Cognitive Biases in UX: Anchoring, Framing, and Defaults
Anchoring, framing, and the default effect change how people judge value and decide. What the evidence says, how to use them in design and research, and limits.
What is anchoring?
Anchoring is the tendency to use the first figure or reference as the point of comparison. If a store shows a crossed-out price first and the final price second, the second looks like a deal even when it is normal. NN/g presents it as a psychological principle that affects how people perceive value and make decisions, both in real life and in an interface (NN/g, “How anchoring influences UX”).
It also happens inside design teams. A first concept, a palette, or a screen layout can become the reference everyone compares against, even when it was only a draft.
What is framing?
Framing is the change in a response depending on how the same information is presented. “95% success” and “5% failure” say the same thing, but they provoke different reactions. In research, framing biases what people say when the moderator asks a question in a certain way.
NN/g published a research example. Just over a thousand UX practitioners received the results of a test of a search function, randomly split into two versions: “4 out of 20 users could not find it” or “16 out of 20 users found it”. Those who read the failure version were 31% more likely to believe the search needed a redesign (51% versus 39%) (NN/g, “Decision framing: cognitive bias for UX pros”). Same data, different frame.
What is the default effect?
It is the tendency to stay with the option that is already selected. The Decision Lab explains that a preselected option creates the path of least resistance and is perceived as a social norm (The Decision Lab, “Can defaults save lives?”).
The classic study is Johnson and Goldstein, published in Science in 2003. They compared the share of people who agreed to be donors in European countries with presumed consent (you must say you don’t want to be one) and with explicit consent (you must say you do). They measured recorded agreement, not donations actually made. In the first group, the great majority ended up as donors; in the second, the rate was much lower: The Decision Lab puts it at about five times lower.
Why should a designer care?
Because every design decision has a starting value. A pre-ticked box to receive a newsletter, a preselected plan on the pricing screen, or a privacy option switched on by default all use the default effect, whether you intend it or not. If you don’t choose it deliberately, the system chooses it for you.
It also affects research. If the moderator presents the first proposal as the best, anchoring makes participants compare the rest against it. And if findings are presented as failures, framing can push the team toward a redesign that may not be needed.
How are they used in practice?
- Anchoring in pricing. Showing the recommended plan first, or with a visible reference, changes how the others are judged. Use it to help people compare, not to inflate a reference price.
- Framing in messages. Pick the frame that describes the real benefit. An error message that says what the person can do is easier to act on than one that only says what went wrong.
- Defaults in settings. Choose the starting value with the best outcome for the person in mind, and make changing it easy and visible. The Decision Lab recommends choosing defaults carefully, with the best social outcome in mind.
- In research. Write questions without adjectives that push the answer, present options in rotated order, and when reporting, state findings in both directions to check your own bias.
Example
Illustrative example: a health app asks at sign-up whether the person wants to share anonymous data for research. The checkbox comes pre-ticked. The team reviews the screen with someone who doesn’t know the product and finds that most people didn’t read the checkbox. They change the starting value to unticked, add a sentence that explains what is shared and why, and keep the option to change it visible in settings. Participation drops, but the people who do take part know what they agreed to. The team prefers it that way.
When NOT to use a bias in the business’s favor
- When the person pays or gives up data without understanding the cost. That is a dark pattern, not persuasive design. The wiki page on dark patterns describes the signals to review.
- When leaving is hard. A default that can be changed, but that nobody can find, deceives even if it is technically reversible.
- In decisions about health, money, or rights. Here the goal should be an informed decision, not conversion.
- As an excuse not to research. Knowing the biases doesn’t replace testing with real people. It helps you ask better questions, not guess.
Common mistakes
- Treating a bias as a trick that always works.
- Showing a reference price that isn’t real.
- Pre-ticking boxes that benefit the business and hide the cost.
- Writing research questions that anticipate the answer.
- Reporting a finding only in the frame that confirms what the team wanted.
How to start tomorrow
Review your purchase or sign-up flow for three things: which figure appears first, how the benefit or the risk is worded, and which option comes preselected. For each one, write down whether the choice serves the person or only the business. If it serves the person, document why. If not, change it. Then validate the change with user interviews or with a preference test comparing the old and new versions, with presentation order rotated.