{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-10-21T13:54:31.602Z"},"content":[{"type":"documentation","id":"0e458b4b-346b-4ea5-abeb-93fcd7dc17d5","slug":"false-consensus-effect-user-research","title":"The False Consensus Effect: Why You Are Not Your User","url":"https://www.koji.so/docs/false-consensus-effect-user-research","summary":"The false consensus effect is the tendency to assume others share your choices and opinions more than they do, first described by Ross, Greene and House (1977). In their sandwich board study, students who agreed to wear an \"Eat at Joe's\" sign estimated around 62% of peers would also agree, while those who refused estimated about 33%. Mullen et al. (1985) pooled 115 hypothesis tests from 23 studies and found a highly significant, moderate effect, and Krueger and Clement (1994) found it held even when other information was available. NN/g's Raluca Budiu notes designers, developers and UX researchers all project their reactions onto users. The guide lists six signs of the bias in product teams, a six-step research routine (write predictions first, recruit outside your circle, ask about past behaviour, count instead of recalling, look for the opposite case, compare predictions with results), when your own experience is still useful, and how Koji's AI-moderated interviews, structured questions and evidence-linked findings support each step.","content":"The false consensus effect is the habit of assuming other people share your opinions, choices and behaviour more than they really do. In product work it shows up as \"I would use this, so our users will too.\" The fix isn't to ignore your own judgement. Write your guesses down before you research, talk to people who are not like you, and let what they actually did overrule what you expected them to do.\n\nThis guide explains where the effect comes from, how to spot it in a product team, a six-step routine for countering it in research, and when your own view is still worth something.\n\n## What the false consensus effect is\n\nThe term comes from a 1977 paper by Lee Ross, David Greene and Pamela House at Stanford. They defined it as people's tendency to see their own choices as common and appropriate, and the opposite choices as unusual, even deviant.\n\nTwo things make it hard to catch:\n\n- **It feels like knowledge, not bias.** When you picture \"a typical user\", the picture is built from your own reactions plus the people you happen to talk to. It doesn't feel like a guess.\n- **It works in both directions.** People who make a choice think it's popular. People who make the opposite choice think *that* is popular. Both groups can't be right.\n\nThe Nielsen Norman Group put the product version plainly. In Raluca Budiu's article \"You Are Not the User\", she writes that designers, developers and even UX researchers \"fall prey to the false-consensus effect, projecting their behaviors and reactions onto users.\" Jakob Nielsen's shorter version, from an interview about the best design advice he ever got, is six words: \"You are not the target audience.\"\n\n## The evidence: a sandwich board and 115 tests\n\nThe best-known demonstration is the sandwich board study. Ross and his colleagues asked Stanford students whether they would walk around campus wearing a sign that said \"Eat at Joe's\". Some agreed and some refused. Then each student estimated how many of their peers would make the same choice.\n\nStudents who agreed to wear the sign estimated that around 62% of their peers would also agree. Students who refused estimated that only about 33% would agree. Each group thought its own choice was the normal one.\n\nThat could have been a one-off. It wasn't. In 1985 Brian Mullen and colleagues pooled 115 hypothesis tests from 23 studies in a meta-analysis and found the effect was highly significant, with a moderate effect size. Joachim Krueger and Russell Clement (1994) went further and called it \"ineradicable\": in their studies the bias held even when people had other information about the group available to them.\n\nThe useful lesson for researchers is the second half of the sandwich board result. It isn't that people overestimate agreement in general. It's that your estimate of what's common is pulled towards whatever you yourself would do.\n\n## How it shows up in product teams\n\nProduct teams are a near-perfect setting for the effect. Everyone in the room uses the product every day, knows how it was built and usually shares a background with the people sitting next to them. Six signs it's shaping your decisions:\n\n![Listicle of six signs of the false consensus effect in product teams, from \"I'd never use that\" to skipping research because the answer feels obvious](/visuals/false-consensus-effect-signs-product-teams/false-consensus-effect-signs-product-teams.png \"If you hear these in a planning meeting, the team is guessing about users from its own habits · Source: Ross, Greene and House (1977)\")\n\n1. **\"I'd never use that.\"** A feature gets cut because the people in the room wouldn't use it, without anyone asking the people who might.\n2. **Internal dogfooding treated as user research.** Feedback from colleagues stands in for feedback from customers, even though colleagues know the product far better.\n3. **Personas that look like the team.** The \"primary persona\" has the team's job title, tools and level of technical comfort.\n4. **Surprise at the support queue.** Tickets about things \"everyone knows how to do\" keep arriving and get treated as edge cases.\n5. **One loud customer becomes \"users\".** A request from a customer who thinks like the team gets read as a request from the whole base.\n6. **Skipping research because the answer is obvious.** If nobody on the team can imagine a different answer, that's the moment the effect is strongest.\n\n## Why it's so hard to shake\n\nKnowing about the bias doesn't switch it off. A few reasons it sticks:\n\n- **You talk to people like you.** Part of the effect likely comes from who we spend time with. If your friends, colleagues and social feeds all share your habits, your sample of \"other people\" is biased before you start.\n- **Your own experience is vivid.** You remember exactly how you felt using the feature last week. A user you never met has no such weight in your memory.\n- **Expertise widens the gap.** The more you know about a product, the less your experience looks like a new user's. That makes the closely related [curse of knowledge](/docs/curse-of-knowledge-user-research) worse.\n- **Agreement inside the team feels like evidence.** When five people in a room all expect the same thing, it feels like a sample of five. It's really a sample of one shared background.\n\n## How to counter it in your research\n\nYou can't remove the bias from your head, but you can design research that catches it. This six-step routine works for interviews, surveys and usability tests alike.\n\n![Process showing six steps to counter the false consensus effect in research, from writing predictions down to comparing them with what people said](/visuals/false-consensus-effect-research-process/false-consensus-effect-research-process.png \"Writing your guesses down first is what lets the data prove you wrong\")\n\n### 1. Write your predictions down first\n\nBefore any fieldwork, have each person on the team write what they expect: the most common answer, the share of people who will pick each option, the reason people will give. Keep the predictions separate so nobody anchors on the loudest voice. Without a written guess, \"we already knew that\" is impossible to argue with.\n\n### 2. Recruit outside your own circle\n\nFriends, colleagues and power users from your community are the people most likely to think like you. Go further out: customers from other segments, people who tried the product and left, people who chose a competitor. See [finding research participants](/docs/finding-research-participants) for channels.\n\n### 3. Ask about what people did, not what they'd do\n\nHypothetical questions invite people to answer the way they imagine a sensible person would, which is its own consensus guess. Ask about the last time they faced the problem, what they tried and what happened. [Grand tour questions](/docs/grand-tour-questions) are a good opener.\n\n### 4. Count, don't recall\n\nAfter ten interviews, the two or three stories that matched your expectations will feel like \"most people\". Tally how many participants said each thing. Where you can, add a structured question so the count is exact.\n\n### 5. Look for the opposite case on purpose\n\nFor each finding, ask who disagreed and why. If nobody did, check whether your sample could have produced anyone who would. [Negative case analysis](/docs/negative-case-analysis) is a formal way to do this.\n\n### 6. Compare the predictions with the results\n\nPut the written guesses next to what you found and talk through the gaps. This is the step that changes how a team thinks. A product manager who predicted 70% would pick an option and saw 25% doesn't need a lecture about bias.\n\n## Hearing it from people who are not you\n\nThe steps above only work if the conversations happen with people whose lives differ from yours. A few habits make that easier in practice:\n\n![A researcher listening to a participant from outside the team's usual circle describe how they actually use a product](https://images.unsplash.com/photo-1512238972088-8acb84db0771?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw3fHxyZXNlYXJjaGVyJTIwaW50ZXJ2aWV3aW5nJTIwb2xkZXIlMjB3b21hbiUyMHRhYmxlfGVufDF8MHx8fDE3OTI1NjE4MTB8MA&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Monica Melton](https://unsplash.com/@monicomelty?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n- **Let the participant set the vocabulary.** If they call it \"the export thing\", so do you. Your internal names are a clue that you're thinking as an insider.\n- **Resist finishing their sentences.** When a participant hesitates, it's tempting to offer the answer you'd give. Wait instead. Our guide to [silence in user interviews](/docs/silence-in-user-interviews) covers how long.\n- **Invite observers from the team.** Engineers and designers who watch a session where someone struggles with the \"obvious\" flow update their mental model of users faster than any report.\n- **Debrief straight away.** Ask each observer what surprised them. Surprises are where the team's assumptions just met the real world.\n\n## When your own view is still useful\n\n\"You are not the user\" can be taken too far. Some practitioners argue that if you never use your own product you lose a lot of useful intuition, and that some teams genuinely are part of their audience: developer tools built by developers, for example.\n\nThe answer is to treat your own experience as one data point, clearly labelled. It's a fine source of hypotheses. It's a poor source of estimates about how common something is. Use it to decide what to ask, then let the participants decide what's true.\n\n## How Koji helps\n\nKoji runs AI-moderated interviews by voice or text, so you can hear from many more people outside your own circle than a team could schedule by hand. Here's how it maps to the routine above:\n\n- **Write your predictions first.** Every Koji study starts from a research brief with a stated goal and questions. Upload your team's predictions as a context file so they sit next to the study.\n- **Recruit outside your circle.** Share an interview link anywhere, import a CSV of customers to give each one a personal invite link, or recruit from Koji's research panel from the Panel recruit tab. Screening questions keep each study to the people you meant to reach.\n- **Ask about past behaviour.** The AI interviewer follows up on each answer, up to three times per question, so participants get asked \"what happened next?\" instead of being left at the hypothetical.\n- **Count, don't recall.** Structured questions (single choice, multiple choice, scales, ranking and yes/no) sit inside the conversation and produce exact counts in the report, while open questions get thematic analysis.\n- **Look for the opposite case.** Each finding in the report shows the share of participants behind it, and you can open the supporting responses to read the quotes, including from people who said something different.\n\n![Koji high-impact finding card backed by 26% of participants with a link to supporting responses](/email/shots/shot-finding.jpg \"Each finding shows how many participants back it, so a hunch can't pass as a majority\")\n\nThere's one more detail that matters for this bias in particular. When you try your own interview in Koji's preview, that run never counts as a response and never appears in the report. Your own answers stay out of your users' data.\n\n## Related Resources\n\n- [The curse of knowledge in user research](/docs/curse-of-knowledge-user-research)\n- [Confirmation bias in user research](/docs/confirmation-bias-user-research)\n- [Cognitive biases in user interviews](/docs/cognitive-biases-user-interviews)\n- [Negative case analysis](/docs/negative-case-analysis)\n- [Product dogfooding guide](/docs/product-dogfooding-guide)\n- [Structured questions guide](/docs/structured-questions-guide)\n\n## Sources\n\n- Budiu, R. (2017). You are not the user: The false-consensus effect. Nielsen Norman Group.\n- Krueger, J., & Clement, R. W. (1994). The truly false consensus effect: An ineradicable and egocentric bias in social perception. _Journal of Personality and Social Psychology_, 67(4), 596–610.\n- Mullen, B., Atkins, J. L., Champion, D. S., Edwards, C., Hardy, D., Story, J. E., & Vanderklok, M. (1985). The false consensus effect: A meta-analysis of 115 hypothesis tests. _Journal of Experimental Social Psychology_, 21(3), 262–283.\n- Nielsen, J. The best design advice I ever got. Interview, Peachpit.\n- Ross, L., Greene, D., & House, P. (1977). The \"false consensus effect\": An egocentric bias in social perception and attribution processes. _Journal of Experimental Social Psychology_, 13(3), 279–301.\n","category":"Research Methods","lastModified":"2026-10-21T05:50:10.964127+00:00","metaTitle":"False Consensus Effect in UX: You Are Not Your User (2026)","metaDescription":"What the false consensus effect is, the sandwich board study behind it, six signs it is shaping product decisions and a six-step routine to counter it.","keywords":["false consensus effect","you are not the user","false consensus effect examples","false consensus bias ux","cognitive bias user research","projection bias product teams","ross greene house 1977"],"aiSummary":"The false consensus effect is the tendency to assume others share your choices and opinions more than they do, first described by Ross, Greene and House (1977). In their sandwich board study, students who agreed to wear an \"Eat at Joe's\" sign estimated around 62% of peers would also agree, while those who refused estimated about 33%. Mullen et al. (1985) pooled 115 hypothesis tests from 23 studies and found a highly significant, moderate effect, and Krueger and Clement (1994) found it held even when other information was available. NN/g's Raluca Budiu notes designers, developers and UX researchers all project their reactions onto users. The guide lists six signs of the bias in product teams, a six-step research routine (write predictions first, recruit outside your circle, ask about past behaviour, count instead of recalling, look for the opposite case, compare predictions with results), when your own experience is still useful, and how Koji's AI-moderated interviews, structured questions and evidence-linked findings support each step.","aiPrerequisites":["Basic familiarity with user interviews or surveys"],"aiLearningOutcomes":["Define the false consensus effect and where the term comes from","Describe the sandwich board study and what the 1985 meta-analysis found","Spot six signs the bias is shaping product decisions","Apply a six-step routine to counter it in interviews, surveys and usability tests","Decide when your own experience is a useful input"],"aiDifficulty":"beginner","aiEstimatedTime":"9 min read"}],"pagination":{"total":1,"returned":1,"offset":0}}