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.

Lazyweb Research · n=332 · Published 2026-07-07

cancellationwebmobilesaasretention

Two universes, kept separate

UniverseScreensBacking
App (canonical Cancel-Subscription)3321,685 screenshots, 81 apps [1][3]
Web (in-product cancellation)135all 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 Do not quote an app-derived save-offer rate as if it applied to web, or vice versa — the samples are separate by design.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

StatComputed from
332 canonical Cancel-Subscription app screenscancel_canonical_screens
135 in-product web cancellation screens, all flagged in-productweb_cancellation_screens
1,685 screenshots across 81 apps back the app corpuscancel_screenshots_and_apps
48-app / 401-screenshot genuine-cancel-intent subset for tactic ratesgenuine_cancel_intent_subset
Methodology. Universes: 332 canonical app cancel screens (1,685 screenshots, 81 apps) and 135 labeled in-product web cancellation screens, July 2026. Method: report each universe separately with its own denominator. Caveat: the two are not pooled; the app category is partially paywall-polluted, so tactic rates use a 48-app clean subset.

Sources & citations

  1. [1] Lazyweb Research analysis of 332 canonical cancel screens (app corpus), July 2026. App-side Cancel-Subscription canonical screen count.
  2. [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. [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.

Related questions

Explore the underlying screens, flows, and A/B tests inside Lazyweb. More research