The Screen in the Aisle Runs Chromium 56. Put the Game on the Shopper's Phone.

Retail signage screens run pinned Chromium builds, with no input device and no session, so most licensed HTML5 games will not run in-store. Here is what does. The budget question underneath it is a separate decision, and getting the two confused is what makes these projects expensive.

The brief usually arrives from a marketing team that has just walked a store refit. There are forty new screens above the aisles, a pair of touch kiosks by the entrance, and someone has said the word "engagement" in a meeting. Could we put games on them? The honest answer is yes, but almost never in the way the brief imagines, and the two reasons have nothing to do with creative.

💸 The Budget Pool Is Smaller Than the Screen Count Suggests

If your plan is to sell game content into retail media inventory, look at the size of that pot before you build anything. EMARKETER's H1 2026 forecast puts US retail media ad spending at $72.97 billion in 2026, growing 20.3% year over year. In-store is $0.58 billion of it.

That is not a rounding error caused by newness. EMARKETER's June 2026 analysis of the demand-execution gap makes the case that spending is small because execution is hard, not because advertisers are uninterested. Three structural reasons it cites are worth carrying into any in-store proposal:

• Retail is fragmented in a way ecommerce is not. The top 15 US retailers accounted for under 40% of overall retail sales in 2024, against nearly 70% of online sales for the top 15 ecommerce players. There is no in-store equivalent of buying Amazon.
• The networks are siloed internally. 86% of grocers report that their print circulars, digital media networks and in-store media networks are completely siloed or only partially integrated.
• The hardware and operational lift is the barrier, not the creative pipeline.

For calibration on how slowly this moves: back in June 2024, EMARKETER forecast that US in-store retail media would reach $1.06 billion by 2028 and still not clear 1% of total retail media spend. Two years on, the trajectory has not been rewritten.

The practical read is that in-store screens are a poor place to build a content-syndication business and a reasonable place to spend a brand's own activation budget. Those are different sales, with different buyers, and conflating them is how these projects get mispriced.

📺 The Screens Are Real, and the Analysts Disagree About How Real

The installed base is genuinely large, but be careful quoting a single number at it. Grand View Research valued the global digital signage market at $31.1 billion in 2025, with retail the largest vertical at over 19% share. Global Market Insights, in a February 2025 report, put the same global market at $23.7 billion in 2025 — and had retail at $9.6 billion, roughly 40% of the total.

Those two houses are about 30% apart on market size and roughly double apart on retail's share of it. They are measuring different things: hardware-plus-software-plus-services scopes vary, and "retail" absorbs everything from quick-service restaurant menu boards to shop-window video walls. If a deck you are reading quotes one of these figures without the scope, treat the precision as decorative. Growth rates cluster in the 8–9% CAGR band across both, which is the part they actually agree on.

🧱 The Constraint Nobody Checks Until Install Day

Here is the thing that kills more in-store game projects than any commercial factor: the screen is not a modern browser.

Commercial displays and signage players ship with a browser engine welded to their OS version, and that OS version is set by the model year of the panel bolted to the wall. signageOS maintains a public map of browser engine versions by platform, and it is sobering reading:

• Samsung Tizen: Tizen 4.0 ships Chromium 56. Tizen 5.0 is Chromium 63. Tizen 6.5 reaches Chromium 86, and Tizen 7.0 Chromium 94.
• LG webOS: webOS 3.x is around Chromium 38. webOS 4.0–4.1 is Chromium 53. webOS 6.0 gets to Chromium 79.
• BrightSign: firmware 7.x is Chromium 45; 8.0.x is 65; 8.1–8.4.x is 69; 8.5.x–9.0.x is 87, with newer builds upgradable to 110.

Chromium 53 shipped in 2016. A shop fitted out in 2019 and never touched since is running an engine that predates a large fraction of what a modern HTML5 game assumes — and ScreenCloud's own compatibility guidance states plainly that Chromium 56 and lower no longer receive security updates, which makes running arbitrary web content on them a risk decision as well as a technical one.

This is a harder floor than the low-end Android problem, because on Android the browser eventually updates itself. A signage panel does not. Its engine is frozen until someone schedules a firmware pass across the estate, and estate-wide firmware passes are the exact operational cost retailers avoid.

What to do about it

Ask for the device inventory before the creative brief, not after. You want make, model, OS version and firmware for every screen in scope. Then set your build target to the oldest engine in the estate, or explicitly scope those screens out of the interactive portion and run them as video. Testing on a laptop and a phone tells you nothing about the panel.

🕹️ Three Things Every Game Assumes That a Store Screen Does Not Have

Even on a screen with a current engine, a catalogue title is the wrong shape for the environment. Ordinary web games are built on three assumptions that fail in-store.

An input device. The overwhelming majority of retail screens are display-only: no touch layer, no controller, no keyboard. A game whose first instruction is "tap to start" is a static image of a game.

A session. Web games assume someone chose to be there and will stay for minutes. A shopper walking an aisle gives a screen seconds, in an unpredictable order, with no beginning or end. There is no onboarding window.

An identity. No accounts, no saved progress, no leaderboard that means anything, and — importantly — no defensible way to personalise. The Grocery TV In-Store Shopper Perception Report 2026, produced with Media, Ads + Commerce and covered by EMARKETER in June 2026, found personalised ads ranked last among content preferences at 27.8%, with shoppers citing privacy discomfort about public displays. Whatever you build, the screen is a public object.

Strip those three away and what is left is not a game. It is an attract loop: a short, silent, self-restarting visual that reads in three seconds from six metres, in a store with the audio off. That is a real deliverable and worth commissioning. It is not what a licensed game is.

📱 The Playable Part Belongs on the Shopper's Phone

The workable pattern is a handoff. The wall screen does the job it is good at — being large, bright and unmissable in a zone where people already slow down — and the play happens on the device already in the shopper's hand, which has a current browser, a touchscreen, an account, and a session that survives leaving the store.

The scan behaviour supports it. Bitly's 2025 trends research, summarised in its QR code statistics roundup, reported QR usage up 25% year over year, with scans rising in every major region it measured: Europe up 42%, LATAM 40%, APAC 21%, MENA 18%. On the consumer side, Bitly found the leading motivation for scanning was exclusive content or information at 39%, ahead of discounts or promotional offers at 33%, with contest entries at 14% and loyalty points at 12%.

Read that ordering carefully, because it inverts the usual retail assumption. "Scan for 10% off" is the second-best offer. "Scan for something you cannot get otherwise" is the first — and a playable game is, structurally, exclusive content. That is the strongest argument for games in this channel, and it comes from scan-motivation data rather than from anyone's engagement deck.

Design consequences of the handoff model:

• The screen never needs to run the game. It needs to run a five-second loop and a code that scans from three metres in overhead lighting.
• Your build target becomes the shopper's phone, not the panel. All the browser-version pain disappears.
• The game can have a backend, a prize mechanic and a data capture step — none of which a wall screen could ever support.
• Play continues after the shopper leaves, which is the one thing an in-store screen genuinely cannot buy on its own.

🧍 Where a Real Kiosk Still Earns Its Place

Touch kiosks are the exception, and placement decides whether they work. The same Grocery TV research found that store entrances, checkout, deli and pharmacy zones all reached around 84% shopper favourability, specifically because those are places people already stand still. Cooler screens scored 58.9%, and shelf blades and shelf talkers came in below 72% — they obstruct the product while a task-focused shopper is trying to get past.

That maps cleanly onto where an interactive unit can work. Put a playable kiosk where there is a queue or a wait — a service counter, a pharmacy collection point, a click-and-collect desk, a fitting-room corridor — and you are filling dead time people already resent. Put one mid-aisle and you have built an obstacle.

The broader receipt for in-store screens is strong: that report found 62% of shoppers bought something directly after seeing it on an in-store screen, and 95% make at least half their purchase decisions while in-store. Acceptance held across generations, from 74% among 18–29s to 81% among 30–44s. Screens work. The question is only ever what you put on them and where they hang.

If you are specifying a kiosk build, the discipline is close to what a good trade show game brief demands: fixed short round length, automatic reset when abandoned, no text entry, an attract state that recovers itself, and hardware chosen for the orientation you actually have. The difference is permanence. A show stand runs for three days with staff beside it. A retail kiosk runs for two years with nobody watching, so unattended recovery is not a nice-to-have.

🚫 Five Ways These Projects Waste the Budget

1. Licensing a catalogue before auditing the estate. Buying 200 titles for screens that cannot render them is the most expensive version of this mistake, and the most common. The device list costs nothing to request.
2. Treating "interactive" as a creative choice. Interactivity is a hardware fact. A screen either has a touch layer wired to a player that supports it, or it does not.
3. Measuring plays. On a wall screen, plays are unmeasurable and meaningless. Measure scans, then what happened after the scan. On a kiosk, measure completed rounds and abandonment point, not sessions started.
4. Assuming the store has usable network. Retail Wi-Fi is provisioned for point-of-sale and stock terminals, and store IT will not open it up for your content. Anything you ship should cache locally and survive the connection dropping mid-day.
5. Personalising on a public screen. The privacy discomfort is documented, the creative upside is thin, and the compliance exposure is real. Personalise after the handoff to the phone, where there is a consent step and an actual user.

🎮 Where a Licensed Catalogue Fits

Forestry Games has licensed HTML5 games since 2017, with a catalogue of 1,049 titles across HTML5 and Android, and builds HTML5 games in-house — including work with branded IP for brand partners such as Disney, Nickelodeon, Cartoon Network and Warner Bros.

For a retail programme, the useful split is straightforward. Licensed titles from the catalogue are the right economics for the phone side of a handoff, where you want several games behind one scan and a reason to come back — buying that content is far cheaper than commissioning it. A custom build makes sense for the kiosk itself, or where the mechanic has to carry a specific product story, which is closer to how marketing game briefs usually run. The wall-screen attract loop is a motion design job, not a game licence.

🧭 Before You Sign the Installation Order

Three things, in this order.

Get the device inventory — make, model, OS, firmware — and find out the oldest engine in the estate. That single document determines whether an interactive plan is possible at all, and it is usually obtainable in a day from whoever manages the signage CMS.

Then decide, explicitly, whether you are selling into retail media inventory or spending a brand activation budget. The first is a $0.58 billion US market in 2026 with structural execution problems; the second has no such ceiling, because it is your own money buying your own attention in your own stores.

Then place the interactive units where people already stop moving, and put the actual game on the phone. Every constraint that makes a store screen a bad games platform disappears the moment the play moves to the device the shopper brought with them — and unlike the screen, that device leaves with them, still holding whatever you gave it.