Every Network Caps a Playable at 5 MB. That's Why It Can't Double as Your Campaign Game.

A playable ad and a branded campaign game are different products: one lives in a 5 MB ad slot, the other on your own site. Here's how to brief and budget each. The confusion costs agencies real money, because the brief usually arrives asking for "a playable" and nobody establishes which of the two things is meant until the build is halfway done.

I have watched this go wrong in the same way twice. A brand team sees a competitor's interactive ad unit, likes it, and asks for something similar. Somewhere between the brand team and the studio, "something similar" becomes a three-minute game with a leaderboard, a prize draw and a data capture form. Then the media agency tries to traffic it, and discovers it is roughly forty times too large to serve.

📦 The Spec Sheet Decides the Product, Not the Creative Deck

Playable ad units are not loosely defined. They are constrained by the container they run in, and the constraints are published. Unity's playable asset specifications are representative of what every major network asks for:

• Under 5 MB. Total. Everything.
• A single index.html, inlined and minified, "with no links to other files or folders." Your sprite sheets, audio and fonts are base64 strings inside the document.
• MRAID 3.0 compliant. The ad talks to its container through a standard API, not to your servers.
• No network requests. Unity's wording is that ads "should not need any network requests (XHR)," with narrow tolerance for analytics calls carrying no personal data.
• Wait for viewableChange before the playable content starts, and route the call-to-action through mraid.open rather than a plain link.
• Both orientations, and never obstruct the container's close button.

Read that list as a product spec rather than a technical annex and the shape of the thing becomes obvious. No backend. No persistence between sessions. No account. No server-side leaderboard. No form post. A single self-contained file that has to load, teach, and convert inside an interstitial slot the player is trying to close.

Now hold that against a typical campaign game brief: multiple levels, saved progress, a leaderboard visible across sessions, an email capture gate, a prize mechanic with terms and conditions, and analytics that feed a CRM. Every one of those requires a server, and the format explicitly cannot have one.

This is not a limitation you engineer around. It is the boundary between two products.

📈 What the Playable Format Is Genuinely Good At

The performance case for playables is real, and it is better documented than most things in ad tech. Liftoff's 2025 Mobile Ad Creative Index, its seventh annual edition, was published in July 2025 and drew on 4.7 trillion impressions, 263 billion clicks and over 1.1 billion installs recorded between January 2023 and May 2025.

Its headline finding on the format: impression-to-install conversion for playable ads ran at roughly 8x non-playable formats for top-spending game advertisers, and 16x for everyone else. Spend followed. Top-spending games put about 35% more of their budget into playables than other games did, and share of spend on interactive formats grew around 20% year over year through 2024.

Two caveats belong next to those numbers, and neither is in the press release.

First, it is network data. Liftoff measures campaigns running on its own demand infrastructure. That is a large and credible sample, but it is not an independent audit, and the same is true of every creative benchmark published by a company that sells the format.

Second, and more important for a brand: the metric is installs. The whole documented advantage of a playable is that letting someone try a game before downloading it converts better than showing them a video of it. That is a mechanism specific to advertising an app with the app's own mechanics. A bank, a soft drink or a supermarket has no app loop to preview. The 16x does not travel with you.

Treat the brand-recall figures with more suspicion than the install figures

Search for playable ad brand recall and you will find confident percentages — a 36% lift here, a 53% recall rate there — recycled across vendor blogs with no linked methodology, no sample size, and in several cases no date. Some trace back to studies that are the better part of a decade old, run by companies that no longer exist under that name. If you cannot find the methodology, do not put the number in a client deck. Make the argument qualitatively instead; it survives scrutiny better than a statistic that falls apart when someone asks where it came from.

💰 Nobody Agrees How Big This Market Is, and the Gap Is Instructive

Ask two research firms to size games advertising and you get answers an order of magnitude apart. eMarketer's 2026 briefing on video game advertising forecasts US game ad revenues surpassing $10 billion by 2029, and puts in-game advertising at about 2.3% of total US digital ad spending in 2026. Market-research houses publishing "in-game advertising market" figures put the 2026 global number north of $100 billion.

Both can be defensible. They are counting different things: one is media dollars bought against game inventory in a single market, the other is a global figure that sweeps in adjacent spend on advergames, sponsorships, esports and product placement. The lesson for anyone writing a business case is to check the definition before quoting the number, because the two figures imply wildly different things about how much budget is actually available for the thing you are selling.

What is more useful than either is the demand-side signal. The IAB's Changing the Game study, published in March 2024, found 86% of surveyed advertisers agreed games advertising was of growing importance to their company and 86% considered it brand safe, while 79% called it relatively easy to buy — still behind other digital channels on ease of planning and purchasing. Interest is not the bottleneck. Operational friction is.

🧱 What a Campaign Game Actually Is

The other product goes by advergame, branded game, campaign game — the name matters less than the fact that it lives on a surface you control. A microsite, a page inside the brand's own site, a QR destination on packaging, a tablet at a stand. Once you are off ad inventory, the constraints invert:

• No 5 MB ceiling. There is a load-time budget, which is a real constraint, but it is measured in seconds of patience rather than megabytes of container limit.
• A backend is allowed. Accounts, saved progress, cross-session leaderboards, server-validated scores for anything with a prize attached.
• Sessions run in minutes, not seconds. Somebody who typed your URL or scanned your code has already opted in. The playable's fifteen-second tutorial-and-endcard structure is the wrong shape for them.
• Data capture is native to the experience rather than smuggled past an ad container that forbids it.
• The asset is reusable. The same build can run on the campaign site, on a stand at a trade show, on a kiosk in a retail space, and inside a loyalty app. A playable ad unit has exactly one life, in an ad slot, until the flight ends.

And the constraint the playable never has: you have to bring the traffic yourself. An ad unit is bought into an audience. A campaign game sits there until media, email, packaging, social or floor traffic delivers someone to it. This is the single most common reason branded game projects underperform, and it is a media planning failure rather than a game design one.

🔗 The Honest Version: They Work Together, in One Direction

The sequence that does work is playable as the acquisition unit, campaign game as the destination. The 5 MB unit runs in-feed or interstitial, teaches one mechanic in a few seconds, and its call-to-action opens the full experience on your own domain, where persistence, capture and depth are all legal.

That requires building two things from one design, which is a legitimate brief — and a very different one from "make the campaign game smaller until the network accepts it." The compression job is where projects die. You cannot inline a 40 MB build into a single HTML file and stay under 5 MB, so what actually happens is that features get amputated in the last week of the schedule, and the surviving unit is a worse ad than one designed for the slot from the start.

📊 Measure Them on Different Things

Applying install-economy metrics to a brand campaign is how good work gets killed in the readout.

For the playable ad unit: completion rate of the interactive segment, click-through on the end card, cost per landing on the destination, and creative fatigue — how many days before performance decays enough to warrant a new variant. Refresh cadence is short in this format; plan for variants, not for one perfect build.

For the campaign game: time on the experience, rounds per visitor, return visits within the campaign window, opt-in rate at whatever gate you set, and cost per qualified opt-in against the same figure from the campaign's other channels. That last comparison is the one clients actually care about and the one nobody instruments in time.

Do not report plays. Plays is a vanity number that a single autoplay bug can multiply. Report people, and what those people did afterwards.

🚫 Five Ways Brands Waste This Budget

1. Briefing one product and buying the other. Decide, in writing, before design starts, whether the deliverable is trafficked ad creative or a hosted experience. If it is both, price two builds.
2. Commissioning a campaign game with no traffic line in the media plan. A game on a microsite with no paid, owned or physical route to it will get the visits the microsite was already getting, which is usually close to zero.
3. Custom-building a mechanic that already exists. If the brand idea is "a match-3 with our packaging in it," a bespoke build is spending most of the budget re-solving a solved problem. Reskinning a proven HTML5 title puts the money into the brand layer and the campaign instead.
4. Ignoring the approval calendar. Any campaign carrying a licensed character or a regulated product adds approval rounds that are not in the studio's timeline. That schedule belongs in the plan on day one.
5. Treating launch as the end. Ad creative needs variants across the flight. A campaign game needs a reason to return — a weekly reset, a new level, a leaderboard that closes on a date. Static campaign games die in week two.

🎮 Where a Licensed Catalogue Fits

Most brand game briefs do not need original game design. They need a mechanic that already works, wrapped in the brand's world, live in weeks rather than quarters. Forestry Games licenses a catalogue of 1,049 HTML5 and Android titles and develops HTML5 games in house, which covers both routes: reskin something proven from the catalogue when the mechanic is standard, or commission a build when the idea genuinely is the game. The company has worked on branded game projects and has brand partnerships including Disney, Nickelodeon, Cartoon Network and Warner Bros, so the approval-and-compliance side of a branded build is familiar ground. If you are scoping a campaign, the marketing games pages are the place to start.

🧭 What to Put in the Next Brief

Four lines at the top of the document, before any creative description:

1. Where does this run? Ad inventory, an owned URL, or physical hardware. Name the networks if it is inventory, because their specs are the build constraints.
2. What is the success metric? One primary number. Cost per landing, cost per opt-in, or dwell time — not all three.
3. What brings people to it? For an ad unit this is the media buy. For a hosted game, name the specific channels and the budget behind them.
4. What happens after the flight? Archive, or reuse at events, retail and internal activations. Reuse changes what is worth building.

Answer those and the format question answers itself. Skip them, and you will find out which product you were building the week the media agency asks for the trafficking sheet — which, in my experience, is exactly when there is no time left to change it.