Dark Patterns: What Product Managers Should Never Ship
Amazon paid $2.5 billion in September 2025 to settle FTC allegations that internal emails showed executives deliberately designed confusing cancellation processes that enrolled millions of customers into Prime through a checkout flow harder to decline than accept. Epic Games paid $245 million in 2023 for confusing, inconsistent button placement that tricked Fortnite players into unwanted purchases. These aren’t abstract UX critiques anymore. The detail in the Amazon case — executives explicitly discussing how much friction to add to cancellation flows — is what turned this from a design debate into a fraud case with named individual defendants.
Dark patterns are interface designs built to manipulate users into decisions that benefit the business at the user’s expense, and the product manager sitting in the room when someone proposes one is the last line of defense before it ships. Understanding where the line sits, and why crossing it now carries real legal exposure on top of the ethical cost, has become a core part of the job rather than a nice-to-have.
What Counts as a Dark Pattern
The FTC’s own taxonomy is more specific and more useful than the general public conversation about “sneaky design.” The agency names, among others: interfaces that obscure or subvert a privacy choice, subscription flows that auto-charge unless affirmatively canceled, hidden fees revealed only late in checkout, and cancellation processes deliberately more tedious than the signup process that created the subscription. The common thread across all of them isn’t cleverness or persuasion — it’s that the design works specifically by making the user’s actual intent harder to execute than a different outcome the business prefers.
That definition matters because it excludes a lot of things people casually call dark patterns that aren’t. A well-designed default setting that most users would rationally choose isn’t manipulative just because it benefits the business too. A prominent “upgrade” button next to a smaller “no thanks” isn’t automatically a violation if declining is still one clear, undelayed click away. The test regulators and courts increasingly apply is whether a reasonable person would say they had a real, unobstructed choice — not whether the design happened to nudge them toward the option the business wanted.
The Legal Risk Is No Longer Theoretical
The enforcement pattern over the past few years has moved from warnings to real money, and the trend line points toward more, not less:
| Company | Pattern Alleged | Penalty |
|---|---|---|
| Amazon | Difficult Prime cancellation, unclear enrollment | $2.5 billion (FTC, 2025) |
| Epic Games | Confusing button configuration causing unwanted purchases | $245 million (FTC, 2023) |
| Publishers Clearing House | Misleading sweepstakes entry requirements, small-font disclosures | $18.5 million (FTC, 2023) |
| Various EU cases | Cookie consent design, tracking without valid consent | Up to 4% of global annual revenue under GDPR |
The FTC’s amended Negative Option Rule now requires cancellation mechanisms to be “at least as easy to use” as the signup flow that created the subscription — a specific, testable standard rather than a vague principle, which means a product team can actually audit their own cancellation flow against it before regulators do. State-level privacy laws including the CCPA, Colorado Privacy Act, and Texas Data Privacy and Security Act have added their own versions of the same requirement, and the California Privacy Protection Agency has been explicit that dark patterns are evaluated by effect, not intent — a team that never meant to manipulate anyone can still be found in violation if the design’s actual effect does so.
The scope of what counts keeps expanding, too. A 2024 sweep by 26 international enforcement authorities reviewed more than 1,000 websites and apps and found that nearly 40% created obstacles for users trying to make privacy choices, with a third repeatedly asking users to reconsider after they’d already decided to delete an account. That’s not a handful of bad actors; it’s evidence that the pattern is widespread enough that regulators are now treating it as a systemic industry problem rather than isolated misconduct, which is exactly the environment where enforcement intensifies rather than tapers off.
The individual liability angle is newer and worth taking seriously. Regulators have begun naming specific executives as defendants in dark-pattern enforcement actions, which changes the risk calculation for whoever signs off on a growth-hacky cancellation flow from “the company might get fined” to “I personally might be named in a settlement.” That’s a different conversation to have with a growth team proposing a friction-heavy retention flow than a purely reputational one.
The Line Between Persuasive Design and Manipulative Design
Every product uses persuasive design, and that’s not automatically a problem. Highlighting a recommended plan, showing social proof, or defaulting to an annual billing toggle are all legitimate influence techniques that respect the user’s ability to choose something else easily. The line dark patterns cross is asymmetry: making the business-preferred option effortless while making the user-preferred option deliberately harder, slower, or more confusing to find and execute.
The clearest gut-check a PM can run is this: would the team be comfortable explaining this specific design choice, in plain language, to the user it affects? “We made the recommended plan visually prominent because most users benefit from it” survives that test easily. “We made the cancel button gray and small, and put it three screens deep, so fewer people would find it” does not survive that test, and if a team can’t say the second sentence out loud without flinching, that’s the signal the design has crossed from persuasion into manipulation. This connects directly to the kind of standard a good product principles document should encode explicitly, so this test doesn’t have to be re-derived from scratch every time a growth team proposes something aggressive.
Common Dark Patterns PMs Should Never Ship
A short list of patterns worth refusing on sight, regardless of the growth case attached to them: pre-checked boxes for anything beyond the core transaction a user explicitly initiated; cancellation flows requiring a phone call or multiple confirmation screens when signup took one click; hidden fees that appear only at the final step of checkout after a user has invested time in the flow; countdown timers or “only 2 left” messaging that isn’t actually true; and confirm-shaming, where declining an offer is framed as a moral failing (“No thanks, I don’t want to save money”) rather than a neutral choice.
Each of these shares the same structural flaw: they work by exploiting a cognitive bias or a moment of low attention rather than by making a genuinely better offer. A discount that’s actually good doesn’t need a fake countdown timer to convert; the timer exists specifically because the underlying offer isn’t compelling enough to work on its own merits, which is itself a signal worth listening to about the offer, not just the interface around it.
Two of these deserve special attention because they show up in B2B SaaS as often as consumer apps, just dressed up differently. Fake urgency in enterprise software often takes the form of artificial scarcity around a pricing tier (“this rate expires at midnight”) on a deal that’s actually negotiable indefinitely. Confirm-shaming shows up in downgrade and cancellation flows as guilt-based copy questioning a user’s judgment for wanting to leave, rather than a straightforward acknowledgment of their choice. Both are just as measurable, and just as actionable to remove, as their more obviously manipulative consumer-app equivalents.
Where Teams Rationalize Their Way Into Dark Patterns
Small Frictions That Compound Into a Pattern
No single design decision looks like a dark pattern in isolation, but a series of small, individually defensible frictions compounds into something that clearly is one. A slightly buried cancel link, plus a confirmation screen, plus a “are you sure” retention offer, plus a second confirmation, none of which looks egregious alone, adds up to exactly the kind of obstruction the FTC has fined companies for. Recovery: periodically walk your own cancellation or opt-out flow start to finish as a first-time user would, counting clicks and screens, rather than reviewing each step in isolation during separate design reviews.
“The Data Shows It Converts” as a Justification
Treating a conversion lift as sufficient justification without asking why it converts is a common trap. A flow that increases signups by making the terms deliberately hard to read isn’t good growth work; it’s manipulation that happens to show up as a positive number on a dashboard, and the metric alone can’t distinguish between the two. Recovery: for any growth win attributed to a friction or urgency mechanic, explicitly ask whether the same lift would hold up if the mechanic were fully transparent to the user — if the answer is no, the “win” is coming from deception, not genuine appeal.
Assuming B2B Products Are Exempt
Assuming a product is exempt because dark pattern enforcement targets consumer-facing companies is a costly assumption. Regulators and reporting have made clear that companies operating business-to-business are facing scrutiny too, particularly for any consumer-facing aspects of billing, account management, or cancellation, even inside an otherwise enterprise product. Recovery: apply the same cancellation and consent standards to B2B self-serve flows that you would to a consumer product, rather than assuming enterprise contracts insulate the design from the same scrutiny.
No One Owns the End-to-End Experience
A growth or lifecycle team optimizing a single funnel step in isolation without anyone owning the end-to-end user experience lets friction accumulate across teams that never see the whole picture. This is the same coordination failure that shows up when deciding whether to kill a feature gets siloed by team instead of evaluated holistically — no one owns the full picture, so nobody notices when the sum of everyone’s individually reasonable decisions adds up to something none of them would have approved on their own.
Building Guardrails Into How Your Team Ships
Bake the “at least as easy as signup” standard into your own design review checklist before a regulator makes you do it reactively. If a flow takes users through three steps to subscribe, the cancellation flow should be auditable against that same three-step benchmark before it ships, not discovered as a problem after a support ticket volume spike or a public complaint.
Connect this discipline to how you justify roadmap investment more broadly. When a growth team proposes a retention mechanic, run it through the same rigor you’d apply to any other business case — what’s the actual mechanism generating the lift, and does that mechanism survive being made fully transparent to the user. If a proposed feature’s north star metric impact depends specifically on users not noticing what’s happening, that dependency itself is the red flag, and it should stop the conversation before the feature ever reaches a design review.
The gap between a persuasive product and a manipulative one usually isn’t a single dramatic decision; it’s an accumulation of small, individually rationalized ones that nobody stepped back to evaluate together. Build the habit of walking your own critical flows as a first-time user, question growth wins that depend on users not fully understanding what’s happening, and treat the regulatory risk as real rather than theoretical. Do that consistently, and the question of whether something is a dark pattern stops being a judgment call made under deadline pressure and becomes something your team already has a clear, tested answer for.
