An OEM Preload Slot Holds One App. Your Thousand-Title Catalogue Does Not Fit.

An OEM preload slot holds one Android app, not a catalogue. That decides what a preinstall deal is worth and what belongs in it. Everything else is logistics.

Preinstall pitches arrive with a big number attached. Hundreds of millions of devices. Whole-country reach. A distribution channel that skips the app store queue entirely and puts your game on the phone before the buyer has finished peeling the screen protector off.

The number is usually real. The thing people get wrong is what unit it applies to. A device maker is not offering to carry your catalogue. It is offering one line in a setup screen, or one entry in a system image, and it is offering it to everyone else too. If you own a thousand Android titles, the preload channel does not scale with your catalogue at all. It scales with how good your single best app is.

That reframing changes the entire commercial conversation, so it is worth walking through what these deals actually are before quoting one.

🏭 Two Different Things Both Get Called a Preinstall

They have almost nothing in common operationally, and confusing them is how projects lose a quarter.

A factory preload is an app baked into the system image. It ships on the /system partition, it occupies fixed storage on every unit whether the owner wants it or not, and it exists because a firmware build was cut with it inside. Changing it means cutting another firmware build.

A dynamic preload — on Google-certified devices, usually Play Auto Install — works differently. As Jason Bayton's breakdown of PAI explains, it is a partner service that lets an OEM configure which apps a device pulls down, without those apps being in the Android build at all. The OEM ships a small system app carrying a default configuration and then manages the rest through Google's device configuration portal, targeting by region, carrier, device build fingerprint or custom OEM properties. New configurations deploy without an OS rebuild.

Read the consequence rather than the mechanism. On a factory preload, your release calendar is the OEM's firmware calendar. On a dynamic preload, it is a portal change. That is the difference between a channel you can iterate on and a channel you can only launch into once.

There is a third variant worth knowing: out-of-box placements delivered through the OEM's own store rather than Google Play. Those matter for categories Play restricts, and they run on the manufacturer's own distribution rails.

💳 What You Pay For, and Which Event Triggers It

Traditional factory preloads are typically sold against a minimum annual commitment — you buy device volume in advance, and the volume commitment is what buys the rate down. You are paying for placement on units, not for humans.

Dynamic preloads are normally sold on cost per install, and the install is an opt-in: the user sees a curated list during setup and chooses. Fewer devices, better devices.

The detail that should decide your model is when the meter reads. Per Adjust's PAI measurement documentation, preload attribution runs through the Google Play Install Referrer API, and the install is recorded when a user launches the preinstalled app for the first time — not when the device is activated. The referrer arrives carrying utm_source=play-auto-installs and utm_medium=preload, and Adjust lists Oppo, Transsion, vivo and Xiaomi among supported partners.

First launch is the honest event. Insist on it. A deal priced per shipped device pays for phones in boxes; a deal priced on first open pays for people who at least got as far as tapping the icon. If a partner will only sell per-device, that is not automatically a bad deal, but it means you are buying inventory risk, and it should be priced like inventory risk.

🔢 The Slot Is Singular, So Pick the Right Occupant

Here is the part licensing buyers consistently mis-plan. You cannot preload a catalogue. You get an app.

Which means the strategic question is not "how many of my titles can I get on this device", it is "what single artefact do I want sitting on ten million home screens". Realistically there are three answers:

• Your single strongest title. Simple, measurable, and it inherits all the retention problems of a one-game app. When the player is done with it, the slot is dead.
• One hub app containing many games. The slot holds one icon; behind it sits a catalogue. Retention comes from breadth rather than from any one loop, and you can add titles after the device has shipped.
• An operator or OEM surface you supply rather than own. The manufacturer or carrier ships their own games destination; you are the catalogue behind it. Less brand equity, far less negotiation.

The hub option is worth dwelling on, because it lines up with a constraint you already have elsewhere. Google Play's spam policy explicitly pushes developers with many small, similar apps toward aggregating them into one — the same structural answer, arrived at from a different direction. If you are already consolidating a portfolio of licensed Android games for policy reasons, that consolidated app is also the only version of your catalogue that fits a preload slot.

And a hub does not have to be a heavy native build. A shell wrapping a well-optimised HTML5 catalogue updates its content without a store release, which matters when your only lever on a shipped device is what the app fetches next.

🌍 Where the Devices Actually Are, and How Much Trackers Disagree

Preload economics live or die on device volume, so use current shipment data rather than the deck you were handed.

The trackers do not agree on the size of the contraction, and the gap is instructive. Counterpoint Research put Q2 2026 global smartphone shipments down 11% year over year — the weakest second quarter since 2013 — with Samsung at 24% share, Apple 20%, Xiaomi 12%, OPPO 11% and vivo 8%. Omdia measured the same quarter at 272 million units, down 6%, with Samsung at 22% and Xiaomi at 11%. Both agree on direction and on which vendors took the damage; they disagree on magnitude by roughly a factor of two. Treat any preload volume forecast built on a single tracker with suspicion.

Regional concentration is where this channel earns its reputation. Canalys data for Q1 2025 put the three Transsion brands — TECNO, itel and Infinix — above 46% of African smartphone shipments combined. One group of relationships covers close to half a continent's new devices. Nothing in Western Europe looks remotely like that.

Vendor-published reach figures point the same way, with the usual caveat that they are marketing material: the OEM advertising firm AVOW's 2026 OEM advertising guide claims roughly 86% coverage of Android devices across its partner set, with reach it puts at 91% in Africa, 87% in Southeast Asia and 84% in India, against 57% in Europe and 55% in LATAM. Those are the firm's own numbers, not an independent audit. But the shape — emerging markets far ahead of Western Europe — matches what the shipment trackers show, and it is the shape that should drive where you spend.

The same guide cites large active-user bases for the manufacturer app stores themselves: Samsung's Galaxy Store, Huawei's AppGallery and Xiaomi's GetApps. Again, vendor figures. But if you are licensing APKs, the manufacturer store is a distribution surface you can pursue independently of any preload deal, and it is chronically under-worked by Western publishers.

⚖️ The Legal Layer Moved Under This Channel

Two regulatory shifts changed what a preload slot is and who can pay for one.

The first is the Epic v. Google injunction. Under provisions that took effect in August 2025 and run for three years to 1 November 2027, Google may not condition a payment, revenue share, or access to its products on an OEM or carrier agreement to preinstall the Play Store in any specific device location, and may not pay developers to launch exclusively on Play. That is a direct constraint on the biggest cheque in the preinstall market. It does not hand you free placement, but it does mean the manufacturer's negotiating position is no longer propped up by an exclusivity payment it can lose.

The second is the EU's Digital Markets Act. Article 6(3) requires gatekeepers to let users easily uninstall any application on the operating system, excepting only software essential to the functioning of the OS or device that cannot technically be offered standalone by a third party. Your game is not essential to the functioning of the device. In the EU, a preload is a suggestion.

Combine those with PAI's own behaviour — dynamic preloads are ordinary installs that a user can remove — and the durable-placement version of this channel has mostly stopped existing. What is left is a very good top-of-funnel buy, which is a different product.

🗑️ The Uninstall Is the Number That Decides the Deal

Since placement is removable and CPI fires on first launch, everything after that first launch is your problem.

Academic work on Android usage has flagged the risk for years: a 2017 study of device-specific app usage patterns across large-scale Android user data found that manufacturer-customised and preloaded apps were more likely to be uninstalled on lower-end devices, and that lower-end users were less likely to adopt them in the first place. That study is old enough that you should not lean on its specifics, and the market has moved. The direction still matters, because it points precisely at the devices where preload reach is highest and CPI is lowest.

So instrument the two numbers that actually settle whether the channel worked: the share of preload installs that produce a session at all, and day-7 retention of preload cohorts against your paid-UA cohorts. If preload users retain materially worse, a lower CPI has not bought you a cheaper user — it has bought you a cheaper number.

🚫 Five Ways Preload Deals Waste the Budget

1. Buying reach when the catalogue has no hub. Signing device volume before deciding what occupies the slot leaves you shipping whichever title was closest to finished.
2. Pricing per device instead of per first launch. On a factory preload with an annual minimum, you carry all the risk that the phones sit in a warehouse.
3. Ignoring firmware lead times. A factory preload tied to a marketing date only works if the build cycle was in the plan from the start. Dynamic placement is the answer when the date is fixed.
4. Treating EU reach as equivalent to emerging-market reach. Both the shipment trackers and the vendors' own coverage claims say it is not, and Article 6(3) makes an EU preload more fragile besides.
5. Reporting installs to the board. Preload installs are the cheapest impressive chart in mobile. Retention curves are the only version of the story that survives quarter two.

🎮 Where a Licensed Catalogue Fits

If the slot holds one app and that app needs depth behind it, the constraint is content volume you can refresh without shipping a build. That is the ordinary case for licensing rather than building: Forestry Games has licensed games since 2017 and carries a catalogue of 1,049 titles spanning HTML5 and Android APK, which is the raw material for a hub app rather than a substitute for one. The company also works with branded IP and has brand partnerships including Disney, Nickelodeon, Cartoon Network and Warner Bros.

Depending on which side of the deal you are on, that means the catalogue as content for your own hub, source-code licences where you need to own and modify the build, or a carrier and operator games portal where the manufacturer or network owns the surface and you supply what runs inside it.

🧭 Before You Sign a Preload Deal

Do these four things in order, and most bad preload contracts stop being tempting.

1. Decide the occupant first. Single title, hub app, or supplying someone else's surface. Nothing else can be priced until this is settled.
2. Ask which mechanism is being sold. Factory image or dynamic configuration. If the answer is vague, it is a factory image and the calendar is not yours.
3. Move the trigger to first launch. Get the Install Referrer attribution wired and agreed before signing, not during the first invoice dispute.
4. Pick the region on shipment data, not on the deck. Two trackers, not one, and weight the markets where a single group of relationships covers a large share of new devices.

Preinstalls are a legitimately strong channel for Android games in exactly the markets where paid user acquisition is worst value. They are simply not a catalogue distribution channel, and the operators who do well out of them are the ones who worked that out before signing, built the one app worth preloading, and then measured what happened on day seven rather than day zero.