An Arcade Catalogue Is Not Courseware: What Licensed Games Can Actually Do Inside an LMS

Licensed HTML5 games can work inside an LMS, but not as courseware. The blockers are cmi5 packaging, ad-based monetisation and student privacy law. None of them is the game.

Every few months someone in edtech asks whether they can drop a catalogue of casual games into a learning platform and call it engagement. The answer is yes, narrowly, for about three jobs — and the reason most of these projects stall has nothing to do with whether the games are any good. It has to do with how a learning platform launches content, who is allowed to see the data that comes back, and whether the thing paying for the game is allowed to exist in front of a thirteen-year-old.

This is worth getting right, because the failure mode is expensive in a specific way: the games get licensed first and the integration questions get asked second, at which point the licence is already signed and the answer is no.

📊 The Research Says It Works. It Does Not Agree on How Much.

Start with the evidence, because it is usually quoted at you selectively.

Gamification research is one of the most heavily meta-analysed corners of education technology, and the recent meta-analyses land all over the place. Li, Ma and Shi's 2023 meta-analysis in Frontiers in Psychology pooled 41 studies and 49 independent samples and reported an overall effect of g = 0.822, with a 95% confidence interval running from 0.567 to 1.078 — a large effect, and a wide one. Kurnaz's 2025 meta-analysis in Psychology in the Schools, looking specifically at K-12 motivation, reported g = 0.654. A 2025 ETR&D meta-analysis of game-element combinations reported d = 0.566. A 2026 SAGE Open meta-analysis of higher education reported d = 1.12 — and, tellingly, heterogeneity of I² = 98.84%.

That last number is the honest headline. An I² near 99% means the studies are not measuring the same thing. The moderator analyses make it worse rather than better: some meta-analyses find effects decay as interventions run longer, which is the classic novelty-effect story; Li, Ma and Shi found the opposite, with interventions running beyond one semester producing the largest effect in their sample. Both cannot be a general law.

What survives all of this is modest and still useful: adding game mechanics to a learning context reliably moves motivation and engagement measures, the size of that movement depends enormously on implementation, and the effect on motivation is better evidenced than the effect on competence. Kurnaz found gamification moved extrinsic motivation (g = 0.713) more than intrinsic (g = 0.638), which should tell you what you are actually buying.

Do not walk into a procurement meeting quoting a single effect size. Quote the range and say why it is that wide. Anyone who has read the literature will trust you more for it.

🧩 Three Jobs a Licensed Game Can Actually Do Here

A licensed arcade catalogue is not instructional content. It contains no learning objectives, no assessment, no alignment to a curriculum, and no subject-matter accuracy guarantee. Trying to sell it as teaching material is how you lose the deal and the reference. There are three jobs it does well.

An instant, cheap reward you can actually deliver

Learning programmes constantly need small rewards and have almost nothing to give. Points and badges are free and feel free. A two-minute game unlocked at module completion is a real, deliverable, non-cash incentive with no fulfilment cost, no shipping, and no tax treatment. This is the single most defensible use, and it is the one that survives a sceptical L&D director.

Break structure inside long compliance modules

Mandatory compliance training is long, disliked, and measured by completion rather than comprehension. Interleaving short, unrelated play between heavy segments is a pacing decision, not a pedagogical claim. It reduces the drop-off in the middle third of a module — the part where completion rates die.

The top of the funnel, where nobody is a student yet

Open days, recruitment microsites, careers pages, campus portals, alumni properties, summer bridge programmes. These sit outside the LMS entirely, carry no student records, and have none of the constraints below. If you are testing whether games belong in your education stack, test here first. The integration is an embed, the legal review is short, and you will learn something real about whether your audience plays.

Notice what is not on this list: teaching a concept. If you need a game that teaches long division, you need a serious game built against learning objectives, and that is a commissioned development project, not a catalogue licence.

📦 An LMS Does Not Want an iframe

This is where most projects discover the real cost. Learning platforms are not websites; they are systems of record. They expect content to announce itself, report completion in an agreed vocabulary, and be launchable with an identity attached. A licensed HTML5 game ships as an HTML file, a bundle of assets, and possibly an ad SDK. It does none of those things.

Four standards matter, and they do different jobs.

SCORM: the portability floor

SCORM packages content into a zip an LMS can store, launch and track, with data held inside the LMS. It is old and it is not where new work should start, but two decades of installed content means almost every platform still reads it. If your buyer's requirement document says "must be SCORM-compliant", it usually means "must import without a developer", and that is a solvable wrapping problem rather than a rebuild.

xAPI and cmi5: tracking, and tracking with rules

xAPI defines how any system reports what a learner did, to a Learning Record Store rather than to the LMS. Its strength is that it can describe anything; its weakness is exactly the same, and in practice every organisation invents its own xAPI dialect, which is an interoperability problem dressed up as flexibility.

cmi5 is the fix. It is an xAPI profile that adds SCORM-style launch and completion semantics on top — so you get structured launch, interoperable completion and mastery rules, and free-form telemetry to an LRS. Rustici's comparison of the three is the clearest summary of the split, and cmi5 is the sensible target for anything built new. It also requires an LRS, which is a procurement line item your buyer may not have.

LTI 1.3: the one with a hard deadline that already passed

LTI is how an external tool gets launched from inside a course with identity and context attached, and it is the right mechanism when the game stays on your infrastructure rather than being packaged into theirs. The version matters. 1EdTech stopped certifying legacy LTI after 30 June 2021 and ended all support for LTI 1.0, 1.1, 1.2 and 2.0 on 30 June 2022, on security grounds. LTI 1.3 and LTI Advantage are the minimum certifiable versions. If a vendor offers you an LTI 1.1 integration in 2026, they have not touched that code in four years.

Practically: pick one target platform before you build any of this. In North American higher education the market is concentrated enough that this is an easy call — Edutechnica's Spring 2025 LMS data update puts Canvas at roughly 39% of institutions and about half of total enrolment, with Canvas, Blackboard, Moodle and Brightspace together accounting for the large majority of implementations. Build for one, certify against it, and expand.

⚖️ Ad-Funded Games Do Not Belong in a K-12 Deployment

This is the constraint that kills more education game deals than any technical issue, and it is almost never raised until legal review.

The standard monetisation model for a licensed HTML5 catalogue is advertising — interstitials, rewarded video, display. In a US K-12 context, that model is largely unavailable. California's Student Online Personal Information Protection Act (SB 1177, 2014) prohibits operators of sites and apps designed for K-12 school purposes from knowingly engaging in targeted advertising to students or their parents, from using information gathered through the service to amass a student profile except for school purposes, and from selling student information. More than 20 states have since passed laws modelled on it, with variations — some of them carve out exceptions the California law does not.

On top of that sits FERPA. When a district lets a third party handle student records, it is typically relying on the school-official exception, which requires the vendor to be under the district's direct control and to use the data only for the authorised educational purpose, with redisclosure restricted. An ad SDK phoning home is difficult to reconcile with "direct control".

The consequence for licensing is concrete: education deployments need an ad-free build and a licence that permits one. That changes the commercial shape. You are no longer selling a revenue-share on ad inventory; you are selling a flat licence, per-seat or per-institution, with the ad stack stripped. Price it that way from the start rather than discovering it in month three.

💻 The Device on the Desk Is Probably a Chromebook

Education hardware is not the phone market. Futuresource Consulting's K-12 tracking put Chromebooks at 39% of global K-12 device volume back in its January 2023 industry update, against 37 million units shipped in 2022; more recent secondary citations of the same tracker put the share higher still, though the underlying reports are paywalled and the numbers quoted around the web vary by how they count districts versus units. The structural point holds regardless: a large share of the install base is low-specification ChromeOS hardware bought on a multi-year refresh cycle, often several years into it.

That has three implications. Assume constrained memory and modest GPUs, not current mid-range phones. Assume the network is a school network — filtered, proxied, and capable of blocking domains your game loads assets from, which is a real and frequent cause of "the game doesn't work" tickets. And assume the session is bounded by a bell, not by the player. A title designed around a 25-minute engagement loop is the wrong title for a five-minute gap between activities.

🏢 Corporate Training Is a Different Buyer Entirely

Move from K-12 to corporate L&D and most of the legal constraint disappears — adult learners, no student privacy statutes, no age gating. What replaces it is procurement rigour and a demand for measurement.

Corporate buyers will ask what the game reports and where it reports to. This is where cmi5 earns its keep: if you can emit structured completion plus meaningful telemetry into their LRS, you are a system that participates in their analytics. If you can only say "the iframe was loaded", you are a novelty they will cut at renewal.

Ignore the market-size numbers you will be handed in this segment. Vendor forecasts for game-based learning disagree wildly — one 2025 report put the global market near USD 29.6 billion growing to USD 181.8 billion by 2034, another sized the corporate segment alone at USD 9.52 billion in 2025 heading to USD 59.24 billion by 2034. Different scopes, different definitions, no reconciliation. These are marketing documents, not measurements, and quoting them in a pitch to an L&D director who has seen the same three charts will not help you.

🚫 Five Ways These Projects Fail

• Licensing before checking the ad clause. You buy a catalogue monetised through advertising, then discover the education deployment requires the ads gone and your licence does not grant an ad-free build. Ask before you sign.
• Selling entertainment as instruction. The moment you claim a match-three game teaches something, a curriculum specialist will ask which objective it maps to, and the meeting ends. Sell it as reward, pacing and engagement — claims you can defend.
• Shipping an iframe and calling it integrated. No completion signal, no identity, no data back. It works in the demo and gets removed in the first platform review.
• Ignoring the school network. Assets served from a domain the district's filter does not recognise, third-party fonts, an analytics call that times out. All of it fails silently in a classroom and loudly in a support queue.
• Measuring plays. Play counts prove nothing to an education buyer. Completion of the module the game is attached to, drop-off in the segment before it, and repeat sessions per learner are the numbers that renew a contract.

🎮 Where a Licensed Catalogue Fits

Forestry Games has licensed HTML5 and Android titles since 2017, across a catalogue of 1,049 games, and the education-adjacent question comes up regularly — usually from a portal operator or an agency serving a school, university or corporate client rather than from a curriculum publisher. The practical fit is the reward and engagement layer described above, not courseware: a school-oriented selection for age-appropriate deployments, browser-based titles that run without an install on managed devices, and licence terms that can accommodate an ad-free build where the deployment requires it. Whether the games are packaged into SCORM or cmi5, or launched over LTI 1.3, is an integration decision to make with your platform team before the licence is scoped — not after. If you are sizing a broader deployment, the full catalogue and the white-label portal option are the two starting points worth comparing.

🧭 What to Do Before You Pitch Games Into a Learning Stack

Four things, in this order.

1. Name the job. Reward, pacing, or top-of-funnel. Write it down. If the answer is "teaching", stop and scope a commissioned build instead.
2. Confirm the audience's legal shape. Under-13 K-12, over-13 K-12, higher education, or adult corporate. This single question determines whether ad monetisation is available to you, and therefore what the commercial model is.
3. Pick the launch mechanism and the platform. SCORM for maximum portability, cmi5 if the buyer has an LRS and wants analytics, LTI 1.3 if the games stay on your infrastructure. One target LMS first.
4. Test on the worst device your audience actually uses, on a filtered network, in a five-minute window. If it survives that, the rest is contract work.

The teams that make this work are the ones that treat the learning platform as an integration target with rules, rather than as a website that happens to have students on it. Get the packaging standard and the monetisation model settled before the licence is signed, and the games are the easy part. Get them backwards, and no catalogue in the world will rescue the project.