calendar
Edited on October 29, 2026

Simulation Training: What It Is, Types, Examples and How to Build One

Simulation training explained: the types, what the evidence says, how real it needs to be, and how to build and debrief a workplace simulation.

Alex Hey
Alex Hey
Chief Operating Officer
Key takeaways

Simulation training lets people practise a model of a real task in a controlled setting, with feedback on what happens. It can rehearse procedures, systems, conversations or decisions while reducing exposure to the consequences of real work.

  • In a review of 24 health-profession studies, most comparisons found no statistically significant advantage for higher-fidelity equipment. Match the task requirements rather than assuming a more lifelike appearance improves learning.
  • A meta-analysis of 46 samples estimated an average performance benefit of about 25% from debriefs over controls. It is a pooled estimate, not the expected gain from every drill.
  • Choose the type of simulation by the decision people need to practise, not by how realistic the equipment looks.
Instructional Designers

Go from storyboard to an interactive course in minutes

A new support hire's first live billing outage is the worst place to learn the incident runbook. Card payments start failing on a Friday afternoon. Three customers write in within 10 minutes, the status page still shows green, and the new hire has to decide whether this is an incident or three unlucky cards.

A useful simulation makes the learner act on relevant information and gives feedback on the consequences. For the payment-outage example, the next situation should reflect the earlier choice, including any time already lost.

I'm the COO at Coursebox and I manage most of our Perth team. Hiring for a small team teaches you to look at what a person does, so I often start with a low-tech version of the real task. My budget question is what the practice needs: the relevant information, actions, feedback and time to review. Some tasks require accurate physical equipment; an outage decision drill may not.

This guide covers the types of simulation, examples by industry, what the evidence shows and how real a simulation needs to be. Then it builds one in six steps, using a hypothetical payment-outage drill as the worked example. At Coursebox we build branching scenario software for workplace training, and near the end I report what happened when I asked our course agent, the AI that plans and builds a course from a prompt, to build that drill.

What is simulation training?

Simulation training lets people practise a model of a real task in a controlled setting, with feedback on what happens. It can rehearse procedures, systems, conversations or decisions while reducing exposure to the consequences of real work.

Simulation researcher David Gaba describes simulation as a technique rather than a technology: guided experiences that replicate substantial parts of the real world in an interactive way. It need not mean high-tech equipment, although some tasks require accurate tools or environments. I plan the practice and the feedback together.

Two close relatives need separating. Role play is one kind of simulation, where the task is a conversation, and our role-play simulation guide covers it. Scenario-based learning builds a whole course around a realistic problem and may or may not include a simulation; the scenario-based learning guide compares the two.

Types of simulation training

One useful grouping is by what is being simulated. The formats overlap: a VR setting can contain equipment practice, conversations or decisions. Choose the format that reproduces the parts of the task the learner must practise.

TypeWhat it simulatesMain cost driverUse it when
Equipment or physicalA machine, vehicle or body, such as a flight simulator, patient manikin or forklift trainerBuying and maintaining hardwareThe skill is in the hands, and a mistake on the real thing is dangerous or expensive
Software or systemA practice copy of a work tool, such as the CRM or billing systemBuilding and resetting the practice environmentPeople must learn a workflow without touching real records
Decision or crisis drillAn incident unfolding over time, with timed updates read out or releasedWriting the scenario and the facilitator's timeThe hard part is judging what is happening and who to call
ConversationA customer, patient or colleague, played by a person or an AIA trained partner's timeThe skill is what you say and how you listen
Business gameA whole company or one department, run over several roundsBuilding the model behind the gameManagers need to see how decisions trade off over time
VR or immersiveA full 3D setting viewed through a headsetHeadsets, content production and updatesThe real setting is hard to reach and spatial awareness is part of the skill

Equipment and VR simulations cost mainly hardware. Decision drills and conversations cost mainly people's time. Conversation simulations have their own guides: the role-play examples and these customer service role-play scenarios. Software simulations have a long history in computer-based training. Business games come in two kinds: functional games that simulate one department, such as marketing, and total-enterprise games that run the whole company.

Simulation training examples by industry

Each example below names what is simulated and the decision people practise in it.

Health care

  • What is simulated: a deteriorating patient, played by a manikin or by an actor.
  • The decision practised: spotting the change early and calling for help before the patient crashes.

Aviation

  • What is simulated: a full-flight simulator reproducing one aircraft type's cockpit and handling.
  • The decision practised: handling an engine failure or a rejected take-off, events too dangerous to rehearse in a real aircraft.

In the US, 14 CFR Part 60 governs the qualification and use of flight simulation training devices when they are used to meet flight-crew certification or qualification requirements. That is a specific regulatory use, rather than a requirement for every informal cockpit practice activity.

Emergency services and safety

  • What is simulated: a multi-vehicle crash staged on a training ground, or a tabletop exercise where radio updates arrive on a timer.
  • The decision practised: who takes command, which casualties come first and when to call for more crews.

Manufacturing and equipment

  • What is simulated: a forklift or crane simulator, or a practice line with a fault planted in it.
  • The decision practised: recognising the fault and stopping the line, or operating safely before touching a real machine.

Corporate and office work

  • What is simulated: a practice copy of the billing tool, a crisis drill or a business game for new managers.
  • The decision practised: processing a refund without touching live accounts, escalating an incident, or trading price against volume.

Benefits of simulation training, and what the evidence actually shows

The benefits come from what a simulation lets you do that real work does not:

  • Practise the rare event: people rehearse failures that happen once a year before they happen for real.
  • Repeat until it holds: a drill can be re-run the same afternoon, which a real incident cannot.
  • Measure performance: an observer scores what people did, not what they say they would do.
  • Reduce exposure to real-work harm: practise choices in a controlled environment, with appropriate supervision and feedback. The session still has costs and needs suitable safeguards.

The strongest evidence comes from health care. A JAMA meta-analysis led by David Cook pooled 609 studies of technology-enhanced simulation for doctors, nurses and other health professionals, enrolling 35,226 trainees.

609 studies
of simulation training for health professionals, with large effects on knowledge, skills and behaviours and moderate effects on patient outcomes, compared with no training
Source: Cook et al. (2011), JAMA, systematic review and meta-analysis of 35,226 trainees

The comparison in that review is with no training at all. It shows simulation beats doing nothing. It does not show that simulation beats a good workshop or on-the-job coaching.

Two more limits apply. Most of this evidence comes from health care, so treat it as a strong signal for other workplaces rather than proof. And patient outcomes are harder to move than test scores: a 2010 review by Lateef noted that no study had yet shown simulation improves patient care outcomes directly. Cook's review a year later found moderate effects across 32 studies that measured direct effects on patients.

An instructor in a navy cardigan talks with two students in blue scrubs beside a patient manikin seated in a wheelchair in a simulation lab
Sahil Singh / Pexels

Case study: US nursing schools that swapped clinical hours for simulation

Nursing students in the US practise on real wards before they are licensed. The US National Council of State Boards of Nursing tested how much of that clinical time simulation could replace.

  • The intervention: in the NCSBN National Simulation Study, new students at 10 nursing programs were randomised in August 2011 to clinical as usual, 25% simulation or 50% simulation in place of clinical hours, across all their core clinical courses.
  • The evidence: 666 students completed the study by graduation in May 2013. They were assessed on clinical competency and nursing knowledge, and tracked through the licensing exam and their first 6 months as registered nurses.
  • The result: there were no statistically significant differences between the three groups in clinical competency, knowledge or exam pass rates. Managers' ratings of competency and readiness for practice also showed no differences at 6 weeks, 3 months and 6 months.
  • Limitation: NCSBN limits its recommendation to conditions comparable with those in the study, including the way simulation was designed and delivered. The reported lack of statistically significant differences does not, by itself, prove the outcomes were identical or establish a replacement percentage for other professions.
  • Lesson: Under the studied programme conditions, substantial simulation replacement was feasible without a statistically detected disadvantage on the reported measures. Any substitution in another programme needs evidence and standards appropriate to that setting.

How real does a simulation need to be?

High-fidelity equipment can be expensive. Geoff Norman and colleagues reviewed 24 studies comparing high-fidelity simulators with lower-fidelity versions of the same health-profession training. The review notes that some higher-fidelity equipment can approach six-figure costs.

1% to 2%
average difference in performance between high-fidelity and low-fidelity simulators; nearly all the studies found no significant advantage for the high-fidelity option
Source: Norman, Dore and Grierson (2012), Medical Education, review of 24 studies

Both higher- and lower-fidelity versions improved performance against no-intervention controls in that review. The small average difference does not establish that physical accuracy never helps; check the functional and physical requirements of the task you are teaching.

Stanley Hamstra and colleagues went further and proposed dropping the word "fidelity". They split it into physical resemblance, meaning how much the simulator looks and feels like the real thing, and functional task alignment, meaning how closely it makes the learner do the thinking and actions the job requires. Their reason: several studies found that the degree of fidelity appears to be independent of how well people learn.

From the COO's chair, I budget for equipment or software, scenario design and feedback or facilitation. The payment drill needs no new dedicated simulator if the team already has the relevant materials. Staff time still has a cost, and a paper drill is not a substitute when operating the real interface is itself the skill.

tip:Match the task requirements

Before buying equipment, identify the decisions, physical actions and cues people need to practise. Check whether a simpler version preserves those functions and supports access. Physical resemblance may help when spatial, sensory or equipment-handling details are part of the skill. Meet any applicable simulator standards and test the chosen design.

The debrief is where people learn

Scott Tannenbaum and Christopher Cerasoli combined 46 samples of debriefing research from medical and non-medical settings, including the after-action reviews the US Army has used for decades, to measure what a debrief adds.

About 25%
average improvement in performance from debriefs over a control group, with similar effects for teams and individuals and in simulated and real settings
Source: Tannenbaum and Cerasoli (2013), Human Factors, meta-analysis of 46 samples (N = 2,136)

A debrief is a relatively inexpensive way to examine an attempt, but the pooled estimate is not a promise for this drill. In the example, the facilitator uses the changing clock and queue to ask what the learner noticed, why they acted and what they would change. The three prompts each have a job:

  1. What happened: "When did you know it was an incident, and what told you?"
  2. Why it went that way: "What did customers know at 4:00 pm?"
  3. What changes next time: "What will you do at report number three next time?"

For a full practice session with a partner and an observer, the role-play guide sets out the method.

How to build a simulation training session in 6 steps

I keep measurement as its own step, because it is the one that gets skipped.

Step 1: Pick one decision the job asks for

Choose a decision, not a topic. "Handling outages" is a topic; "deciding whether three failed payments are an incident" is a decision. If you are not sure which decisions are going wrong, start with a training needs analysis questionnaire.

Step 2: Write the trigger and the facts

The trigger is the moment the simulation starts. Supply the information the person would have, such as a runbook and contact details, or a realistic way to obtain it. If seeking missing information is part of the skill, build that into the exercise rather than accidentally leaving out the inputs needed to act.

Step 3: Add consequences that carry forward

Each choice should change what happens next. If waiting costs 20 minutes, the next decision should start 20 minutes later, with the queue to match.

Step 4: Decide what good looks like

Write three or four behaviours an observer could see or check afterwards. "Escalates within 10 minutes" is observable. "Shows good judgement" is not.

Step 5: Run it, then debrief within minutes

For an assessment-style attempt, let the person act without coaching, then debrief while the choices are fresh. During earlier learning, guidance or pauses may be useful. Match the support to the purpose of that run.

Step 6: Re-run until the behaviour holds, and measure

Run the drill again with a suitable variation and record the actions, conditions and criteria. Later, examine real-work evidence against the relevant procedure, allowing for differences in circumstances rather than comparing unlike scores as if they were equivalent.

Worked example: a Friday-afternoon payment outage drill

This hypothetical drill trains a new support hire at a software company. The hire knows the product and the ticket queue but has never handled an outage. In 20 minutes, they must decide whether failing card payments are an incident, who to escalate to and what to tell customers. The drill uses the company's own runbook, so the hire is judged against the procedure they will follow on the job.

  • The trigger: it is 3:40 pm on a Friday, Perth time. Three customers have reported failed card payments in the last 10 minutes, and the status page shows green.
  • The facts the learner gets must be usable, not just named. For this fictional drill, three failed-payment reports within ten minutes trigger incident escalation by phone to the nominated on-call engineer within ten minutes of the third report. The holding message says reports are being investigated without promising a fix time; a confirmed status-page incident requires engineering confirmation. Fill in the approved contact and message details before running a local version.
  • The decisions: whether this is an incident or three tickets to answer one by one; who to escalate to, and when; what to tell customers, and when to update the status page.
  • Consequences that carry forward: if the hire keeps answering tickets for 20 minutes, it is 4:00 pm and the queue has tripled. If they post a status update before engineering confirms the problem, they have to retract it.
An outage drill branches from a third report at 3:40 pm to escalation by 3:50 or a twenty-minute wait that misses the deadline.

View the full-size diagram

  • What good looks like: escalate within ten minutes of the third report, issue the holding message without inventing a fix time, record a timeline and hand over before leaving. Record Demonstrated, Missed or Not observed, with the time and action seen. An unseen behaviour needs evidence rather than a guessed score.
  • The debrief questions: "When did you know it was an incident, and what told you?" / "What did customers know at 4:00 pm?" / "What will you do at report number three next time?"
Simulation drill kit

An editable Word kit for the payment-outage drill: the scenario card (trigger, facts, three decisions, two carry-forward consequences), the four-behaviour observer sheet and the three-question debrief sheet. A blank copy of each follows for your own drill. No email required.

Building a simulation with an AI course builder

I gave this drill to our course agent in a test workspace. My prompt asked for a 20-minute decision simulation for a new support hire, built from the trigger, facts, decisions, consequences and four-behaviour rubric above, ending with the three debrief questions.

Before planning, the agent asked for the runbook excerpt, engineer details and message template because my original brief named them without their contents. I told it to use only the supplied facts. It therefore built an incomplete drill rather than inventing those resources. The one-page simulation had two six-decision runs, a comparison, scorecard and the three written debrief questions. Its hand-over note flagged an empty comparison, unresolved Demonstrated/Missed results and a learner-visible replay instruction. The platform estimated about eight minutes against the planned twenty.

Then I played it. On Decision 1 I kept answering tickets, and the simulation said 20 minutes had passed with the queue tripled, yet Decision 2 still offered to escalate "within 10 minutes of the third report". Each step reset, and every option showed its outcome under the button before I chose, which turned the decision into reading.

My first fix asked the AI editor for three changes to Decision 2 at once. The chat showed "Edited 1 page", but after two reloads only the intro text above the simulation had changed. The simulation itself was the same.

So I sent one narrow request instead. It told the editor to store my Decision 1 answer and, on the wait path only, start Decision 2 with "It is now 4:00 pm and the queue has tripled." It also gave the escalation outcome: "You escalate at 4:00 pm, 20 minutes after the third report. The 10-minute escalation behaviour is missed." The confirm path was to stay exactly as it was, and nothing else on the page was to change.

This time it landed. I reloaded and played the whole first run, all six decisions, on the wait path. Decision 2 opened at 4:00 pm with the queue tripled and marked the 10-minute escalation as missed. In a control run after another reload, choosing to confirm the incident left Decision 2 as it was, so the branch depends on the earlier choice, as I asked.

One check by hand caught what the confirmation did not. On the wait path, the summary at the end of the run said "It is 4:00 pm and the queue is tripled", then "Escalation occurred within 10 minutes of the third report." The turn I edited had changed, but the record behind the summary had not, so the scorecard would credit a behaviour the learner missed. Outcomes also still show under each option before the learner chooses. I did not replay the second run or open the comparison after this edit.

The lesson is to check the page, not the confirmation. "Edited 1 page" appeared both times, and only playing the drill to the end showed what had changed. One narrow, named request made a later decision follow an earlier one. Then I had to check the end summary as well as the turn I edited.

Limitations of simulation training

  • Build cost: scenarios take time to write and maintain. Facilitated drills also use staff time; self-guided versions still need suitable feedback and review.
  • Wrong scenario, wrong lesson: a drill built on an outdated runbook trains people to follow it.
  • No automatic transfer: doing well in the drill does not prove someone will do the same on a real Friday.
  • Health-care skew: most of the evidence comes from health professions and compares simulation with no training.
  • Guessable branches: over-scripted choices teach people to hunt for the "right" button instead of reading the situation.

Most of these limits are managed with two habits. Rewrite the scenario whenever the process changes, and check what people do on the job after the drill, not only how they scored in it.

Putting simulation training to work

Simulation training gives people a controlled place to practise the task and examine the result. Start with the simplest format that preserves the relevant information and actions, then check whether equipment accuracy, guidance or facilitator support is needed. Use workplace follow-up to assess transfer rather than treating a drill score as proof.

Pick one decision that went wrong in your team last quarter. Write its trigger, the facts, two consequences and four observable behaviours, run it with one person this week and debrief before they leave the room. If you cannot spare a facilitator for every run, keep the debrief questions written down and review the answers yourself, knowing the consequences will be less vivid without someone calling them out.

Frequently Asked Questions

Simulation training lets people practise a model of a real task in a controlled setting, with feedback on what happens. It can rehearse procedures, systems, conversations or decisions while reducing exposure to the consequences of real work.

Pilots handle engine failures in full-flight simulators, nurses manage a deteriorating patient manikin, and forklift operators train on simulators before touching a real machine. In offices, support staff run outage drills, finance teams practise in a copy of the billing tool and new managers play business games.

For the decision drill in this guide, define the task, supply the trigger and usable facts, connect choices with consequences, and agree observable criteria. Run the attempt, debrief it and plan another attempt or workplace check. Other simulations may need different preparation and support.

Simulation training is effective compared with no training: a JAMA meta-analysis of 609 health-profession studies found large effects on knowledge, skills and behaviours. Most of the evidence comes from health care and compares simulation with doing nothing, so check the results in your own workplace.

Role play is one type of simulation training, where the task being practised is a conversation with a customer, patient or colleague. Simulation training also covers equipment, software systems, crisis drills and business games. Our role-play simulation guide covers conversation practice in depth.

Alex Hey

Alex Hey

Chief Operating Officer

Alex Hey is a technical marketer with 10 years of hands-on growth and digital marketing experience. He cofounded Coursebox together with Travis Clapp who he met online when looking for a cofounder to build a globally impactful software product. He works in the Perth, Western Australia office as the COO. He built the world's first automated funnel for patient recruitment before building Coursebox. After launching Coursebox in mid 2023, he drove +400k organic sign ups for Coursebox. Alex leads the marketing and digital growth of Coursebox and works with the product development team bringing a passion for data-driven product decision and exceptional user experience. Alex is determined to build the world's easiest way to build the perfect course.