Your Retail Media Network Only Has Inventory While Someone Is Shopping. License HTML5 Games for the Rest of the Week.

Retail media inventory only exists while someone is shopping. License HTML5 games and your site earns new sessions instead of slicing the same ones thinner.

Retail media is the best-funded idea in commerce right now, and for most retailers it is also the most quietly capped. The money is real. The growth is real. But the supply side of a mid-size retail media network is bounded by something nobody writes on the slide: a shopper has to be on your site, in a session, before you have anything to sell at all.

That constraint decides more about your ceiling than your ad server does. And it is the one constraint a licensed games catalogue actually moves.

📈 The Money Is Growing and Almost None of the New Money Is Coming to You

Start with the numbers, because they are unusually clear. eMarketer's retail media ad spending forecast for H1 2026 puts US retail media spend at $60.32 billion in 2025, rising to $71.09 billion in 2026 — up 17.8% year over year. On a chart that looks like a rising tide.

Look at where the increment lands and the picture changes. Of roughly $10.53 billion in net-new retail media investment in 2026, eMarketer expects Amazon and Walmart to take about $9.42 billion — more than 89% of the growth. Two companies absorb nearly nine dollars in ten of the new money entering the category.

If you run a retail media network at any other retailer, that is the number that describes your year. Category growth is not your growth. You are competing for the remaining sliver, against every other network with a loyalty file and a display unit, and the advertisers you are pitching have already spent their easy budget on the two endemic platforms where their products are literally sold.

The instinctive response is to add formats: off-site, connected TV, in-store screens, a retail-powered social buy. Those are all legitimate. They are also all things the big two do at greater scale with better data. Adding a format you are structurally worse at is not a strategy.

🔍 Two-Thirds of Retail Media Is Search, and Search Is Capped by Queries You Do Not Have

Retail media is not one business. eMarketer put US retail media search spend at $39.64 billion in 2025 — 63.6% of the category total. Sponsored products and sponsored brands, the units that sit against a query, are the bulk of what advertisers buy.

Search inventory has a hard supply function: impressions are a product of query volume. You cannot manufacture more searches for "protein bar" than your shoppers type. You can raise ad load, and every retailer eventually does, but that trades conversion against yield and there is a limit before the shelf stops looking like a shelf.

So the two-thirds of retail media demand that is easiest to sell is the two-thirds you can least expand. What is left — display, video, sponsorship, brand formats — is exactly the part where you need somewhere for a shopper to be for more than a few seconds. And that brings you to the other cap.

⏱️ Your Second Cap Is Called "Session End"

Every impression your on-site network can serve is bounded by sessions × pages × ad load. You know your own figure for the first two. I would not trust the published benchmarks: across the studies I looked at, average pages per session for retail sites ranges from roughly 1.5 to over 4, and average session duration lands somewhere around two minutes, with the numbers varying enough between methodologies that they are useless for planning. Use your own analytics; treat any vendor benchmark as a conversation opener.

What matters is the shape, not the value. Sessions arrive when someone wants to buy something. That is one to four visits a month for most non-grocery retailers, concentrated in a few weeks of the year, and it is a number your merchandising team is already fighting for with email, paid search and price.

An ad tech investment does not change that number. A better ad server, a cleaner identity graph, a curated PMP — all of it divides the same sessions more efficiently. Useful work. It is not supply growth.

🎮 A Games Shelf Is On-Site Inventory That Makes New Sessions Rather Than Splitting Old Ones

Here is the argument in one line: a licensed casual game catalogue is the cheapest way to give a shopper a reason to open your app or site on a day they are not buying anything.

This is not theoretical. Walmart has been building gaming into its engagement stack for two years — its Roblox presence added a Discovered Pass in May 2025, pitched as the first branded game pass on the platform. Loyalty-app operators outside retail proper have done the same thing more plainly: Fetch and GasBuddy both bolted a games section onto a rewards app, as adjoe's roundup of retail loyalty programmes documents. The mechanic is old and well understood — the app has a reason to be opened between transactions.

What a games shelf changes on the supply side:

• Session count goes up without a media spend. The visit is triggered by the game, not by an intent to purchase.
• Session length goes up sharply. A casual game session is minutes, not the seconds a category page gets. That is video and rewarded inventory you did not have.
• Dayparting flattens. Games get played in evenings and on weekends, which is exactly when your commerce traffic thins and your CPMs are softest.
• You get brand-safe context you own. Unlike an off-site buy, the surrounding page is yours, and so is the first-party signal from who played.
• It is defensible. Amazon and Walmart's advantage is transactional data. Attention on a non-transactional day is a different product, and scale does not automatically win it.

None of this requires a games studio. A catalogue of licensed HTML5 titles drops into a route on your existing web property or a webview in your app, runs client-side, and does not touch checkout.

🧾 Where the Ads Actually Sit, and Why That Detail Decides the Build

A licensed HTML5 game is a build you run, not a build you edit. Practically every monetisable surface therefore lives in the page around the game, not inside it: the interstitial before play, the sponsorship frame around the canvas, the banner under it, the branded tile on the games index, the takeover of the category shelf that lists them.

That is good news for a retail media network, because those are the units your ad stack already serves and your rate card already prices. It is bad news if your plan depended on a rewarded video that grants an in-game reward — that needs a callback into the game, which needs either a licensor who supports a rewarded-ad hook or source-level access. Settle it in the licence, before the media plan is sold.

The other thing to settle is what a sponsor is allowed to do. A brand paying to own the games section for a month will ask for its logo on the shell, a custom skin, or a title chosen for thematic fit. Whether you may apply a sponsor's branding around a licensed title is a term, not a courtesy, and it is far easier to negotiate up front than to retro-fit against a signed insertion order.

📏 Measure It Like Media, Price It Like Media

Do not sell game-adjacent inventory on the same basis as sponsored products. It will lose, and it should — the shopper is not in a buying state. Sell it as reach and attention against a first-party audience, and measure it under the standards the buy-side already accepts.

The reference documents exist. The IAB/MRC Retail Media Measurement Guidelines, first published in January 2024, set out how retail media impressions, attribution windows and reporting should be handled; the MRC display standard behind them is the familiar one — 50% of pixels in view for one continuous second. IAB and IAB Europe followed with In-Store Retail Media Definitions and Measurement Standards in December 2024 and a framework for maturing in-store measurement in December 2025. If your games inventory reports as viewable impressions and completed video under those definitions, an agency planner can buy it without an argument.

One honest caveat on attribution: a game play will show weak last-click sales lift, and any RMN that quietly reports otherwise is doing itself harm. Report it as an upper-funnel product with an audience-based measure, or hold it back until you can.

🚫 Five Ways Retailers Get This Wrong

• Launching supply into a demand shortfall. If your network cannot sell out the display inventory it already has, more inventory is worth nothing. Fix demand first. This is the single most common reason a games shelf fails, and it has nothing to do with the games.
• Burying it three taps deep. A games section reachable only from a footer link gets the traffic a footer link gets. It needs a nav slot and at least one lifecycle message.
• Ten titles and no rotation. A fixed set decays fast. Plan the refresh cadence in the licence term, not as a later request.
• Selling the launch as a sales-lift product. You will fail the post-campaign report and lose the second flight. Pitch reach, frequency and dwell.
• Ignoring the kids question. Casual catalogues skew young. If under-13s can reach the games, your consent, data and ad-personalisation rules change materially — decide the age gate before launch, not after a complaint.

🧭 Where a Games Shelf Is the Wrong Answer

Grocery and pharmacy retailers with genuinely high purchase frequency already get the sessions; their gap is usually ad-load discipline and better search relevance, not new surfaces. A pure marketplace with heavy seller-funded search demand should spend its next quarter on auction mechanics. And if you need one high-production, mechanic-specific experience tied to a single campaign, a commissioned build from a studio serves you better than a catalogue — you are buying bespoke, not breadth. Catalogue licensing wins on breadth, refresh and multi-surface reuse, and you should not buy it for anything else.

📋 Questions to Put to a Licensor Before You Sign

1. How many unique titles, and how many are mobile-portrait playable without a rotate prompt?
2. Are HTML5 and Android builds available under one agreement, so an app team and a web team are not running two procurements?
3. Can we host the files ourselves, and are asset paths relative enough to allow it?
4. What ad hooks exist — pre-roll, rewarded callback, in-game placement — and which need source access?
5. May we apply sponsor branding around a title, and does that right extend to a third party paying us?
6. What refresh cadence is included in the term, and who selects?
7. What metadata arrives with the builds: category, age suitability, orientation, file size, language?
8. Does the licence cover both our website and our app, in every territory we operate in?
9. What happens to hosted titles at the end of the term?

🎯 What You Get When You License HTML5 Games From a Direct Licensor

A direct licence is a commercial agreement over a catalogue you deploy, not a per-file purchase. Forestry Games has licensed games since 2017 and holds a catalogue of 1,049 titles spanning HTML5 and Android APK builds, with source available on applicable titles, so a retail media team can put a games section on the website and inside the app under one conversation rather than two. Hosting can sit with you or with the licensor, branding can be applied around titles where the terms allow, and a white-label portal is available if you would rather not build the shell. For a network sizing its first games surface, the practical starting point is a shortlist filtered by orientation, language and file size — you can browse the catalogue or ask for a licence scope covering both your web and app properties.

🧸 Licensing Branded Games for Retail Campaigns, Apps and Portals

Sponsors ask for recognisable characters, and a generic catalogue cannot always answer that. Forestry Games works with branded IP and has brand partnerships including Disney, Nickelodeon, Cartoon Network and Warner Bros, and businesses can license branded game content through it for campaigns, portals, events and apps. For a retailer, that is the difference between selling a games sponsorship as a banner package and selling it as a named seasonal activation. If a branded tentpole is on your calendar, start the scoping conversation early — branded licences carry approval steps that generic ones do not. You can review the games built for marketing campaigns or request a portal demo to see how a sponsored shell is assembled.

✅ Do the Session Arithmetic This Week

Open your analytics and write down three numbers: monthly sessions, average pages per session, and the share of your on-site display inventory that went unsold last quarter. If the third number is large, your problem is demand and a games catalogue will not save you. If it is small and the first two are flat, you have a supply problem, and there is no ad tech purchase that fixes a supply problem.

The next move is a scoped pilot, not a platform decision: thirty to fifty licensed titles on one surface, a nav entry, one lifecycle message, ninety days, measured on incremental sessions and viewable impressions rather than sales lift. That is small enough to run inside a quarter and specific enough to argue about afterwards. Get a title manifest with orientation and language fields, filter it against your own audience, and license HTML5 games for the surface you can measure first.