Why cancelling a subscription is so much harder than starting one

What people run into

In June 2023 the Federal Trade Commission sued Amazon.com, alleging the company had enrolled tens of millions of customers into Prime without clear consent and then made the service difficult to leave. The FTC's own case summary is blunt about intent: "The primary purpose of its Prime cancellation process was not to enable subscribers to cancel, but to stop them," and it adds that Amazon leadership slowed or rejected changes that would have made cancelling easier because those changes hurt the company's bottom line. A stipulated settlement order was entered in September 2025. According to the FTC's refunds page, Amazon is required to pay $1.5 billion in refunds to affected customers and a $1 billion civil penalty, which together make up the $2.5 billion settlement. The FTC's case page still lists the matter's status as pending, with a further stipulated order on residual refund payouts entered in September 2026. The refunds page says that by September 2026 Amazon had issued over $845 million in refunds, and that payments continue through April 2027.

Amazon is the biggest name attached to the behavior, not the only one. The pattern has a name in interaction-design writing: "roach motel," a term the cataloguing site deceptive.design credits to Harry Brignull's 2010 work. On October 3, 2026 the site's "hard to cancel" page listed 135 examples in its hall of shame, naming companies across unrelated industries, among them a CRM tool, gym chains, an insurance platform, a design-inspiration site, a professional network, a meal-kit service, a whisky subscription and a car-breakdown service. The site's definition names the common mechanics: hiding the cancellation option, requiring users to call customer services to cancel, and making the process complex and time-consuming enough that some users give up and keep paying.

Regulators have tried to write a specific rule against this, and the attempt has had a rough run. In October 2024 the FTC finalized an amended Negative Option Rule, widely called "click-to-cancel," which mandated immediate cancellation through the same medium a customer used to sign up. On July 8, 2025, the Eighth Circuit Court of Appeals vacated the rule on procedural grounds, shortly before its July 14 compliance date: the court held that the FTC should have conducted a preliminary regulatory analysis once an administrative law judge found the rule's economic impact would exceed $100 million. As the law firm Sidley Austin summarized it that month, existing federal and state laws remain in place, including the Restore Online Shoppers' Confidence Act (ROSCA), which carries the same requirements for simple cancellation mechanisms. According to a later Sidley alert, on January 30, 2026 the FTC submitted a draft Advance Notice of Proposed Rulemaking on the Negative Option Rule to the White House's Office of Information and Regulatory Affairs, which had to clear it before it could be published for public comment. This analysis did not check what happened to that filing after February 2026.

How it plays out

Signing up for most online subscriptions takes a card number and a few taps; the deceptive.design page puts it plainly, describing a test in which creating a subscription took a matter of seconds.

Cancelling starts with finding the option at all. The same page describes the cancellation option being hidden, and gives the example of New York Times subscribers who reported being required to call customer service, waiting on hold for long periods, or being redirected to other pages without managing to cancel; in one test cited there, cancelling took roughly 8 minutes of conversation with a customer service representative.

A California consumer who signed up online is, since the state's amended Automatic Renewal Law took effect on July 1, 2025, legally entitled to cancel online: according to the state Attorney General, "if a consumer enrolled in an auto-renewal or continuous-service plan online, they must be able to cancel it online at will," and the business cannot "engage in any steps that obstruct or delay" that. This analysis did not check the rules of other states or countries, so it is a guess rather than a confirmed fact that elsewhere users fall back mainly on persistence, or on disputing the charge with their bank.

What already exists

• Trim (now OneMain MyMoney)WIRED's April 2026 review describes Trim as connecting to a user's bank account, listing recurring payments, and cancelling some subscriptions on the user's behalf, along with negotiating bills for a commission fee. That is the core of what a new cancellation tool would propose. Trim's own website now presents it as OneMain MyMoney and advertises finding recurring subscriptions; it does not mention cancellation on the user's behalf, so whether that service is still offered under the new name was not confirmed here.

A possible approach

It is plausible that most of the unmet demand here is not for a new cancellation tool but for broader coverage and lower friction in the tools that already exist: handling services outside the US, working without bank-account access for privacy-conscious users, or simply being free rather than charging a fee of their own to help a user escape a subscription. This is a hypothesis, not something the sources measured.

What we don’t know

Sidley Austin's July 2025 alert noted the FTC had 90 days to ask the Supreme Court to review the ruling; this analysis did not check whether it did. What happened to the FTC's January 2026 draft notice restarting the rulemaking after February 2026, and whether any new rule would change much given ROSCA already covers similar ground, was not checked here.

How often cancellation-concierge apps like Trim actually succeed at cancelling a resistant subscription versus merely informing the user how to do it themselves.

Whether users who distrust connecting a bank account to a third-party app have any equivalent option that does not require that access.

How this problem looks outside the United States, where neither ROSCA nor California's law applies.

Arguments against

The regulatory and commercial response is already substantial: the FTC's case against Amazon ended in a $2.5 billion settlement, California now specifically requires online cancellation for online sign-ups, and at least one reviewed app has sold cancellation on the user's behalf. A new solo-built tool competing on cancellation itself would be entering a space where an existing product already has bank-account connections; how that product handles the risk of acting for a customer was not examined here.

The vacated FTC rule, while a setback for click-to-cancel specifically, does not reopen the underlying question: according to Sidley Austin, ROSCA already carries the same simple-cancellation requirement and state laws such as California's remain in place, which weakens the case that cancellation friction is an unaddressed regulatory gap rather than a slow-moving enforcement one.

Could one person with AI build it?

A narrow, free directory of "here is where the real cancel option is for service X, and whether it requires a phone call" looks within reach of a solo builder and would not require bank access or acting as anyone; it is closer to a wiki than a financial tool. Building the cancel-it-for-you mechanism itself, especially anything that contacts a company on a user's behalf, carries data-access, trust and liability risks that a solo AI-assisted project would take on from a standing start. Both points are estimates, not findings from the sources.

The next check

Look for independent reporting or complaint data (not vendor marketing) on how often Trim, now OneMain MyMoney, or similar services actually complete a cancellation, and whether Trim's cancellation service still exists under the new name.

Check whether any directory-style, bank-access-free "how to actually cancel X" resource already exists and is current, since that is the narrower and more buildable idea surfaced here.

When to stop

If a focused search turns up a maintained, free, bank-access-free directory of real cancellation paths for major subscription services, the narrower idea should be shelved as already covered; if the existing cancellation-concierge apps are shown to reliably succeed even against phone-only retention flows, the broader idea should be shelved as a solved problem rather than an opportunity.