Confirmation Bias
Confirmation bias is the tendency to seek, interpret, or remember evidence in ways that favor an existing belief or hypothesis.
Download Markdown (.md)Soroush’s Product Wiki · https://wiki.soroush.uk/topic/confirmation-bias/
01What it means
Confirmation bias influences how people gather and evaluate information. A team may ask leading questions, notice supportive cases, or explain away contradictory observations without applying the same scrutiny to favorable evidence. It differs from simply having a hypothesis: the problem is testing that hypothesis unevenly.[1][2]
02Why it matters
Product teams often invest effort and identity in an idea before investigating it. Biased evidence gathering can make a weak approach appear validated, delaying a useful change of direction and wasting implementation effort.
03When to use it
Consider this risk when planning interviews, interpreting experiments, reviewing analytics, or deciding whether a favored concept works. The response should be a better evidence process, not assuming that every disagreement proves another person is biased.
04How it works
- Write the belief and plausible alternative explanations before collecting evidence.
- Specify observations that would weaken or contradict the belief.
- Use neutral questions and consistent inclusion criteria.
- Review negative cases alongside supportive cases, preferably with another analyst.
- Record uncertainty and revise the decision when the evidence warrants it.
05A practical example
A team believes checkout abandonment comes from an unclear button. Instead of asking whether the button is confusing, it observes complete checkout attempts. Unexpected delivery fees emerge as a recurring issue, leading the team to investigate price disclosure before redesigning the button.
06Where it breaks
Awareness alone does not remove bias, and additional reviewers can share the same assumptions. Evidence can legitimately favor an existing belief. The aim is consistent scrutiny and openness to alternatives, not manufacturing equal support for every explanation.
Common mistakes
- Asking users to confirm a preferred explanation.
- Discarding contradictory sessions as unusual without a consistent rule.
- Treating a colleague’s disagreement as proof of bias.
07Sources
- Confirmation Bias in UXNielsen Norman Group · Professional guidanceConfirmation Bias in UX
- Confirmation and Cognitive Bias in Design CognitionASME / Autodesk Research · Research paperConfirmation and Cognitive Bias in Design Cognition