Summary: UX analysis of the FTC lawsuit accusing Amazon.com of employing dark design patterns in its subscribe and unsubscribe workflows.
“Dark Patterns” in user-interface design intentionally mislead users, either to take actions they didn’t intend or to overlook information that would have caused them to make different decisions. Did Amazon.com employ dark patterns in the subscribe/unsubscribe flows for its Prime service until recently?
The Federal Trade Commission (FTC) filed suit against Amazon.com, alleging that it did step over the line, intending to dupe customers into subscriptions they never sought and then preventing them from escaping the traps they had unwittingly entered. FTC acknowledges that Amazon recently improved its UI to be less misleading, so the case concerns how the design used to be, not how it is now.
For a quick intro to dark design patterns, watch this 3-minute video by one of my favorite UX experts, Alita Joyce.
The FTC legal filing: https://www.ftc.gov/legal-library/browse/cases-proceedings/2123050-amazoncom-inc-rosca-ftc-v (The full legal filing runs to 159 pages. I read it so that you don’t have to.)
Unfortunately, much of the legal filing is redacted, making it hard to glean all the details about the case. My comments can only address the unredacted information in the FTC document. (Amazon should be treated as innocent unless found guilty, so it’s fair that the company’s confidential information has been hidden from my prying eyes.)
Leaving aside the possible merits of the FTC case, I’m pleased to see terms like “user experience” and “dark design patterns” used prominently in official government documents from an important agency.
The FTC makes much hay from Amazon having codenamed the Prime unsubscribe process the “Iliad Flow” in reference to Homer’s epic poem the Iliad, which is famously long with many twists and turns in the tale, such as the bad management practices of Agamemnon, King of Men. I have always found the Iliad to be a lusty and bloodthirsty story that’s highly engaging, but of course, I agree with the FTC that in UX design, we must get to the point and not distract users with side plots. (Late in the Iliad, swift-footed Achilles kills Hector, tamer of horses. This story peak must be the metaphor for Achilles finally getting out of his Prime subscription after realizing that he didn’t order many packages delivered to the black ships of the Achaeans.)
I am not a lawyer, so I can’t say whether Amazon’s design broke any laws. The judge will decide that. I am strictly commenting in my capacity as the world’s leading expert on web usability, so I will limit myself to discussing whether Amazon’s design in fact represents dark patterns that are likely to mislead users. Here’s my hot take on the UX aspects of the case.
Subscribing: Claimed to Mislead Users
One part of the lawsuit addresses how Amazon enticed new customers to subscribe to the Prime service.
FTC doesn’t like that Amazon showed the subscribe button as a big yellow button, whereas the opt-out option was presented as a small text link. Such a “nudging” design of making the preferred option huge makes for bad usability when the concept of what’s “preferred” is determined by the company’s business needs and not the users’ likely wants. That said, I don’t think the size discrepancy is a true dark pattern; it’s more of an annoyance, especially on mobile, where small links are hard to click.
Worse, the yellow subscribe button is on a screen that follows a series of previous screens where that same button design is used for the default or natural action, such as “add to cart” or “continue.” This sequence of steps habituates the user to consider yellow buttons as harmless and the natural thing to click to proceed at pace to the desired end goal of having completed the order. I would call this semi-dark because it’s a diversion from the user’s intent to sign up for a different service.
Much worse, the design concealed the fees that follow from signing up. On mobile, the fee disclosure is presented below the fold, meaning busy users won’t see it because they won’t scroll. (All users are busy.) Above the fold users only see that they get free shipping if they click the yellow button and that they will save $5.99 instantly.
Also alarming is the comparison table that shows the differences between shopping without and with Prime. The “with” column that supposedly lists the features of Prime doesn’t state its price, meaning that users may reasonably assume that it costs the same as the “without” option because most comparison tables on the web include the price for each option as a major comparative element. This is even true for Amazon’s own comparison tables on its product pages:
Typical Amazon comparison table, screenshot taken June 22, 2023. Note the prominent display of prices.
Now I feel we’re getting into dark pattern territory. As Jakob’s Law of the Internet User Experience states, users spend most of their time on other sites, so they assume that a new design works the same as they have learned from using all these other designs. Amazon exploited those expectations.
A close reading of the page shows a text line disclosing the price. But users’ eyes are drawn by actionable page elements like comparison tables, so many are likely to overlook that line.
The FTC also complains about Amazon tricking users who wanted only Prime Video into the costlier Prime subscription. Of course, it’s already dubious from a usability perspective to denote the difference between options by proprietary product names. But I’ll clear Amazon for this usability sin since many other companies commit similar sins. At least the video-only option has the word “video” in its name.
Worse, the ability to choose between paying for Prime and Prime Video is presented on a screen that only shows the Prime option, with Prime Video relegated to a separate screen that is only reached after clicking a rather discrete greater-than sign (“>”) placed off to the side. This interface element doesn’t have a strong perceived affordance of clickability. This is a dark pattern because users can’t tell from the design that they can choose between multiple options since only one option is visible. (It’s OK to make one option bigger or selected by default. It’s unacceptable to hide other options behind a non-standard way of displaying them.)
The misleading nature of the design is exacerbated by the fact that users can reach the signup page from the Prime Video page after clicking a button labeled “Watch with Prime.” Yes, if one understands Amazon’s internal marketing terminology, then one might make a distinction between Prime and Prime Video. But it’s pretty reasonable to expect that many consumers will not draw that fine line and assume that if they click a button to watch videos, then this is the service they will be paying for and not a more expensive service with more features.
In conclusion, Amazon’s sign-up design included many misleading elements and ways in which important information or options were hidden, but it wasn’t wholly egregious. I award it 4 out of 5 dark-design skulls:
Unsubscribe Flow: Claimed to Entrap Users
Once you’re in, can you get out? FTC claims that Amazon’s previous (now improved) unsubscribe flow was “labyrinthine” (a word I didn’t expect to see in an official government legal document). That’s the workflow Amazon itself supposedly likened to the Iliad in length and plot twists.
The FTC complains that users needed to traverse 4 pages with 6 clicks to unsubscribe. However, this is not necessarily bad in itself. I have never been a proponent of the “3-click rule,” which says that any important action should be possible in 3 clicks. The real question is whether each click (and each user decision) is straightforward and easy, so that it can be made quickly and with low risk of diverging from the user’s intent. 6 fast clicks have higher usability than 3 slow clicks. And 6 clicks that are all completely free of likely errors are much better than 3 error-prone clicks. In itself, the many-clicks design doesn’t rack up any of my skulls.
However, the actual workflow is terrible. And when each step is difficult and often misleading, having many steps compounds the problem. Let’s optimistically say that each step only has a 10% likelihood of leading the user astray. If so, then the string of 6 clicks only has 0.9^6 = 53% chance of being error-free.
Thus, a relatively low level of lousy usability (10% risk at each step) turns into an overall workflow that only goes right half the time.
(I should emphasize that the 10% level is only an example. I have not measured the actual error rate when average consumers used the old Amazon design. The true number could be larger or smaller.)
The FTC also complains that many steps involved a button labeled “End Membership,” but that clicking this button did not end the membership. Instead, it took the user to yet another page in the long workflow. I don’t view this as a terrible design, since the button has the correct information scent for progressing toward the user’s goal. However, it is reasonable for users to assume that clicking a button labeled with an explicit action would result in that action being executed. (After all, people have been habituated to this from countless “Add to Cart” clicks.) Thus, some users might not review the subsequent page thoroughly and might not realize that their action had not been completed.
As users progressed through the overly long workflow to cancel their membership, they were shown multiple hard-selling screens with information about how much that user had saved in the past from being a Prime member, as well as an offer for a reduced renewal price. At each step, the option to proceed was one of many buttons and not the most prominent one, even though it likely represented the user’s goal. After all, the user entered the unsubscribe workflow by choosing “End Membership.” This is a low-usability design but not a genuinely dark pattern.
On the final screen of the arduous cancellation process, Amazon gave the user an “End Now” button that would terminate their subscription. However, this button was at the bottom of a medium-length screen and was preceded by 4 buttons that would keep the membership active.
At this point in the workflow, it’s pretty much guaranteed that the user wanted to terminate the subscription since they had repeatedly been pressing buttons saying as much. Thus, a high-usability design would have shown the “End Now” button at the top of the screen and not at the bottom. For desktop users, this design is not a catastrophe; more of a nuisance.
The mobile design is much worse because the option the user is guaranteed to want — at least during the second half of the cancellation process — would be below the fold and necessitate much error-prone scrolling and hassle to locate and press.
The FTC concludes, “Amazon did not design the Iliad Flow to be simple and easy for consumers.” This is an understatement, and if I were doing a usability review, I would not hesitate to award the Amazon design team 5 bad-design skulls. However, the designers likely did as management told them, so we should judge the design for its potential dark-pattern side: is it misleading?
The unsubscribe process is unpleasant. I would even term it directly user-hostile.
But purely judged for its dark design elements, I only award Amazon 2 out of 5 skulls:
Conclusion: Amazon Is Bad, But Not the Worst
My verdict was 4 skulls for the signup process and 2 skulls for the unsubscribe process. I don’t think this averages out to 3 skulls because being less-misleading in one element of the website doesn’t excuse being quite misleading elsewhere.
Even so, Amazon doesn’t rise to the level of 5 out of 5 skulls for dark design.
While egregious, Amazon has not plumbed the stygian depths I have witnessed in other companies. I am damning them with faint praise because Amazon could obviously have designed much better workflows for both onboarding and offboarding. Whether laws were broken remains for the courts to decide. I pass judgment solely for dark UX — and find Amazon bad but not the worst.
Quiz: Test Your Comprehension
Take this short test with 10 questions to check that you have understood the UX analysis in this article. The correct answers are given after the illustration below.
Question 1: What is the FTC's allegation against Amazon.com in relation to its Prime service?
A. Amazon provided too many options for subscribing and unsubscribing.
B. Amazon used dark patterns to trick customers into subscriptions and made it difficult for them to unsubscribe.
C. Amazon did not provide enough information about the benefits of the Prime service.
D. Amazon made the unsubscribe process too easy, leading to accidental unsubscriptions.
Question 2: What is Jakob’s Law of the Internet User Experience?
A. Users spend most of their time on other sites, so they assume that a new design works the same as they have learned from using all these other designs.
B. Users prefer designs that are simple and easy to navigate.
C. Users are more likely to subscribe to a service if the benefits are clearly outlined.
D. Users are more likely to unsubscribe from a service if the process is complex and time-consuming.
Question 3: According to the article, what is the main issue with the "End Membership" button in Amazon's unsubscribe process?
A. Clicking the button ends the membership immediately.
B. Clicking the button does not end the membership, but takes the user to another page in the workflow.
C. The button is too small and difficult to click.
D. The button is not clearly labeled and can be confusing for users.
Question 4: How does the article describe the design of the final screen of Amazon's cancellation process?
A. User-friendly and straightforward.
B. Cumbersome, with the "End Now" button at the bottom of the screen and preceded by buttons that keep the membership active.
C. Clear and easy to navigate, with the "End Now" button prominently displayed.
D. Confusing, with too many options and unclear instructions.
Question 5: What does Jakob Nielsen specifically state is the problem with the size discrepancy between the subscribe button and opt-out link?
A. It prevents users from unsubscribing.
B. It represents a true dark pattern.
C. It makes for bad usability on mobile devices.
D. It tricks users into subscribing.
Question 6: What does Nielsen say about the sequence of steps leading to the subscribe button?
A. It is highly deceptive.
B. It habituates users to click yellow buttons.
C. It prevents users from seeing the fees.
D. It is not misleading.
Question 7: What does Nielsen say is the worst aspect of how Amazon presents Prime and Prime Video options?
A. No prices are shown.
B. The options are labeled with proprietary names.
C. One option is shown while the other is hidden.
D. The options are not clearly comparable.
Question 8: What aspect of the unsubscribe process does Nielsen say is not necessarily bad in itself?
A. The number of clicks required.
B. The hard selling at each step.
C. The misleading "End Membership" button.
D. The placement of the final "End Now" button.
Question 9: According to Nielsen, what makes the likelihood of errors compound in a multi-step process?
A. The number of clicks alone.
B. The complexity of the steps.
C. The hard-selling approach.
D. That all steps in the sequence must be completed correctly.
Question 10: What is Jakob Nielsen’s overall verdict on Amazon's use of dark design patterns?
A. Amazon is the worst offender of dark design patterns.
B. Amazon does not use any dark design patterns.
C. Amazon uses dark design patterns, but is not the worst offender.
D. The author does not provide a verdict on Amazon's use of dark design patterns.
Quiz Answers
Question 1: What is the FTC's allegation against Amazon.com in relation to its Prime service?
Correct response: B. Amazon used dark patterns to trick customers into subscriptions and made it difficult for them to unsubscribe.
Question 2: What is Jakob’s Law of the Internet User Experience?
Correct response: A. Users spend most of their time on other sites, so they assume that a new design works the same as they have learned from using all these other designs.
Question 3: According to the article, what is the main issue with the "End Membership" button in Amazon's unsubscribe process?
Correct response: B. Clicking the button does not end the membership, but takes the user to another page in the workflow.
Question 4: How does the article describe the design of the final screen of Amazon's cancellation process?
Correct response: B. Cumbersome, with the "End Now" button at the bottom of the screen and preceded by buttons that keep the membership active.
Question 5: What does Jakob Nielsen specifically state is the problem with the size discrepancy between the subscribe button and opt-out link?
Correct response: C. It makes for bad usability on mobile devices.
Question 6: What does Nielsen say about the sequence of steps leading to the subscribe button?
Correct response: B. It habituates users to click yellow buttons.
Question 7: What does Nielsen say is the worst aspect of how Amazon presents Prime and Prime Video options?
Correct response: C. One option is shown while the other is hidden.
Question 8: What aspect of the unsubscribe process does Nielsen say is not necessarily bad in itself?
Correct response: A. The number of clicks required.
Question 9: According to Nielsen, what makes the likelihood of errors compound in a multi-step process?
Correct response: D. That all steps in the sequence must be completed correctly.
Question 10: What is Jakob Nielsen’s overall verdict on Amazon's use of dark design patterns?
Correct response: C. Amazon uses dark design patterns, but is not the worst offender.