Game-Based Learning: What It Is and When It Works in Training

Turn boring content into gamified learning instantly

Start Free

The brief I see most often for a learning game is one word long: fun. A manager wants new starters engaged, and the request reaches me as "can we make the induction a game?" The question I ask back is what the player has to do to win, because that answer tells me whether we are building practice or packaging.

Game-based learning uses games to support learning through their goals, rules, challenges and feedback. At work, a useful design aim is to make progress depend on practising the knowledge or actions the role needs.

My design check is whether the winning move practises what the learner needs. It helps expose a mismatch, such as rewarding fast clicking when the objective is a careful decision. A recall game can still be game-based learning, and gamification can reward useful performance; neither label tells you whether the design is good.

I'm an instructional designer at Coursebox, where we build tools for practice activities such as branching scenarios for workplace training. I review plenty of "games" that turn out to be click-to-reveal pages with a score. This guide is about adults at work.

What is game-based learning?

Game-based learning uses games to support learning through their goals, rules, challenges and feedback. At work, a useful design aim is to make progress depend on practising the knowledge or actions the role needs.

I will carry one hypothetical example through this guide. A café group with 12 sites is practising how staff respond when a customer declares an allergy. The fictional exercise uses four actions: ask which allergen, check the supplied information, explain what can and cannot be established, and flag the order. Its invented matrix lists walnut and wheat in banana bread and a tree-nut facility warning for oat milk. This is a limited content-design exercise, not a complete food-allergen procedure. Real service needs current approved information, kitchen and cross-contact controls, and an escalation route when safety cannot be established.

I inspect four parts when sketching a learning game. Here they are for the fictional Order Up example:

• The goal: follow the fictional information-checking and communication steps, including stopping or escalating when the information is insufficient.
• The rules: each round is an order card, such as "Oat flat white and a banana bread, I have a tree nut allergy", and the player picks the next action against a time limit.
• The challenge: serving fast is always tempting, and the matrix has traps, like oat milk that cannot be guaranteed nut-free.
• The feedback: a wrong choice shows the consequence, such as a customer with a tree nut allergy eating banana bread that contains walnut.

View the full-size diagram

A serious game is a game designed for a purpose beyond entertainment, such as education or training. It may be digital or physical. Digital game-based learning refers specifically to learning through digital games.

Game-based learning vs gamification

Gamification adds game elements, such as points, badges and leaderboards, to an activity that is not itself a full game. Game-based learning uses a game as the learning activity. Either can reward recall, decisions or participation. In this café example, points for opening a policy reward access; a scenario can practise using the information. Judge the practice and feedback as well as the label.

Gamification still has effects. A meta-analysis of gamification in education by Sailer and Homner (2020) found small positive effects on cognitive, motivational and behavioural outcomes. The cognitive effect held in the most rigorous studies, but the motivational and behavioural effects "were less stable". Motivation is usually why a team buys points and badges. For the points-and-badges route, see our guide to gamification in e-learning.

Serious games, simulations and gameful design

These categories can overlap. Use the final column to examine the learning design rather than assuming a label guarantees useful practice.

Types of game-based learning at work

Each format practises a different kind of skill:

• Quiz-style recall games: timed question rounds for remembering facts.
• Sequencing and sorting games: putting steps in order, for procedures.
• Decision and scenario games: choices that change the story, for judgement (see our guide to scenario-based learning).
• Role-play and conversation games: saying what you would say, for conversations (see these role-play and simulation examples).
• Tabletop exercises: a team works through an incident in stages. They can support game-based learning when adapted with game goals, constraints and feedback; a discussion exercise is not automatically a game.
• Simulations: a working model of equipment too costly or dangerous to learn on.

For games to run in a live session, I've collected 12 training games picked by skill, each with a debrief.

What the research says game-based learning does

I start with the practice a game creates. Traci Sitzmann's review of computer-based simulation games, published in Personnel Psychology in 2011, compares learning outcomes with other methods. Her university also summarised it for a general audience.

The same paper found trainees learned less from games than from the comparison method when that method also actively engaged them, which suggests the advantage came from activity. The university's summary adds that 16% of the games were too passive and "no more effective than other teaching methods".

Wouters and colleagues (2013) found average learning and retention benefits over conventional instruction, without a statistically clear motivational advantage. Clark, Tanner-Smith and Killingsworth (2016) found differences by study and game design; their single-session subgroup did not show a statistically clear advantage over non-game instruction. That is not proof a single session cannot help. Most participants in those reviews were school or university students, so I use the findings as design prompts rather than promised workplace results.

Case study: The same maths game, built two ways

Educational software has often used games as a "sugar coating" for learning content. Jacob Habgood and Shaaron Ainsworth tested the alternative: building the maths into how the game is played. They made a maths game, Zombie Division, for children aged 7 to 11, in an intrinsic version (the maths was the core of play) and an extrinsic version (the maths bolted on), plus a control.

• The evidence: two studies in the Journal of the Learning Sciences (2011) measured learning gains for 58 children who played for two hours with their teacher, then time on task for 16 children given free choice of which version to play.
• The result: the children learned more from the intrinsic version under fixed time limits, and spent 7 times longer playing it when they could choose.
• The limitation: the learners were children, the subject was school maths and the samples were small. It shows a design principle, not a workplace result.
• The lesson: the same game taught more and held attention longer when winning required the skill, which is the winning-move test in a controlled comparison.

The conditions that made games work

The reviews found differences associated with activity, repeated access and surrounding instruction. Those comparisons suggest questions to ask when designing a game, rather than proving each feature caused the result:

• Active: the player makes the decisions rather than watching.
• Replayable: trainees could play as many times as they wanted.
• Embedded: teaching came before and after, with the game as a supplement.
• Spread out: played over more than one session.

I add a debrief. It wasn't one of the conditions the studies measured, but it is where the teaching after the game happens.

When is a game worth building?

A custom game can take substantial design, testing and maintenance. Compare those costs with another activity that practises the same objective before choosing the format.

Game-based learning examples in workplace training

1. Ice cream portions at Cold Stone Creamery. The chain was losing money on staff serving too much ice cream, so it built an interactive video game that taught how much goes into each scoop, according to the university summary above.

• The job: serving the right portion.
• The winning move: not described; the summary says only that the game taught portion size.
• What the player gets back: not described either.

The same 2010 summary says Miller Brewing was developing Tips on Tap, a game showing bartenders how to pour a beer. Players lose points if the mug hits the tap, where it could become contaminated, so the penalty sits on the work mistake.

2. A ransomware tabletop exercise. The US Cybersecurity and Infrastructure Security Agency publishes over 100 free tabletop exercise packages, with scenarios including ransomware, insider threats and phishing, plus discussion questions and a feedback form.

• The job: responding to a cyber incident as a team.
• A possible game adaptation: give the team a shared response goal and explicit constraints, then ask it to decide who acts, what they do and who they tell at each stage. That adaptation is illustrative, not a claim that every CISA package has a win condition.
• What the player gets back: the facilitator's questions and an after-action report that show gaps in the plan.

3. A hypothetical customer complaint conversation scenario.

• The job: calming a customer whose delivery arrived damaged and agreeing a fix.
• The winning move: acknowledging the problem before offering a remedy the policy allows.
• What the player gets back: a reply that cools down or escalates depending on what was said.

There are more prompts like this in our customer service role-play scenarios. Published workplace evidence for individual games is thin, so treat any single company story, Cold Stone's included, as an example.

How to design a learning game, step by step

Here is how I'd build Order Up. The steps work for any format above.

1. Write the on-the-job action as verbs. "Take an allergy order safely" becomes ask, check, offer or state, flag. These SMART learning objective examples help with the wording.
2. Make that action the win condition. A round is won only when all four actions happen in order and the customer is told the truth. Speed alone never wins.
3. Write feedback that explains the consequence and the next useful action. In the fictional café task, name the ingredient information the player overlooked and explain why they need to ask before proceeding.
4. Allow replay, and start each attempt fresh. New order cards each round, so players practise the routine rather than memorise answers.
5. Teach before and debrief after. Show the matrix and the four steps first. Afterwards ask: what happened, so what, and now what will you do at the counter? These experiential learning activities use the same debrief.
6. Pilot with people from the role, including a range of experience and access needs. Five is a manageable first planning number, not validation. Watch the choices, ask about confusing rules and revise before testing again.

Limitations: when a game is the wrong choice

These are the cases where I push back:

• A one-off game day without follow-up. A single session can be useful, but the Clark review did not find a statistically clear advantage for its single-session subgroup. Plan practice around the objective rather than assuming a game day will produce transfer.
• Passive designs. Sitzmann's review found weaker results for passive games. A reveal page with a score still needs a learning purpose and suitable practice; the score alone does not establish either.
• Leaderboards that shame low scorers. I avoid them: practice is where people should be free to get it wrong. If you use one, let players opt out.
• A simpler activity meets the objective. Compare a recall game with a quiz or flashcards rather than assuming a game adds enough benefit to justify its cost.
• Build and upkeep cost. A scenario game needs rewriting every time the procedure changes.
• Learners who find games childish. Tie the win to their real job so the game reads as work practice.

Building game-based practice in Coursebox: what I tested

I gave Coursebox's course agent, the AI that builds a course from a prompt, the Order Up brief. I wanted to see whether it builds practice where the winning move is the work move, or falls back on reveal-style pages. This was one course in a test workspace.

What I asked the course agent. A prompt with the four steps as the goal, the four-line fictional matrix, a drag-and-drop ordering question, a three-customer scenario where every choice has a consequence and an ungraded quiz with feedback, plus: "Do not add points, badges or leaderboards."

What it produced. Four pages, about 15 minutes. The four steps kept my order and wording, and the matrix appeared word for word. The ordering task was a real drag-and-drop, not flip cards. The scenario was a branching block with six turns, and the quiz had three questions with feedback. I searched every page for allergens outside the matrix and found none.

What was wrong. The scenario looked like a game but didn't play like one. When I chose "Offer banana bread immediately because it may suit the customer", I got a one-line explanation, and then the customer volunteered the allergen anyway. You could win without doing step 1, which fails my own test. The correct option was always the longest, so picking the longest option won every later turn. The end screen showed a five-star rating, though I'd asked for no points or badges, and the agent reported an extra graded quiz I hadn't asked for.

The fix, and how I checked it. I asked for one revision: a wrong first choice should lead to a consequence where the customer doesn't volunteer the allergen, then send the learner back to ask, with no scores. Three things are still weak. The wrong path explains and returns rather than showing the order going ahead, the correct option is still the longest, and one stem now does the matrix lookup for the learner. I haven't rechecked the star rating or removed the extra quiz. But after a reload, playing as a learner, the core fix held. The wrong path now says banana bread contains walnut and wheat, "so its suitability cannot be established while the allergy is unknown", and its only button is "Return to: Ask which allergen." The customer no longer gives the allergen away, so step 1 is the only way forward. The test shows what the pages contain, not that baristas take safer orders.

To try a similar brief, start with our AI quiz generator or AI tools for employee training.

Putting a learning game to work

A learning game needs a clear learning purpose, useful practice and feedback. Connect what players do to the objective, whether that is recalling information, making a decision or coordinating with others. Then compare the design with a simpler activity before committing to a full build.

Before agreeing to make it a game, sketch one round on the design sheet. Check the rules and feedback with the subject expert, pilot with people from the role and revise the confusing parts. Plan how you will check the skill afterwards, because completing the game does not establish workplace competence.