Why Cancelling Subscriptions Is So Hard (And That's Intentional)

When Amazon's internal teams needed a name for the process they'd designed to stop Prime subscribers from leaving, they chose the Iliad. Homer's epic poem. The one about a war so long and brutal it took a decade to resolve. They named their cancellation flow after it internally, apparently without irony, and kept it running for years.

The Iliad Flow, as the Federal Trade Commission documented in exhaustive detail before Amazon settled for $2.5 billion in September 2025, required subscribers to navigate four pages, make six clicks, and choose from fifteen separate options to complete a cancellation. At each step, the interface presented discount offers, benefit reminders, and strategically designed friction, each element calibrated to produce a different kind of abandonment. Signing up for Prime, by contrast, took two clicks. The asymmetry was not accidental. Internal documents showed that whenever Amazon simplified its cancellation process, Prime sign-ups dropped, and executives rolled back the changes.

This is the story of why cancelling subscriptions is so hard. Not because the technology is complicated. Because the difficulty is a product decision.

Roach Motels and Dark Patterns

The practice has a name in interface design: the roach motel. You check in easily. Checking out is another matter. The term was coined by UX researcher Harry Brignull, who in 2010 began cataloging what he called "dark patterns" — design choices that use interface architecture to steer users toward actions that benefit the company and away from actions that benefit the user.

Brignull's initial taxonomy described a handful of patterns. The category has since expanded to cover dozens of documented techniques, from confirm-shaming (the "No thanks, I don't want to save money" cancel button) to obstruction (the cancellation option buried four submenus deep in account settings) to nagging (the repeated interstitial offers of pausing instead of canceling, presented every time a user tries to exit).

Why These Patterns Keep Working

What makes these patterns effective, and what makes them so persistent, is that they exploit the same psychological tendencies that make subscription services appealing in the first place. Canceling requires active effort and deliberate decision-making. The interface's job is to make that effort feel larger than it is. Confirmation screens that ask users to select a reason for leaving, then ask again, then offer a reduced price, then ask one final time, are not providing value. They are wearing down resolve.

A 2024 sweep by the International Consumer Protection and Enforcement Network reviewed 642 subscription websites and apps across 26 countries and found that the majority used at least one dark pattern in their cancellation flows. Eighty-one percent of sites made it impossible to turn off auto-renewal at the point of sign-up. Seventy percent provided no information about how to cancel at all.

These are design choices. Every element of a cancellation flow is designed by someone, tested against alternatives, and measured against a retention metric. The question companies ask when building these flows is not "how do we make this experience fair?" It is "at which step do the most people give up?" The answer, refined through years of A/B testing, is embedded in the product you're using right now.

The Frustration Is Familiar — Because It's Intentional

This is the moment most people feel the frustration most acutely: they've decided to cancel, they go looking for the button, and they can't find it. Or they find it, but the button leads to a form that requires them to call a phone number. Or they call the number and wait on hold long enough that the will to cancel becomes the will to hang up and deal with it later.

That "later" is worth real money to the company. Research from C+R Research found that 42% of consumers have paid for a subscription they forgot to cancel. Forgetting often isn't passive. It follows a failed attempt.

If you've ever found yourself in this situation, running a tool like Substract is one practical way to surface exactly which subscriptions are still active on your accounts, including ones you may have tried to cancel but weren't certain went through. Substract analyzes your bank statement and identifies every recurring charge in under ninety seconds. No bank login required. It doesn't replace the work of canceling, but it gives you a clear picture of what still needs to be done.

The Regulatory Picture (Not Great)

The FTC finalized its click-to-cancel rule in October 2024, which would have required companies to make cancellation as simple as enrollment. The Eighth Circuit Court of Appeals blocked it in July 2025. The FTC restarted rulemaking from scratch in January 2026. In the meantime, the Amazon settlement stands as the most consequential enforcement action to emerge from this space.

The $2.5 billion figure is notable. More notable is what the internal documents revealed: that Amazon executives knew the cancellation flow was producing unwanted subscriptions, called it a problem in writing, and kept it running anyway because the revenue outweighed the risk. One internal document described the enrollment practices as "an unspoken cancer." Another reportedly referenced an executive as the "chief dark arts officer." They were not confused about what they were doing.

How Widespread Are Dark Patterns?

The broader industry is watching. Dark pattern analysis of six major subscription services conducted between March and May 2025, covering platforms including The Economist, Bloomberg, Adobe, and Audible, found that loss aversion manipulation was present in five of the six. Emotional exploitation, hidden options, and navigation barriers appeared in four. These aren't fringe services or scam operations. They are mainstream platforms used by tens of millions of people.

A 2024 study measuring the effects of dark pattern-heavy cancellation flows found a 28% reduction in user trust and a 54% decrease in usability scores compared to transparent alternatives. The companies running these flows have access to that data too. They made the tradeoff anyway.

What You Can Actually Do

What this means practically is that canceling a subscription you no longer want should not require research, persistence, or a twenty-minute phone call. But for a meaningful number of services, it does.

The appropriate response is not deference to the friction. It's recognizing the friction for what it is and navigating past it. Cancel through the web if the app obscures the button. If the service requires a phone call, call. Ask for confirmation in writing. Check your bank statement in the following billing cycle to verify the charge stopped. And if you're not sure what you're subscribed to in the first place, Substract will show you the full picture before you start.

The interface is not neutral. It was built by people whose job was to make leaving harder than it needs to be. Knowing that doesn't change the design, but it does change how you engage with it. You're not struggling to find the cancel button because the product is complex. You're struggling because someone decided you should.