How is the cancel button styled versus the keep/subscribe path?

Across 78 cancel/keep/pause/resubscribe CTA buttons, 40 are styled as dismiss and 18 as secondary, versus only 7 primary ^[1]. So the cancel action is heavily de-emphasized relative to the primary keep/subscribe path. If you are optimizing for compliance rather than dark patterns, that skew is the thing to check first.

Of 78 cancel/keep CTAs, 40 dismiss and 18 secondary vs 7 primary styling, July 2026.

Lazyweb Research · n=78 · Published 2026-07-07

cancellationretentiondesignux-patternsexperiments

The role split

CTA role	Buttons	Distinct apps
Dismiss	40	18 ^[1]
Secondary	18	14 ^[1]
Unknown	13	— ^[1]
Primary	7	6 ^[1]

Cancel and retention actions skew heavily toward dismiss and secondary styling; only 7 of 78 buttons in this set are styled primary ^[1].

What it means

The visual hierarchy consistently pushes users toward the keep/subscribe path and mutes the cancel action into dismiss-weight styling ^[1]. This is the mechanism regulators scrutinize under click-to-cancel: if the cancel affordance is styled as an afterthought, the flow reads as obstructive even when a cancel option technically exists.

How to apply it

Audit your cancel screen's button hierarchy against this split ^[1]. A compliant flow keeps the actual cancel affordance clearly reachable rather than dismiss-weighted. The field's own experiments trend toward easing progression — DoorDash made a reason button more prominent to reduce abandonment on the cancel path, a completion-friendly direction ^[2]. Style the cancel action so a user can find and complete it without hunting.

The numbers

Stat	Computed from
78 cancel/keep CTAs: 40 dismiss (18 apps), 18 secondary (14 apps), 13 unknown, 7 primary (6 apps)	cancel_flow_cta_roles
DoorDash A/B made 'Select Reason' button solid to reduce abandonment	qualitative / doordash experiment

Methodology. Universe: 78 cancel/keep/pause/resubscribe CTA buttons in the paywall universe, July 2026. Method: count by cta_role and distinct apps. Caveat: 13 buttons have unknown role and role labels are derived from styling, not intent.

Sources & citations

[1] Lazyweb Research analysis of 78 cancel/keep/pause/resubscribe CTA buttons (paywall universe), July 2026. cta_role split for cancel-family buttons across distinct apps. ↩
[2] Lazyweb Research analysis of DoorDash cancel-flow experiment, July 2026. A/B changing reason button from grayed-out to solid to ease progression. ↩

Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.