Amazon's 'Iliad Flow' Was a $2.5B Lesson in What Good UX Actually Means

Someone at Amazon named it the Iliad Flow. That's not a metaphor I'm reaching for — it's the actual internal codename that surfaced during litigation, and whoever coined it knew exactly what they were building: a cancellation journey long enough to deserve an epic poem's name. Six clicks. Four separate pages. Around fifteen distinct points where a user could be redirected into a retention offer, a downsell, a "are you sure" interstitial, before ever reaching an actual cancel button.

The joke landed on the design team first, presumably. It landed on the FTC next, and then it landed on Amazon's balance sheet: a $2.5 billion settlement — a $1 billion civil penalty plus $1.5 billion in consumer refunds — announced in September 2025, closing a case that had been working through the courts since 2023. That's not a fine that companies write off as a cost of doing business. That's a number that gets a UX pattern discussed in board meetings.

The Internal Name Is the Whole Case

Most dark-pattern controversies live in a gray zone of intent. A company can plausibly argue that a confusing flow was a byproduct of legacy architecture, of A/B tests that happened to favor complexity, of nobody quite noticing how bad it had gotten. That defense doesn't survive an internal codename. "Iliad Flow" isn't a description some outside critic applied after the fact — it's what Amazon's own people called it, which means somebody, at some point, sat in a room and thought "this is absurdly long" and shipped it anyway, because absurdly long was the point.

That's the detail that should unsettle every designer who's ever built a cancellation flow with an unnecessary retention offer bolted onto it "just to see." The FTC's underlying legal theory — that friction asymmetry between sign-up and cancellation constitutes an unfair practice under Section 5 of the FTC Act — isn't new law. What's new is a regulator with the appetite and evidence to make an example of the largest retailer in the world for it, and a settlement number large enough that "we'll fix it if we get caught" stopped being a rational bet.

The Asymmetry Was the Whole Design Strategy

Strip away the legal framing and look at the actual UX principle being punished: sign-up for Prime takes one click. Cancellation, before the settlement forced a redesign, took a guided tour through retention offers that most users had to navigate blind, with no visible progress indicator telling them how many more screens stood between them and done. This is the "roach motel" pattern in its purest form — easy in, deliberately hard out — and it's been a known dark-pattern category in UX research literature for close to a decade. What made this case different is that a regulator finally treated the asymmetry itself as the violation, independent of any specific deceptive screen inside the flow.

That's worth sitting with, because it changes what "compliant" means going forward. It's no longer enough to check that no individual screen contains a lie. The FTC's position — echoed later the same year in an $8 million settlement against Care.com over structurally similar obscured-cancellation complaints — treats the shape of the journey as the violation. A cancellation path meaningfully longer or more convoluted than the enrollment path is now a prosecutable asymmetry on its own, regardless of how each individual screen reads in isolation.

Designers Built This. That's the Uncomfortable Part.

The instinct in our industry is to talk about dark patterns as something that happens to a product — imposed by a growth team, a PM under revenue pressure, a stakeholder who overruled the design recommendation. Sometimes that's true. It's also frequently not. Someone wireframed the Iliad Flow. Someone built the fifteen decision points, tested them, watched retention numbers move, and called it a win. The tools of good design — attention to friction, sequencing, decision architecture — are directionally neutral. They work exactly as well pointed at trapping a user as they do pointed at helping one, and for years, "it increased retention" was treated as sufficient justification on its own, independent of the method.

The FTC ruling doesn't just create legal risk. It reframes the actual definition of craft. A cancellation flow engineered to be as frictionless as the sign-up flow isn't a compliance checkbox bolted onto good design after the fact — under this ruling, it is the design requirement, on the same tier as accessibility or performance. Designers who've spent years justifying friction as "considered UX" are going to need a better argument than "it moved a metric," because that argument is now the same one sitting in the FTC's case file against Amazon.

If you found this useful, it pairs directly with our earlier piece on the EU's Digital Services Act turning dark patterns into a compliance category — that piece covered the regulatory trend from the EU side; this is the specific American case that made the trend impossible to dismiss as a European quirk. Worth reading both if you're auditing your own product's cancellation flow this quarter, because increasingly, you should be.

What "Easy as Signup" Actually Requires

The FTC's implicit standard — cancellation should require no more effort than enrollment — sounds simple until you try to build it against a product with retention offers, plan-tier logic, and legal disclosures that all want screen real estate. The honest version of compliance isn't deleting all of that. It's making the path to done exactly as short as the path to committed, and putting anything else — offers, surveys, "are you sure" — behind the cancellation, not in front of it. That's an architectural decision, not a copy tweak, and it's the one Amazon avoided until a $2.5 billion incentive made avoiding it more expensive than building it right.

The Standard Just Got a Price Tag

For years, "good UX" and "ethical UX" were treated as adjacent but separable concerns — you could ship the first without much scrutiny of the second. That gap is closing, and it closed with a number attached: $2.5 billion, one internal codename that became public record, and a regulatory precedent that other agencies are already citing. The next team debating whether to add "just one more retention screen" to a cancellation flow isn't making a UX decision anymore. They're making a decision the FTC now considers big enough to litigate — and the internal Slack thread where someone jokes about the flow's length is discoverable.