Skip to content
The Product Guys
All patterns
Money6 min read

The Cancellation Flow

The path out of a subscription, and the clearest test of whether a company means it.

The problem it solves

Every subscription business eventually faces someone who wants to leave. The flow can save a few of them honestly, or it can hold the rest hostage and burn the brand to hit a quarterly number.

Steps added between the intent to cancel and the cancellation

% of people who set out to cancelOpens billing settings100Finds the cancel link84 (-16)Past the loss-framing screen71 (-13)Past three retention offers58 (-13)Completes cancellation52 (-6)
Each extra screen keeps a few more subscribers and converts more of them into people who dispute the charge instead. Saves rise while the relationship is spent.

This is where otherwise decent companies behave worst. The revenue saved by obstruction is visible this quarter and the damage is diffuse, so the incentive points one way. Resist it deliberately, because nobody drifts into a good cancellation flow.

The honest version

  1. 1Put Cancel subscription in the billing settings, at the same depth as the upgrade button.
  2. 2Confirm once, on one screen, stating the date access ends and what happens to their data.
  3. 3Ask one optional question about why they are leaving, with a skip that works.
  4. 4Offer at most one alternative, and only one that genuinely fits: pause for three months, or a cheaper tier.
  5. 5Cancel it, send a confirmation email with the end date, and leave the account readable until that date.

A pause offer is the one save worth building. Some cancellations are seasonal rather than final, and pausing keeps the data, the settings, and the door open without pretending the person did not ask to leave. Discount offers work too, though they tell every customer who hears about them that the list price was negotiable.

Sludge, whatever it is called internally

  • Call us between 9am and 5pm to cancel
  • Are you sure? You will lose all your data, your history, and your progress
  • Six screens of offers before the cancel button appears
  • Cancel request received. We will process it within 7 business days

The version that keeps the relationship

  • Cancel subscription
  • Your plan ends on 14 October. Your data stays available to download until 14 January
  • One screen: Cancel now, or Pause for 3 months
  • Cancelled. Confirmation sent to [email protected]. Nothing further will be charged

Where this becomes a legal problem

Obstructed cancellation is now regulated, not merely frowned upon. California's automatic renewal law requires that a subscription bought online can be cancelled online. The US Federal Trade Commission has pursued companies over cancellation obstruction under its enforcement of negative-option rules. The EU Consumer Rights Directive and the UK's Digital Markets, Competition and Consumers Act both push in the same direction: the exit must be as easy as the entrance. Teams should treat parity between signup and cancellation as a requirement rather than a design preference.

If the retention offer only works because the person could not find the exit, it is not a retention offer.
A useful internal rule

Measure the flow on refunds, chargebacks, and win-back rate rather than on saves. A flow optimised purely for saves will find its way to obstruction within two quarters, because that is what the metric rewards.

When it fits

  • You have a genuine alternative to offer, such as a pause or a lower tier that fits the stated reason.
  • The cancellation reason can be captured in one optional question and fed back to the product team.
  • Downgrade and pause can be executed by the product without a support ticket.
  • You are prepared to be judged on refunds and chargebacks rather than on saves.

When it backfires

  • Cancelling requires a phone call, a chat queue, or an email while signing up took two clicks.
  • The flow stacks multiple offers, which raises complaint volume and, in several jurisdictions, regulatory exposure.
  • The confirmation screen overstates the loss, such as threatening deletion of data that is retained anyway.
  • The team is measured on saves, which makes obstruction the rational local choice for whoever owns the flow.

Products using it

  • Netflix

    Cancels in a few clicks from account settings and keeps the profile and viewing history for a period, so returning is easy.

  • Amazon Prime

    Was the subject of an FTC complaint over the design of its cancellation path, and subsequently simplified it.

  • Spotify

    Cancels from the web account page and keeps playlists intact on the free tier, so leaving does not destroy the library.

The psychology under it

The takeaway

Make leaving exactly as easy as joining, offer one honest alternative, and judge the flow by chargebacks rather than saves.

Finished the teardown? Bank it and the day counts toward your run.