Skip to content

Roach Motel

A dark pattern that makes getting into a subscription or account easy and getting out of it deliberately hard, usually through a cancellation maze.

Growth & Metrics

A roach motel is a flow that is easy to get into and deliberately hard to get out of. Signing up takes a minute. Cancelling takes a phone call, a chat with a retention agent, five screens of offers, or a setting hidden three menus deep.

The name comes from the early deceptive design catalogues. Current versions file the same behaviour under hard to cancel and obstruction.

What it looks like in SaaS

  • No cancel option in settings. Billing shows the plan and the card, and nothing that ends either.
  • "Contact us to cancel". Self-serve signup, but cancellation only through sales or support, often during business hours in another time zone.
  • The maze. Survey, offer, second offer, "are you sure", a warning about lost data, then a confirmation link that lands back on the dashboard.
  • Downgrade as a dead end. The free plan exists but cannot be selected from inside the product.
  • Seats that only go up. Adding a seat is one click; removing one needs a ticket.

Easy exit against roach motel

Honest exitRoach motel
WhereIn account settings, where billing livesHidden, or outside the product
ChannelSame as signup, usually onlinePhone, chat or email only
StepsRoughly as many as signupMany more, with loops
Retention offerOne, relevant, easy to declineSeveral, repeated
ConfirmationImmediate, plus emailUnclear whether it worked

Why teams build it

Churn is a board-level number and the cancellation flow is the one place where design can move it overnight. Adding friction reduces completed cancellations, and the report shows lower churn the same month.

Some of that friction is defensible. A short question about why someone is leaving, and a pause option for seasonal teams, keep customers who would have stayed with a little help. The roach motel is what happens when the friction keeps growing because every additional step still moves the number.

Why it backfires

It does not keep customers, only payments. Someone stuck in a subscription they want to leave is not retained in any useful sense. They do not use the product, they do not expand, and they leave the moment they find the exit.

It moves the cancellation elsewhere. Customers who cannot leave through the product go through their bank, finance team or procurement. Chargebacks and disputed invoices cost more to handle than a cancellation and do worse things to the relationship.

It is visible. Cancellation stories end up on review sites and in peer conversations, where the buyers you are trying to win read them.

It attracts enforcement. In the US, ROSCA requires sellers of online subscriptions to provide simple mechanisms to stop recurring charges. In September 2025 Amazon agreed to pay $2.5 billion to settle FTC charges that included making Prime hard to cancel. In the EU, the Digital Services Act, Article 25 lists making termination more difficult than subscribing as a practice the Commission may issue guidance on for online platforms. Scope varies and much of this is consumer law; it is a direction of travel, not a checklist.

The honest alternative

Put cancellation where billing is. Plan, seats, invoices and cancel on the same page.

Match the signup channel. If someone could start online, they can stop online.

Ask once, offer once. One optional question about why. Depending on the answer, one relevant option: pause, a smaller plan, help with setup. Then cancel.

Say what happens next. When billing stops, what happens to data, how long it is kept, how to export it.

Confirm in writing. An email that states the cancellation and the end date.

Make the reasons count. Cancellation reasons are some of the most honest research a product gets. Route them to the people who decide the roadmap.

In practice

A product has self-serve signup and cancellation by email to support. Churn looks steady, but a share of cancellations arrive as card disputes, and finance spends hours on them every month.

The redesign moves cancellation into billing settings. One screen asks why, with five options. "Too expensive" shows the smaller plan. "Seasonal" shows a three-month pause. Everything else goes straight to a confirmation screen stating the end date and the export link.

Completed cancellations rise in the first month, then settle. Disputes drop to almost nothing, pauses bring some teams back, and the reasons list gives the roadmap its clearest signal in a year.

Where teams get it wrong

  • Measuring churn without disputes. The cancellations did not disappear; they moved to the bank.
  • Stacking offers. One relevant offer is a service; four is a maze.
  • Cancel only through sales. Self-serve in, self-serve out.
  • No confirmation. People who are not sure it worked will dispute the next charge.
  • Ignoring the reasons. The most useful part of the flow, unread.

Frequently Asked Questions