Signing up for Amazon Prime took one or two clicks. Cancelling took four pages, six clicks, and fifteen options. That gap is not a design flaw. It is the business model.

Every subscription playbook repeats the same line: "Reduce friction to grow." Strip the steps. Cut the clicks. Make it so easy to buy that the customer barely notices the charge. The advice sounds right. It is not. Friction was never removed from the system. It was moved to the other side of the door.

Amazon's internal code name for the cancellation process was Iliad. Named after Homer's poem about a ten-year siege. That was not a joke. The Iliad Flow forced Prime members through four pages, six clicks, and fifteen separate options before they could leave. Each page offered new reasons to stay. Each click led to another screen built to slow the exit. After the flow launched, Prime cancellations dropped 14%.

A draft memo stated that "clarifying" the enrollment process was not the "right approach" because it would cause a "shock" to business performance. When a system works that well at extracting revenue, you do not change it because it harms the customer. You protect it because it helps the margin.

Employees inside Amazon called the problem what it was. They described unwanted subscriptions as "an unspoken cancer." Two senior executives, Neil Lindsay and Jamil Ghani, oversaw the program. Both faced potential personal liability in the ongoing lawsuit.

The research on this is worth reading. In early 2025, Amit Zac and researchers at the University of Amsterdam and Oxford published a study on dark patterns: the design tricks that push people toward choices they did not intend to make. Their key finding was specific. These tricks are most effective when stored payment details mean a single click completes the purchase. No extra step. No second thought. No pause long enough for the buyer to reconsider. The less friction on the way in, the more power the system holds over the person inside it.

The researchers call this asymmetric choice architecture. One side of the door swings open. The other side is bolted shut.

The cost, when a court finally put a number on it: $2.5 billion. One billion in civil penalties, the largest ever imposed for an FTC rule violation. One and a half billion in refunds to 35 million consumers. The settlement came on day four of what was supposed to be a four-week jury trial. Amazon chose not to fight it in the open.

"Reduce friction" did not fail these people. It failed the system they thought they were signing up for.

The advice only looks in one direction. It treats friction as the enemy of growth and tells you to kill it wherever you find it. The problem is not friction itself. The problem is that stripping friction from the entrance creates room to stack it at the exit. That is not a growth strategy. That is a trap with a clean front door.

Most people treat this as an ethics problem. It is a system flaw. The structure of the advice produces the abuse. Fix the structure and the abuse has nowhere to hide.H3

Friction is a design choice, and it must run the same in both directions. If one click gets someone in, one click gets them out. Once that is clear, three moves follow from it.

Map every step a customer takes to start paying you. Then map every step they take to stop. Count the clicks, the pages, the choices, the calls. Put the two lists side by side on the same sheet of paper. Most operators have never done this. The comparison will show you what your system is actually built to do: retain customers by value, or hold them by maze.

This is the part that takes honesty. You may not like what the side-by-side reveals about your own design.

If your signup runs three steps, your cancellation runs three steps. Not a form with a required comment box. Not a phone call with a hold time. Not a chatbot that loops three times before it shows the cancel button. The same number of steps, the same level of precision. This means giving up the small retention tricks most subscription tools build in by default. Those tricks feel free. They are not. They cost trust, and trust is what keeps a customer past the first renewal.

A customer who stays because they could not find the exit is not a retained customer. They are a liability with a timer on it. Sooner or later they find the door. When they do, they leave angry, and they tell people about it. Track how many people stay after you make leaving easy. That number is the real health of your operation. The other number, the one padded by friction at the exit, was always borrowed time.

Running this for 90 days does something the old advice never did. It shows you which customers want what you sell and where your product is strong enough to hold without a trap. It separates real revenue from revenue borrowed from confusion. And it tells you who would come back, because the people who leave clean are the ones most likely to return.

At the end of 90 days, ask three things.

→ What kept people who had a clear path to leave?
→ What looked like retention but was just friction hiding the exit?
→ What part of the product got better once the maze came down?

That is the difference between advice that sounds right and a system that proves itself.

Amazon called its cancellation maze "Iliad," after Homer's poem about a siege built to outlast the people trapped inside it. They told you what it was in the name. Friction is a design choice. The design always tells you who it serves.