Signing up can take seconds. Leaving can take menus, retention offers, chats, phone calls and persistence. Federal cases show why that difference matters: every customer who stops trying can mean another recurring charge.
By Brandon L. Burley
Systems Explained
Amazon once had an internal name for the online process customers used to cancel Prime.
Iliad.
The name, according to the Federal Trade Commission, referred to Homer’s epic about the long and arduous Trojan War.
The comparison was unusually appropriate.
In its lawsuit against Amazon, the FTC alleged that customers trying to leave Prime were routed through multiple pages, retention offers and other steps before finally reaching cancellation. The agency further alleged that Amazon executives slowed or rejected changes that would have simplified the process because doing so would hurt the company’s financial performance.
Those were allegations.
Then Amazon settled.
In September 2025, the company agreed to a $2.5 billion settlement, including a $1 billion civil penalty and $1.5 billion for refunds to eligible Prime customers affected by the challenged enrollment and cancellation practices. Automatic refunds began in late 2025, and a second claims process continued in 2026.
The size of that settlement turns an everyday annoyance into a much larger question.
Why can subscribing feel almost effortless while canceling sometimes feels like work?
The answer begins with the economics of recurring revenue.
Friction can have value.
A subscription is permission to come back for more money
A traditional transaction ends.
You buy a hamburger.
The restaurant gets the money.
You get the hamburger.
A subscription changes the relationship.
The customer gives a company permission to return next month.
And the month after that.
And the month after that.
Unless someone acts.
Federal regulators describe many of these arrangements as negative-option programs because a customer’s failure to take affirmative action can result in continued billing. They include automatic renewals, memberships, continuity plans and free trials that convert into paid subscriptions.
Negative-option billing is not inherently deceptive.
It can be convenient.
Nobody wants to manually reauthorize a streaming service, cloud-storage account or software subscription every 30 days.
The important moment comes when the customer decides the relationship should end.
That is when convenience and company economics can begin pulling in opposite directions.
One forgotten month is small
Suppose a subscription costs $9.99.
A customer intends to cancel but forgets.
Another $9.99 is charged.
If 100,000 customers remain for one additional month, that would represent $999,000 in additional subscription revenue.
Three months would approach $3 million.
Those numbers are illustrative. They are not estimates for any particular company.
But they expose the basic economics.
A subscription company can increase revenue by finding more customers.
It can raise prices.
It can sell additional services.
Or it can keep existing customers longer.
That last part is called retention.
Its opposite is churn.
Reducing churn can be good business because acquiring a customer costs money. Advertising, introductory discounts, commissions and free trials all have costs.
There is nothing improper about persuading a customer to stay.
Improve the product.
Lower the price.
Offer another plan.
Provide better service.
Remind someone about benefits they may have forgotten.
Those are retention strategies.
The line becomes more important when persuasion turns into obstruction.
The government calls some of that design a dark pattern
The FTC uses the term dark patterns for design practices that can trick or manipulate people into decisions they otherwise might not make.
The agency has specifically identified difficult subscription cancellation processes as one form the problem can take.
Consider two journeys.
To join:
Choose plan.
Enter card.
Click subscribe.
To leave:
Find account settings.
Find membership.
Manage plan.
Decline an alternative.
Decline a discount.
Confirm cancellation.
Decline another offer.
Confirm again.
Maybe contact customer service.
Maybe return to the website.
Every individual screen can look harmless.
Together, they change the probability that a customer finishes.
That distinction matters.
Bad website design is not automatically manipulation.
A long cancellation process does not prove illegal intent.
Motive requires evidence.
Amazon is important because federal regulators said they found evidence about both the design and the economics behind it.
Amazon makes the incentive unusually visible
The FTC alleged that Amazon knew consumers were being accidentally enrolled in Prime and knew customers encountered a complicated cancellation process.
According to the agency’s amended complaint, internal Amazon materials discussed the financial consequences of making enrollment clearer and cancellation easier. The FTC alleged that executives slowed, rejected or reversed changes when those changes threatened business performance.
One company document cited by the FTC referred to a potential “shock” to business performance if enrollment were clarified.
The FTC also alleged that customer-service employees could process Prime cancellations but were directed in some circumstances to send customers back through the online Iliad cancellation flow instead.
The importance is not the literary joke behind the word Iliad.
It is what the allegations reveal about the business tension.
The customer sees a cancel button.
The company sees the possible end of a recurring revenue stream.
Those are two views of the same click.
Amazon is not the only case
The pattern appears in other FTC enforcement actions, although the facts differ and should not be flattened into one allegation.
In September 2025, Chegg agreed to pay $7.5 million to settle FTC allegations that it made recurring subscriptions difficult to cancel and, in some cases, continued charging consumers after they had requested cancellation. The FTC alleged that nearly 200,000 consumers had been charged after requesting cancellation since October 2020.
A month earlier, Match Group agreed to pay $14 million and accept restrictions involving advertising, billing and cancellation practices. The FTC’s case included allegations that Match.com made subscription cancellation difficult.
Then in May 2026, Shutterstock agreed to pay $35 million to settle FTC allegations involving subscription disclosures, consent and cancellation. The FTC alleged that some subscription structures lacked simple cancellation mechanisms and that material renewal and early-cancellation terms were not adequately disclosed.
A month later, the FTC sued what it described as a sprawling network of 15 corporations and eight individuals operating online subscription products ranging from fitness apps to PDF tools and horoscope services.
A federal court temporarily halted the operations while the case proceeds.
The FTC alleges that the companies hid recurring charges, billed people without authorization and failed to provide simple cancellation methods. Five of the products alone generated nearly a quarter-billion dollars in global revenue from early 2023 through mid-2025, according to the complaint.
Again, allegations are not findings.
But the enforcement record establishes something important.
Cancellation is not merely customer service.
It can be a consumer-protection and revenue issue.
Washington tried a remarkably simple rule
The basic idea behind the FTC’s 2024 “click-to-cancel” rule was easy to understand:
Leaving should generally be as simple as joining.
The rule would have expanded federal protections for negative-option programs and required sellers to provide a simple mechanism for ending recurring charges.
But the rule did not survive.
On July 8, 2025, the U.S. Court of Appeals for the Eighth Circuit vacated it.
The reason matters.
The appeals court did not rule that businesses have a right to make cancellation difficult.
It found that the FTC had failed to complete a preliminary regulatory analysis required after the rule was determined to have an annual economic effect of at least $100 million. The court called the agency’s rulemaking process procedurally insufficient and vacated the rule.
The cancellation rule was canceled because of the process used to create it.
The underlying problem remained.
In March 2026, the FTC began again, asking for public input on whether and how to update federal negative-option regulations.
The agency said it had received more than 100,000 complaints involving negative-option and related practices during the previous five years.
That is a useful distinction.
A court stopped one regulatory solution.
It did not make the marketplace problem disappear.
Why one more screen matters
Imagine you have already decided to cancel.
You click the button.
At that moment, the company knows something important.
You are about to stop its recurring revenue.
So another offer appears.
Three months at half price.
A free month.
A cheaper plan.
A pause instead of cancellation.
A reminder of benefits.
None of those offers is inherently improper.
A customer might genuinely prefer one.
But each step requires another decision.
And each decision creates another opportunity not to finish.
That is where human behavior becomes part of the business model.
A customer does not necessarily have to decide, “I want to keep paying.”
They may simply get interrupted.
Become frustrated.
Decide to come back later.
Misunderstand which button completes the process.
Or accept a pause while intending to leave eventually.
The subscription survives.
So does the billing permission.
Forgetfulness and obstruction are not the same thing
This distinction has to remain clear.
If someone subscribes to a streaming service, forgets about it for six months and never attempts to cancel, the company did not necessarily do anything wrong.
The customer forgot.
That is different from a customer attempting to leave and encountering unnecessary obstacles designed to prevent that choice from taking effect.
The difference can be investigated.
Not by asking whether cancellation “felt annoying.”
By measuring it.
How many clicks does enrollment require?
How many does cancellation require?
How long does each take?
Can both happen through the same medium?
Can someone who subscribes online cancel online?
Is a phone call required?
Is a chat required?
How many retention offers appear?
How visually prominent is the cancellation option compared with the option to remain?
Does cancellation require information the company did not request during signup?
Does the final screen clearly say the subscription has ended?
When does billing actually stop?
Those are observable facts.
That gives journalists something better than outrage.
A test.
Measure the friction
Take a group of subscription businesses.
Streaming services.
Gym memberships.
Dating sites.
Meal-delivery programs.
Software subscriptions.
Cloud storage.
Newspapers.
Identity-protection services.
Educational products.
Consumer memberships.
Use the normal enrollment process.
Record it.
Count every click.
Time it.
Then cancel.
Record that too.
One simple measure could be a cancellation friction ratio:
time required to cancel ÷ time required to subscribe
If joining takes 45 seconds and leaving takes nine minutes, the ratio is 12-to-1.
Then record the number of screens, retention offers, chats, phone calls, confirmations and other actions required before recurring billing actually stops.
The result would not establish motive.
It would establish experience.
That distinction matters because an investigation should not begin by assuming a company wants customers to fail.
It should begin by determining whether the path out is materially harder than the path in.
Then ask why.
Investors already understand the other side of the button
Public companies with subscription models routinely discuss metrics such as recurring revenue, subscriber growth, retention, churn, average revenue per user and customer acquisition.
Those terms can sound remote from the person staring at a cancellation screen.
They are not.
When a consumer cancels a subscription, the customer is ending a bill.
The company is losing expected future revenue.
The same action appears differently depending on which side of the screen you occupy.
That does not make retention sinister.
It explains why the cancellation experience deserves scrutiny.
The cancel button is not merely a piece of interface design.
It sits where customer autonomy and company economics meet.
The strongest evidence would be change
Interfaces do not stay still.
A service requiring four cancellation steps today could require seven next year.
Or two.
That creates another way to investigate friction.
Document the process over time.
Compare companies.
Compare industries.
Compare states where automatic-renewal laws impose different requirements.
If a company can provide a simple exit for customers in one jurisdiction, that creates an obvious question when consumers elsewhere receive something more complicated.
Not an accusation.
A question.
What makes the difference necessary?
Consumers can protect themselves — but that is only half the system
There are practical ways to reduce subscription inertia.
Review recurring charges.
Set reminders before free trials end.
Save cancellation confirmations.
Keep screenshots when a cancellation becomes complicated.
Watch for introductory prices that later change.
Look at bank and credit-card statements rather than relying on memory.
Those steps matter.
But “consumers should pay more attention” cannot answer every cancellation problem.
The company built the enrollment process.
The company built the exit.
If one was engineered to require almost no thought, it is reasonable to ask why the other requires considerably more.
Follow the dollar
Take a $14.99 subscription.
The customer decides to cancel in January.
They start the process but do not finish.
February: $14.99.
March: $14.99.
April: $14.99.
Nearly $45 has now been charged after the point when the customer first intended to leave.
For one person, that may be easy to overlook.
Across thousands or millions of subscribers, those additional months become something else.
Revenue.
That does not mean every company deliberately creates friction to produce it.
The government’s cases show why the possibility cannot simply be dismissed, either.
Amazon agreed to a $2.5 billion settlement after the FTC alleged that enrollment and cancellation design kept people in Prime without proper consent and made leaving unnecessarily difficult. Chegg, Match and Shutterstock reached multimillion-dollar settlements in cases that also involved subscription or cancellation practices. Federal regulators are still bringing new cases in 2026.
The next question should not be whether every inconvenient cancellation page is malicious.
That would be too easy.
Measure it.
Count the clicks.
Time the process.
Record every offer.
Compare the path in with the path out.
Then follow the recurring charges that survive somewhere between them.
A company is free to explain why a customer should stay.
The more revealing question is what happens after the customer has already decided to leave.
Advertisement · Practice investigating a claim
Turn “I suspect” into a question you can actually test. OSINT I teaches a deliberate public-source research process: frame the question, trace sources, verify and document the finding.
Review the course before enrolling →
I am a retired detective and criminal justice / government educator based in Tennessee. I am a commentary write for Tennessee Lookout and a weekly columnist with Knox TN Today. My work examines public policy, public safety systems and civic responsibility. My reporting and commentary have also appeared in Governing, The Arizona Capitol Times, South Florida Sun Sentinel, Police1, among other state and regional outlets.
