they say simulation. The difference is consequence: a role play ends when the conversation does, while a simulation carries your decision into the next twenty minutes and shows you what it cost.

How to run one well

• Write the decisions first. Name three or four choices that are genuinely hard in the real job, then build the scenario backwards from them. Scenario-first design produces atmosphere and no learning.
• Buy fidelity only where the decision lives. If the skill is choosing with partial information against a clock, spend on the information and the clock, not on the set.
• Let it go wrong. A simulation nobody can fail teaches nothing. The consequence has to land inside the session, while the decision that caused it is still fresh enough to argue about.
• Make the world reactive but not helpful. The facilitator plays the environment and answers only what is asked. Much of the learning is in what nobody thought to ask.
• Protect the person, not the performance. Nobody risks being visibly wrong in front of colleagues unless the room is safe, which makes psychological safety a design input rather than a happy outcome.
• Do not score it for promotion. Once a simulation counts towards someone's record they play safe, and playing safe is the one behaviour you learn nothing from.
• Give the debrief as long as the run. Debriefing an activity is where an experience becomes a transferable rule, and it is the first thing cut when the agenda slips.
• Run it more than once. Deliberate practice, as Anders Ericsson described it, needs a task at the edge of ability, immediate feedback and repetition. Most corporate simulations skip the third.

How to tell whether it is working

1. Did anyone visibly change their mind mid-run? A group that holds its first theory to the end was not under enough pressure, or did not believe the consequences.
2. Can people name the cue rather than the lesson? "We should communicate better" is a slogan. "We read the silence as agreement" is a finding, and it is portable.
3. Did it surface something you did not already know? If the simulation only confirmed the brief, it was scripted too tightly to tell you anything.
4. Does the behaviour appear in the real thing? The Kirkpatrick levels run reaction, learning, behaviour, results. Most programmes measure reaction and report it as behaviour.

Be honest about what you can claim. End-of-session ratings mostly track how much people enjoyed the facilitator, and confidence often rises fastest in those who learned least, because a simulation hands out vocabulary as well as skill. Triangulate: what the facilitator observed, what the manager notices six weeks later, one operational signal agreed beforehand.

Myths, and when not to simulate

• "More realistic is always better." Past the point where the decision feels real, extra realism buys attention rather than learning, and makes the thing expensive to change when the business changes.
• "Simulation means VR." Some of the most effective simulations ever run are a printed brief, a clock, and one person playing the rest of the world convincingly.
• "People remember ninety per cent of what they do." Those percentages were invented and attached retrospectively to Edgar Dale's Cone of Experience, which carried no numbers. Practice does beat listening for skills. Quoting invented figures makes a sound argument look unserious.
• "The scenario has a right answer." If it does, you have built a quiz with scenery. Real situations offer several defensible answers with different costs, and the debrief is about the cost.
• "It went badly, so it failed." A simulation that falls apart has usually done its job. The failure is free, which is the point of buying it here rather than discovering it live.

Three times not to reach for one. When the objective is knowledge transfer, a document is faster and cheaper. When nobody competent will run the debrief, you are buying an experience and hoping. And when the real problem is a broken process rather than a missing skill, simulating it walks people through a failure that was never theirs. That last case matters most: a badly designed simulation is worse than no training, because people generalise hard from vivid experiences, and they leave with a war story defending the wrong lesson.

Tour De Force runs this as live, experiential training for teams worldwide — online and in person. Talk to us, or play Gamified learning appsThe Weekly Challenge to see the method in ten minutes.