top of page

Dark Patterns: The Design Tricks You Never Agreed To

Jul 20
10 min read

In a narrow sense, he designed the interfaces he collected very well. You have almost certainly caught one. Perhaps it was the subscription that took four minutes to start and twenty minutes to cancel, routing you through a maze of confirmation screens, each one asking whether you were quite sure and offering a discount you had not requested. Perhaps it was the checkout page where a travel insurance add-on had quietly selected itself on your behalf. Perhaps it was the countdown timer insisting that three other people were looking at this room right now, or the newsletter box whose decline option read "No thanks, I prefer to pay full price."


In each case, you probably felt a flicker of irritation and then moved on. Most people stop thinking about the experience at that moment. And it is here that the more compelling questions start. These are dark patterns: interface decisions designed to nudge you toward decisions you wouldn't otherwise make. They are not bugs. They are not the consequence of a rushed dash. They are purposeful, they are wildly pervasive, they work way better than most people know, and until recently they were nearly completely unregulated. In this article, I look at what researchers have actually measured about them, why the evidence is more uncomfortable than the folklore, and what it means to anyone who designs digital things for a living.


Where dark patterns got their name

The term was coined in 2010 by British user experience designer Harry Brignull, who began collecting examples on a public website and provided the industry with a vocabulary for something practitioners had observed but never named. What Brignull found was not incompetence. In a narrow sense, he designed the interfaces he collected very well. They struck their targets accurately. The problem was, whose goals were they? Academic attention followed. Colin Gray and colleagues examined how practitioners themselves discussed these techniques and produced a taxonomy of five broad strategies: nagging, obstruction, sneaking, interface interference, and forced action (Gray, Kou, Battles, Hoggatt, & Toombs, 2018). The categories are worth pausing on, because each one describes a familiar experience.


Bar chart showing that dark patterns raised acceptance of a dubious data protection plan from 11.3 percent in the control group to 25.8 percent with mild dark patterns and 41.9 percent with aggressive ones.
The product never changed. Only the interface did.

Nagging is the repeated interruption that wears you down. Obstruction is the cancellation flow constructed like an obstacle course. Sneaking is when someone puts something in your basket without telling you. Interface interference is the visual hierarchy that makes “Accept all” a bright, confident button and “Manage preferences” a grey whisper. Forced action is the story you have to tell yourself to do the thing you already paid for. What Gray and colleagues found in practitioner conversations was not a guild of villains. It was a profession under commercial pressure, frequently uneasy about what it was being asked to build, and lacking both the vocabulary and the institutional standing to refuse it. That finding matters more than the taxonomy, and we will return to it.


How common are dark patterns, really?

To name a phenomenon is one thing. Measuring it is a different story and this is where the research really impresses. In 2019, a team at Princeton, led by Arunesh Mathur, created automated crawlers that went through about eleven thousand shopping websites, stepping through product pages, cart pages, and checkout flows just like a customer would. Then they had human experts classify what the crawlers found. This resulted in 1,818 instances of dark patterns on 1,254 websites, categorized into fifteen distinct types across seven categories (Mathur et al., 2019).


Two details in that study deserve emphasis. First, the crawler only examined text-based interfaces on product, cart, and checkout pages. It was unable to evaluate images, assess visual hierarchy, or touch the account creation or cancellation flows where some of the most aggressive patterns live. The figure of 1,254 sites is therefore a floor, not a ceiling, and the researchers said so plainly. Second, the team traced many of these patterns back to third-party providers. Companies were not always inventing their own manipulation. They were installing it, as a plugin, from vendors who sold urgency messaging and social proof notifications as a service. Some of those notifications were fabricated outright, displaying invented purchase activity and invented deadlines. The manipulation had been productized.


The mobile picture is worse. Linda Di Geronimo and colleagues at the University of Zurich analyzed 240 popular applications from the Google Play Store and found that 95 percent of them contained at least one dark pattern, with popular apps averaging around seven distinct types each (Di Geronimo, Braz, Fregnan, Palomba, & Bacchelli, 2020). Only 5 percent of gambling apps, and only 5 percent of shady free games, are safe. Ninety-five percent of popular, mainstream, everyday applications. At this point the honest reader should raise an objection. Perhaps these patterns are everywhere because they are ineffective, and firms use them like any cheap tactic that costs nothing to try. Prevalence is not proof of power. That objection held up for a decade, because nobody had run the experiment.


Do they actually work?

In 2021 the legal scholars Jamie Luguri and Lior Jacob Strahilevitz published what they described as the first public evidence on the question, and the answer was not reassuring (Luguri & Strahilevitz, 2021). They recruited a representative sample of American consumers and offered them a dubious data protection plan, randomly assigning participants to one of three conditions. The control group received a neutral interface. A second group received what the researchers called mild dark patterns. A third received an aggressive version, laden with pressure, additional steps, and a countdown timer that prevented participants from clicking through quickly. In the control condition, 11.3 percent of participants accepted the plan. Under mild dark patterns, acceptance rose to 25.8 percent, an increase of 228 percent. Under the aggressive condition, 41.9 percent accepted, close to quadrupling the baseline.


Sit with those numbers for a moment. The product did not change. The price did not change. The value proposition did not change. Only the interface changed, and roughly three in ten additional people made a decision they would not otherwise have made. Whatever we imagine consumer choice to be, it apparently rests on a foundation that a designer can move at will. The second study identified which techniques did the heavy lifting. Hidden information, trick questions, and obstruction were particularly effective at manipulating people successfully. These are, notably, the techniques that operate by degrading a person's understanding rather than by pressuring their emotions. They do not persuade you. They make it harder for you to know what you are agreeing to.


The dark patterns nobody complains about are the dangerous ones

The finding with the sharpest implications is one that is easy to skim past. Aggressive dark patterns provoked a significant backlash. Participants noticed, resented it, and thought less of the company afterwards. Mild dark patterns produced no meaningful backlash and more than doubled acceptance. Strahilevitz directly stated the conclusion when presenting the work to the American Federal Trade Commission: mild dark patterns are the most insidious because they substantially increase acceptance without generating consumer anger.


This finding inverts the intuitive picture. We tend to assume the worst offenders are the loudest ones, the flashing countdown timers and the guilt-tripping decline buttons. But those are self-limiting, because they announce themselves, and a company that annoys its customers pays for it in reputation. The genuinely corrosive patterns are the ones calibrated to sit just below the threshold of notice: the pre-checked box, the slightly quieter decline link, and the confirmation step that seems reasonable in isolation. They extract the behavior without triggering the immune response. There is a second finding that ought to trouble anyone who cares about fairness. Less educated participants were significantly more susceptible to mild dark patterns than their better-educated counterparts. The costs of manipulative design are not distributed evenly across a population. They fall hardest on the people least equipped to absorb them, which converts a design question into a question about justice.


You are not as good at spotting these as you think

A common response at this point is personal exemption. Other people fall for this argument. I notice the tricks. Di Geronimo's team tested that belief directly. Alongside the analysis of 240 apps, they ran an experiment with roughly 589 participants, asking them to evaluate app interfaces. Most participants did not perceive the manipulative designs at all. When researchers first informed participants about dark patterns and told them what to look for, their recognition improved considerably (Di Geronimo et al., 2020). That second half is the hopeful part, and it is also the part that turns the narrative from a lament into an argument. Vulnerability to these techniques is not fixed. It is a function of knowledge. People who have been taught to see the patterns can see them, which suggests that the appropriate response is education rather than resignation. But the default state, for most users, most of the time, is not to notice. This means that the flicker of irritation described at the top of this article is the exception, not the rule. The patterns that irritate you are the ones that failed. The successful ones were never registered at all.


The law arrives slowly.

For most of the past decade, all of these patterns sat in a regulatory vacuum. That is changing.

In December 2022 the American Federal Trade Commission announced a settlement with Epic Games, the maker of Fortnite, and finalized the order in March 2023. Epic agreed to pay 245 million dollars in refunds over allegations that it had used dark patterns to charge players for unintended purchases. The complaint described a counterintuitive, inconsistent, and confusing button configuration that led players to incur unwanted charges from a single press, alongside a practice of locking the accounts of customers who disputed those charges with their credit card companies. The order bars Epic from charging consumers through dark patterns without obtaining affirmative consent, and the commission described the sum as its largest refund in a gaming case (Federal Trade Commission, 2023).

The number is what gets the headlines, and $245 million is a genuinely large figure. But the more consequential part of the settlement is the phrase "bars the use of dark patterns." A regulator has now treated interface design not as decoration or as marketing, but as conduct, and conduct is the kind of thing that carries liability. The European Union has moved in a similar direction under the Digital Services Act, and further research has begun to formalize how these patterns should be defined and measured for exactly this purpose (Mathur, Kshirsagar, & Mayer, 2021). For anyone entering the field, that shift changes the calculation. Manipulative design used to be an ethical question that a designer could privately dislike and publicly build anyway. It is becoming a legal exposure that a company has to price. The person in the room who can identify a dark pattern before it ships is no longer just the conscience of the team. They are risk management.


The designer in the room

Return to the finding from Gray and colleagues, because it is where this argument lands. When researchers listened to practitioners discussing these techniques, they did not hear enthusiasm. They felt discomfort. Designers often recognize when they are asked to create something that works against the user, yet they proceed because the request carries the authority of a business objective, while the objection is based solely on a feeling.

The gap there is not moral. It is evident and professional. A designer who says "this tactic feels manipulative" loses that argument. A designer who says "this is an obstruction pattern; the FTC has already taken enforcement action on this exact configuration, the research shows aggressive versions generate measurable reputational backlash, and here is a version that hits the same conversion target without the exposure" wins it. Both designers had the same instinct. Only one had the vocabulary, the evidence, and the standing to act on it. That is the difference between decoration and a profession. Interface design is not a matter of making screens attractive. It is the practice of shaping how other people understand their choices, which is a serious power and one that most of the industry wields without having ever been asked to think carefully about it.



At Raffles Jakarta, we build our Digital Media Design program on that premise. Students learn interface and interaction design as a discipline with consequences: how attention actually works, how people really read a screen, where persuasion ends and manipulation begins, and how to make the case for the better version in a room full of people who want the numbers to go up. These techniques are taught because you cannot build what you cannot recognize, and designers who understand them will be trusted to build what comes next.


The pattern you cannot unsee

There is a small, reliable side effect of learning this material. You start to notice. You notice that the cookie banner has an "Accept all" button but no equivalent "Reject all" button, and that this difference was a deliberate decision by a paid person. You notice which button is bright and which is grey. You notice that the box was already ticked. You notice how long the cancellation takes compared to the sign-up.


That noticing is not cynicism, but literacy. It is literacy. The research is clear that people who know what to look for are considerably harder to manipulate, which means the act of reading an article like this one has already changed your position slightly. The interfaces did not get better. You did. The wider question is whether the next generation of people building these systems will learn to see what they are doing. The patterns will not disappear by themselves. They are effective, cheap, and available as plugins, and they will persist as long as the people designing them lack the evidence and confidence to argue for something better.


Arman Poureisa

Marketing Manager

Raffles Indonesia



References

Di Geronimo, L., Braz, L., Fregnan, E., Palomba, F., & Bacchelli, A. (2020). UI dark patterns and where to locate them: A study on mobile applications and user perception. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems (pp. 1–14). Association for Computing Machinery. https://doi.org/10.1145/3313831.3376600

Federal Trade Commission. (2023, March 14). FTC finalizes order requiring Fortnite maker Epic Games to pay $245 million for tricking users into making unwanted charges. https://www.ftc.gov/news-events/news/press-releases/2023/03/ftc-finalizes-order-requiring-fortnite-maker-epic-games-pay-245-million-tricking-users-making

Gray, C. M., Kou, Y., Battles, B., Hoggatt, J., & Toombs, A. L. (2018). The dark (patterns) side of UX design is discussed. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems (pp. 1–14). Association for Computing Machinery.

Luguri, J., & Strahilevitz, L. J. (2021). This article examines dark patterns. Journal of Legal Analysis, 13(1), 43–109. https://doi.org/10.1093/jla/laaa006

Mathur, A., Acar, G., Friedman, M. J., Lucherini, E., Mayer, J., Chetty, M., & Narayanan, A. (2019). Dark patterns at scale: Findings from a crawl of 11K shopping websites. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), Article 81. https://doi.org/10.1145/3359183

Mathur, A., Kshirsagar, M., & Mayer, J. (2021). What makes a dark pattern... dark? Design attributes, normative considerations, and measurement methods. In Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems (pp. 1–18). Association for Computing Machinery. https://doi.org/10.1145/3411764.3445610

bottom of page