Somewhere right now, a person is trying to cancel something. They signed up in forty seconds, with a thumb, half-watching television. Canceling is taking longer. There is a page that asks if they’re sure, then a page that asks why, then an offer of a discount that somehow never appeared while they were a paying customer, then a phone number, staffed weekdays until four, for a service they joined at midnight on a Sunday.
Every product has this screen. Not just the cancellation flow — the privacy settings buried three menus deep, the data export that arrives as an unusable file format, the deletion confirmation that warns you, at unusual length, about everything you’re about to lose. These are the screens where the company’s interests and the user’s visibly diverge, and they are, as a result, the most honest interfaces the company ships. Everything else is a stage set.
The mechanism is simple enough that it’s strange we don’t talk about it more plainly. Marketing pages are written for the buyer, and the buyer is hypothetical — a person being wooed, whose goodwill is worth spending money to win. The person trying to leave is already lost. Their goodwill has no resale value. So the welcome flow gets the senior designer, the copywriter, the A/B tests, the animated confetti. The exit flow gets whatever was left over, or worse: it gets deliberate attention of a different kind, from people whose job is to make leaving feel like wading. Harry Brignull coined the term “dark patterns” back in 2010 to describe interfaces designed to trick, and his taxonomy has a name for this particular one — the roach motel. Easy to check in. You know the rest.
What makes the worst screen interesting, rather than merely irritating, is that it can’t be excused as an accident. A confusing settings page might be neglect. But a cancellation flow that offers you a discounted rate the moment you try to quit is proof the company knew the price was negotiable all along and chose to charge the loyal ones more. The obstacle course isn’t a failure of design; it’s a successful design aimed at you. The FTC spent years assembling evidence of exactly this — its 2022 report on dark patterns reads like a catalogue of screens everyone recognizes and nobody will admit to building — and when the agency finally tried to require that canceling be as easy as signing up, industry groups fought the rule into the ground. A federal court vacated it on procedural grounds in 2025, before it ever took effect. The market’s position, stated plainly, was that making leaving easy is an unreasonable burden. Consider what that admits about the value of staying.
There is a psychology underneath this that goes beyond any single company. Daniel Kahneman’s work on the peak-end rule showed that people judge experiences largely by how they end — a colonoscopy study, of all things, established that a longer procedure with a gentler finish was remembered as better than a shorter one that ended abruptly. Companies have absorbed this lesson in reverse. They cannot make you love the product, but they can make the ending so unpleasant that you never come back, and they apparently calculate that a retained-but-resentful subscriber is worth more than a departed-but-grateful one. Sometimes they’re right, in the narrow sense. The resentment just gets outsourced to the culture, where it compounds into the ambient feeling that everything digital is a small scam, that every free trial is a trap, that you should never enter your card number anywhere you can’t imagine escaping from.
That ambient suspicion is the real cost, and it doesn’t appear on anyone’s churn dashboard. When enough companies treat departure as a threat, users start treating commitment as a risk. People keep spreadsheets of their subscriptions. They set calendar reminders for trial expirations. They use virtual card numbers with spending limits, an entire product category that exists solely because trust doesn’t. The dishonest interface doesn’t just damage its own maker; it raises the transaction costs of the whole economy, one defensive habit at a time.
Which is why the corollary matters more than the complaint. A company that makes leaving genuinely easy — one click, no guilt trip, no save-offer gauntlet, your data exportable in a format that opens — is doing something expensive and legible. It is giving up revenue it could have kept through friction, and it is doing so in the one place where no one would have blamed it for the friction. That is a statement about confidence: we believe you’ll stay for the product, not the maze. It is also, not coincidentally, the most credible marketing the company will ever produce, because it’s the only claim that costs money to make. Anyone can write “we respect our users” on a landing page. The cancellation flow is where you find out whether it’s true.
So here is a practical method, free of charge. Before you subscribe to anything, before you trust anything, find the worst screen. Don’t read the homepage; go straight to the help article titled “How do I cancel?” and follow it like a detective. Count the steps. Read the tone. A company can rent a beautiful front door, but the back exit is load-bearing — it has to be built out of what the company actually is.
The person from the first paragraph eventually cancels, probably. Most people do, in the end; the maze delays rather than prevents. But they leave with a story, and the story is never about the product. It’s about the screen where the mask came off. Companies spend fortunes managing their first impression and then file their true face in a menu three levels deep, as if no one would think to look. Look. It’s the last honest interface most of them have.