The criterion is whether the user can easily reverse the choice

What it is

The line between persuasion and manipulation is judged by three parallel criteria — whether the influence serves the user's own goal, whether information is truthful and complete, and whether the user can easily reverse the choice. Reversibility asks about what happens after consent: once the influence has taken effect, what does it cost the user to change their mind? Reversibility is the test, and asymmetry of exit is the evidence — one click to get in, five steps to get out; the asymmetry itself is structural proof of manipulation, however compliant each step looks in isolation. The observable is not "does an option exist" (formally it always does) but the real friction of exercising it. Reversibility downgrades influence to advice: a reversible influence permits trial and error, and the user can say no at any moment at low cost; an irreversible one is a one-time transfer of control.

Why it happens

Exit cost turns consent into an irrevocable transfer: when changing one's mind is expensive enough, agreement stops being an evaluation of the current option and becomes the disposal of one's future self — at which point the "freedom of choice" shown on the interface is formal, every step compliant and every option present, while the effective freedom of the decision is set by the friction of the reverse path. Friction asymmetry counts as structural evidence because it exposes the design's real expectation: friction is dismantled on the way in to close the deal and piled up on the way out because the designer expects the user to regret — the difference between the two directions is a bet placed on the probability that users change their minds, one that can be neither removed nor hidden. The mechanism by which reversibility downgrades influence to advice is trial and error: with advice, accepting and refusing cost about the same, mistakes get corrected, and the long-run net effect of the influence converges toward what the user's own filtering would have produced; when exit costs exceed the value of the information gained, trial and error is impossible and consent becomes a single, unverifiable bet. Each step of a retention flow also adds consistency pressure on the psychological layer, so "easily" fails there too — that mechanism belongs to commitment and consistency; its consequence here is one line: retention interstitials are the price tag on exit friction.

Where it stops holding

The criterion demands cost proportional to real constraints, not zero friction: account deletion involves statutory data retention, refunds involve services already consumed — those steps carry real cost; steps beyond the real constraint, piled on in the name of compliance, fail the test all the same. Symmetry does not mean identical friction for all operations — high-destructive actions such as deleting all data deserve friction on the way in as well; what the criterion compares is whether comparable operations are reachable at comparable cost in both directions, with sign-up/cancellation and save/delete forming their own pairs. The criterion audits only what the interface controls: part of lock-in lives at the ecosystem level — social ties, migration cost, data accumulated across products — and no interface symmetry dissolves it, so audit findings should state this residual explicitly.

Applying it

• Run a symmetry audit: decompose the entry path and the exit path into steps, and build a comparison table of step counts, waiting periods, degree of concealment (is the entry reachable, is phoning support the only route), and number of retention interstitials — asymmetry beyond the same order of magnitude gets fixed.
• Bring automatic renewal up to the symmetric standard: a proactive reminder before renewal (with amount and date), one click either to renew or to stop, and a no-questions cooling-off window after renewal.
• Keep exit entries reachable at the same level as entry: in settings, "turn off / cancel / delete account" appears alongside "subscribe / upgrade", and "contact support" is never the only exit channel in a digital product.
• Set retention limits in advance: the number of interstitials and the intensity of the copy are fixed at design review; exceeding them in production counts as obstruction, not reminders.
• To validate: screen-record and time it — how many seconds and clicks a new user needs to subscribe versus to cancel, and how many retention layers appear in between; put the two numbers side by side in the review deck and the asymmetry speaks for itself.

Related

• Same group: P2.08.1 The criterion is whether the influence serves the user's own goal · P2.08.2 The criterion is whether information is truthful and complete
• Nearby: P3.05 The freedom to exit · P3.10 The freedom to leave · P2.06 Commitment and consistency
• Search terms: reversibility · asymmetry of exit · roach motel