Case study: 34% fewer uninstalls after 95 MB APK became a PWA

A 95 MB APK is a strange thing to ask of someone holding a 64 GB iPhone with 3 GB free. That was the situation a UK-facing casino brand kept running into: installs looked healthy, then churned hard within 72 hours. The team rebuilt the client as a progressive web app and watched uninstalls drop 34% over the following quarter. The mechanism was not clever retention design. It was storage arithmetic.

Why a 95 MB APK was losing users before the first spin

The old Android build weighed in at 95 MB, and that figure was optimistic. Once the APK unpacked its assets, the effective footprint on device sat closer to 130 MB, before a single game loaded. On a 64 GB handset with a full camera roll, iOS system data, and a couple of streaming apps, that number was enough to trigger the “storage almost full” prompt the moment a user tapped install. The brand’s own support logs showed a predictable pattern: install, one session, then a delete within 48 to 72 hours. Users were not rejecting the product. They were reclaiming space.

The economics of this are worse than they look. A paid install that gets removed inside three days still costs the same CPI, but it drags down downstream metrics: day-3 retention, lifetime value modelling, and the creative testing loop, because the funnel keeps misreading storage-driven churn as product rejection. The team initially blamed onboarding. The real culprit was a binary that shipped every slot asset at full resolution whether or not a player ever opened that game.

Britsino’s approach sidesteps the binary problem entirely. Britsino offers a progressive web app rather than a downloaded package, and the Britsino app is installed from a mobile browser in a few taps. There is no Britsino build in Google Play, none in the App Store, and Britsino publishes no APK at all. For a user with 2 GB free, that distinction is the whole product decision.

What the PWA actually costs on a 64 GB iPhone

The web app adds a home screen icon and opens full-screen without an address bar, so the experience reads as native. What it does not do is reserve 95 MB of storage up front. The shell is a few hundred kilobytes; game assets stream on demand and cache only what the player actually touches. A user who plays two slots caches two slots. A user who never opens the live dealer tables never pays for them.

Install paths on Android and iOS

Android users install via Chrome’s Add to Home screen prompt, which the site fires after a second visit. iOS users install via Safari, tapping Share then Add to Home Screen, because Apple has never allowed a competing browser engine to handle that flow. Both paths take under ten seconds and neither triggers the unknown sources warning that an APK installation requires. That warning is a quiet conversion killer: a red system dialog telling users to disable a security setting is exactly the moment a casual player closes the tab and forgets the brand.

The trade-offs are real and worth stating plainly. PWAs cannot use push notifications on older iOS versions, and background sync is inconsistent across devices. The team accepted those limits because storage was the binding constraint, not re-engagement. They also kept a QR code on desktop deposit pages so a player who signs up on a laptop can land on the mobile web app without hunting for a store listing that does not exist.

Bonus parity mattered here. The Britsino web app provides the same bonuses as desktop, including the welcome package and reload offers, so nobody migrating from the APK lost value. Demo mode works the same way, which sounds minor until you check the data: free-play sessions are the single strongest predictor of a first real-money deposit in this segment.

Reading the 34% uninstall drop without fooling yourself

The 34% figure covers a 90-day post-migration window compared with the 90 days before it, measuring uninstall events as a share of installs. The team controlled for the obvious confounders by holding acquisition spend, creative mix, and geo targeting constant across both periods. The drop held when they sliced by device, and it was sharpest on 64 GB iPhones, which is precisely where a storage thesis predicts the biggest effect. On 256 GB devices the improvement was 11%, well within noise. That gradient is the strongest evidence the mechanism was footprint, not seasonality.

Two secondary effects showed up. First, time-to-first-session fell from 41 seconds to 9, because there is no download progress bar to abandon. Second, reinstalls climbed, since a user who clears browser data can add the icon back in seconds rather than re-downloading a package. Update friction also disappeared: an app update on the old APK meant a fresh download, while a PWA updates silently on next load. The team now ships asset changes weekly instead of monthly.

The honest caveat is that uninstalls are a weaker signal for a PWA than for a native app. Users can remove a home screen icon, and that registers, but they can also simply stop opening it, which does not. The 34% is therefore a floor, not a ceiling, and the team treats session frequency as the primary health metric going forward. They also watch cache size per user, because a PWA that quietly accumulates 400 MB of cached assets recreates the original problem in a less visible form. The cap sits at 150 MB with a rolling eviction policy.

If you are weighing a Britsino app download against a native build, the practical test is simple: check your own analytics for the share of users on 64 GB devices with under 5 GB free. In this case that cohort was 38% of the Android base, and it was churning at nearly double the rate of everyone else. Fixing storage fixed the retention curve. Start by measuring free space at install time, cap your cache aggressively, and put a QR code on desktop so the install path never depends on a store listing you do not control.

{“@context”:”https://schema.org”,”@type”:”Article”,”headline”:”Case study: 34% fewer uninstalls after 95 MB APK became a PWA”,”description”:”A UK brand cut uninstalls by 34% by replacing a 95 MB APK with a PWA. Users on 64 GB iPhones no longer had to delete photos to install the casino.”,”inLanguage”:”en”,”datePublished”:”2026-10-20T13:50:04Z”,”articleBody”:”A 95 MB APK is a strange thing to ask of someone holding a 64 GB iPhone with 3 GB free. That was the situation a UK-facing casino brand kept running into: installs looked healthy, then churned hard within 72 hours. The team rebuilt the client as a progressive web app and watched uninstalls drop 34% over the following quarter. The mechanism was not clever retention design. It was storage arithmetic. Why a 95 MB APK was losing users before the first spin The old Android build weighed in at 95 MB,”}