Superwall's cancel-screen data shows one in five departing subscribers can be saved
Superwall shipped Customer Center on October 15, 2026, an in-app screen that asks subscribers why they are leaving before Apple's own cancellation sheet opens. It follows the first production benchmarks for Apple's Retention Messaging API, numbers its larger rival RevenueCat has not published.
Richard Tang · 3 min read
Superwall, the paywall-testing and subscription-infrastructure company behind more than 10,000 apps, shipped Customer Center on October 15, 2026 1. It is a screen inside its iOS and Android SDKs, versions 4.18.0 and 2.8.5, where a subscriber can cancel, switch plans, request a refund or restore a purchase 1 2. The commercially important part runs first: before Apple's+0.16% — Apple, up 0.16 percent today own cancellation sheet opens, a short survey asks why the person is leaving, and every answer lands in the developer's analytics 1. The cancel screen is the last moment an app still owns the customer, and Superwall has moved its fight with RevenueCat, the incumbent in subscription infrastructure, onto it.
Apple opened the cancel screen, and Superwall measured it
Apple's Retention Messaging API lets a developer answer a cancellation in progress with one of three responses: a plain message, a promotional offer or a plan switch 3. Setup guides existed, but nobody had published production results, so on September 29 Superwall did, covering every retention message it served across 30 iOS apps from August 31 to September 28 3.
Across roughly 73,000 sends, a typical app saved 13 to 21 percent of the subscribers who reached the cancel screen 3. More than 80 percent of the people on that screen canceled anyway, and no app ran a no-message control, so these are save rates rather than proven lift 3.
A plain message saves as many canceling subscribers as a discount does
Data
| Value | |
|---|---|
| Plain text message | 20.7% |
| Promotional offer | 17.7% |
| Switch-plan offer | 12.9% |
A recheck of the same subscribers a week later answers part of that doubt. Text and promo saves held up, dropping only 1.5 to 3 points 3. Switch-plan fell from 11.6 to 7.9 percent, so about a third of its saves canceled within the week 3. The best messages named exactly what the subscriber was about to lose: their scans, their history, their streak 3.
Annual subscribers almost never come back
The stakes are in RevenueCat's own research: annual reactivation sits at 5 percent 4. Once an annual subscriber walks, the app has almost no way back.
Superwall matches RevenueCat's screen and adds published save rates
Customer Center presents with one line of code and hands the billing to StoreKit, Apple's in-app purchase framework: plan changes, cancellations and refunds all open Apple's own system screens 1. App Store and web-checkout subscriptions appear side by side, with a warning for anyone paying twice 1.
RevenueCat already ships a Customer Center of its own: offers right before the final cancel button, cancellation reasons captured through feedback prompts, and refunds, plan changes and restores in one place 5. On the screen itself, the two products now overlap. The difference is published evidence: the September 29 study is, on Superwall's own account, the first public set of production save rates 3.
A private fight on ground Apple laid
Neither company trades publicly 6 9. Superwall says it is profitable on $7.5M raised 6, with more than $1.6B in annual subscription revenue running across its platform 7. RevenueCat reports over 115,000 apps and more than $16B processed a year 8, and has raised $100M, its Series C led by Bain Capital Ventures 9. Adjacent, Nico Wittenborn's fund, backs both sides 6 9, and every figure here is the company's own.
For stock investors, the listed company in this fight is Apple, whose StoreKit and Retention APIs decide what either vendor can do at the moment of cancellation 1 3. The value landed with Superwall, which ran the API in production, published what it learned and shipped the tool two weeks later 1 3. For app makers, the first finding is free to test: before offering a discount, name exactly what the subscriber is about to lose 3. Watch next whether the saves hold as more apps go live, and whether RevenueCat publishes numbers of its own 3.
Deepdive
AI-generated from this story and its cited sources. Not investment advice.
