Dark Patterns Online: How Design Shapes Your Choices

Dark patterns online are designs that impair informed choice, including through concealed terms or misleading pressure.

It helps to keep three questions apart. What did researchers see on real websites? What actually changed choices in experiments? And what can you check yourself? A recorded click is not the same as an unwanted purchase. It is also not proof of lasting financial harm.

What are dark patterns online?

Dark patterns online are interface choices that can block an informed decision. An official synthesis of workshop findings describes tactics like hidden terms, fake urgency and unwanted recurring charges (Federal Trade Commission, 2022). That report gathers examples and enforcement concerns. It is a synthesis, not a controlled test, so it cannot show how common these designs are, and it does not prove that any single company uses one today.

A bright button or a real sale is not automatically deceptive. Ask whether key terms are clear and the choices are accessible.

What can they look like in everyday shopping?

In shopping, these designs can take the form of unequal choices. One option is bright and simple. The fair alternative takes extra steps, hides behind vague wording, or sits under a link that says "see more choices."

A team crawled about 53,000 product pages across roughly 11,000 shopping sites and found 1,818 instances on 1,254 of them, sorted into 15 types (Mathur et al., 2019). That work shows these designs existed and varied in the pages they sampled. It is a historical snapshot. It cannot tell you how common they are today, and it cannot measure anyone's losses.

A single website can contain multiple design instances. Counting instances therefore differs from counting sites. Also, pages are not people. One page may reach many shoppers. The study sampled selected pages. It did not measure every shopper’s exposure. Nor did it measure financial loss. So instance counts and site counts answer different questions. Check what a headline counts before treating it as a claim about people.

Suppose, as a hypothetical, a checkout page preselects a monthly plan while the one-time option sits behind an extra link. Whether that design is deceptive or unlawful depends on the full page, the terms you agreed to and the law where you live. This article does not settle that. What you can say is simpler: the layout leans your attention one way, and that lean is worth a second look.

Do these designs actually change people's choices?

Yes, some do, at least in the short term. In one preregistered experiment, UK volunteers were recruited through an online panel in 2022 and shown a fictional investment product, leaving 2,252 complete observations. Designs that created a false hierarchy or blocked the exit raised initial acceptance compared with a neutral version (Zac et al., 2025).

Some of those effects weakened at the payment step. Notice what the study did not show. Nobody was charged, the product was fictional, and the volunteers were told the full picture before full card details went in. So the results cover clicks and payment choices, not proven long-term loss.

For a calmer way to weigh a hard choice, see this reflection on making decisions under uncertainty.

How can privacy choices be steered?

Privacy screens can be steered too, and the evidence here comes in two parts. One group observed 680 implementations of five popular consent tools across leading UK websites: the same handful of platforms, many different pages. The same study then ran a small field experiment with 40 people on eight designs. Removing the opt-out from the first page raised consent by 22 to 23 percentage points, while putting granular controls on that first page lowered it by 8 to 20 points (Nouwens et al., 2020).

Those are percentage points in a small experiment, not relative jumps. And the banner-versus-barrier comparison showed no detected effect in that test. The placement of refusal and fine-grained choices mattered in this experiment.

Observation and experiment answer different questions. One shows what sits on the page. The other shows what moved a choice. Neither one tells you what a company meant to do.

Does age, income or education predict who is affected?

The investment experiment found weak differences between age, income and education groups (Zac et al., 2025). It did not test carelessness or inexperience.

Weak differences do not establish equal vulnerability.

For related reading, see loss aversion. It is one related concept, not a full explanation of how these designs work.

What should you check before and after you agree?

Before you agree, slow down at the final screen. Read what recurs, when it recurs, and how to stop it. US consumer guidance suggests checking renewal terms and cancellation deadlines first, and looking closely at any prechecked box (Federal Trade Commission, 2024). A short set of checks:

• Note the renewal date, price and exact cancellation method in your calendar.
• Save the confirmation screen or email whenever you cancel.
• Check statements after the free period or trial ends, while the details are fresh.

A dispute is not a guaranteed refund, and banking procedures vary by country. Stopping a card payment does not by itself end a contract you signed.

This article is for informational and educational purposes only and does not constitute financial, legal, tax, medical, or professional advice. Individual results vary.