How does cancellation-flow coverage compare between web and app?
This research tracks two separate universes: 332 canonical Cancel-Subscription app screens (1,685 screenshots, 81 apps) and 135 labeled in-product web cancellation screens [1][2][3]. All 135 web screens are flagged as real in-product cancel screens, not marketing pages [2]. Report the two universes separately — they are not a single pooled sample.
332 canonical app cancel screens vs 135 labeled in-product web cancellation screens, July 2026.
Two universes, kept separate
| Universe | Screens | Backing |
|---|---|---|
| App (canonical Cancel-Subscription) | 332 | 1,685 screenshots, 81 apps [1][3] |
| Web (in-product cancellation) | 135 | all flagged in-product [2] |
The app corpus is screenshot-backed across 81 apps; the web corpus is 135 labeled in-product cancellation screens, every one flagged is_in_product=true [1][2][3].
Why not pool them
The two are collected and labeled differently, so pooling would produce a false blended rate. The web set is deliberately filtered to real product cancel screens (not marketing pages) [2], and the app canonical category is partially polluted with paywall/win-back screenshots, which is why app-side tactic rates use the narrower 48-app genuine-cancel-intent subset [4]. Keep denominators honest by citing each universe on its own.
How to apply it
When benchmarking a web cancel flow, anchor to the 135-screen web universe; for a mobile app flow, anchor to the app corpus and its 48-app clean subset for tactic rates [1][2][4]. Do not quote an app-derived save-offer rate as if it applied to web, or vice versa — the samples are separate by design.
The numbers
| Stat | Computed from |
|---|---|
| 332 canonical Cancel-Subscription app screens | cancel_canonical_screens |
| 135 in-product web cancellation screens, all flagged in-product | web_cancellation_screens |
| 1,685 screenshots across 81 apps back the app corpus | cancel_screenshots_and_apps |
| 48-app / 401-screenshot genuine-cancel-intent subset for tactic rates | genuine_cancel_intent_subset |
Sources & citations
- [1] Lazyweb Research analysis of 332 canonical cancel screens (app corpus), July 2026. App-side Cancel-Subscription canonical screen count. ↩
- [2] Lazyweb Research analysis of 135 in-product web cancellation screens, July 2026. All 135 flagged is_in_product=true (real product cancel screens). ↩
- [3] Lazyweb Research analysis of 1,685 screenshots (81 apps), July 2026. Screenshot and app coverage backing the app canonical category. ↩
Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.