Products are built on assumptions about what people want, how they think, and\nwhy they behave — and most of those assumptions are wrong in ways the team\ncannot see from the inside. A UX researcher exists to replace guesses with\nevidence: to find out what real users actually need, do, and struggle with, and\nfeed that truth back into decisions before the team spends months building the\nwrong thing. The discipline exists because builders fall in love with their own\nideas, and someone has to systematically check those ideas against reality.
\n","wordCount":91},{"heading":"Core Mission","id":"core-mission","markdown":"Reduce the team's uncertainty about users to the point where the next decision\nbecomes obvious — and defend the gap between what people say and what they\nactually do.","html":"Reduce the team's uncertainty about users to the point where the next decision\nbecomes obvious — and defend the gap between what people say and what they\nactually do.
\n","wordCount":28},{"heading":"Primary Responsibilities","id":"primary-responsibilities","markdown":"The visible work is running studies — interviews, usability tests, surveys — but\nthe actual work is shaping which questions get asked and making sure the answers\nchange what gets built. A UX researcher scopes research questions with\nstakeholders, picks the method that fits the question and the timeline, recruits\nand screens the right participants, moderates without leading, synthesizes messy\nqualitative data into patterns the team can act on, quantifies behavior when\nnumbers are warranted, and translates it into findings that land. They also build\nresearch operations — panels, repositories, consent and incentive workflows — so\ninsight compounds rather than evaporating. Underneath it all is advocacy: keeping\nthe user a real presence in rooms full of business and engineering pressure.","html":"The visible work is running studies — interviews, usability tests, surveys — but\nthe actual work is shaping which questions get asked and making sure the answers\nchange what gets built. A UX researcher scopes research questions with\nstakeholders, picks the method that fits the question and the timeline, recruits\nand screens the right participants, moderates without leading, synthesizes messy\nqualitative data into patterns the team can act on, quantifies behavior when\nnumbers are warranted, and translates it into findings that land. They also build\nresearch operations — panels, repositories, consent and incentive workflows — so\ninsight compounds rather than evaporating. Underneath it all is advocacy: keeping\nthe user a real presence in rooms full of business and engineering pressure.
\n","wordCount":116},{"heading":"Guiding Principles","id":"guiding-principles","markdown":"- **What people say is not what people do.** Self-report is a hypothesis, not\n evidence. Trust observed behavior over stated preference, and design studies\n to catch the difference.\n- **The question dictates the method, never the reverse.** Pick the tool for the\n decision at hand; don't run a survey because surveys are easy or interviews\n because you like talking to people.\n- **Insight that doesn't change a decision is a hobby.** Research earns its keep\n by altering what the team does next. Tie every study to a pending decision.\n- **Bias is the default; rigor is the intervention.** Leading questions,\n confirmation bias, and the order of your prompts contaminate data silently.\n Assume you are biasing the result and design against it.\n- **Small n, deep truth.** Five users surface most usability problems; you don't\n need a thousand to learn that the button is invisible. Match sample size to\n the claim you want to make.\n- **Generative before evaluative.** Understand the problem space before you test\n a solution; testing the wrong solution flawlessly is still wrong.\n- **Triangulate.** One source lies; three that agree are hard to dismiss. Combine\n qual, quant, and behavioral data.","html":"A UX researcher sits at a crossroads. They work with product managers (who own\nscope and often want the answer yesterday), UX and product designers (who own the\nsolution and need actionable findings fast), data scientists (who own the\nbehavioral and experimental numbers and make natural triangulation partners),\nengineers (who reveal what's feasible to test), and marketing (who hold their own\ncustomer signal). The recurring tension is the timeline: research is most\nvaluable upstream, before commitment, but teams reach for it downstream, after.\nGood researchers embed early, teach stakeholders to run lightweight studies\nthemselves (democratization), and stay the steward of quality so scaled research\ndoesn't become scaled bias.
\n","wordCount":108},{"heading":"Ethics","id":"ethics","markdown":"Research puts a person's attention, data, and sometimes vulnerability in the\nresearcher's hands. Core duties: obtain genuine informed consent, including how\nrecordings will be used and stored; protect privacy and anonymize by default;\ncompensate fairly for time; and never deceive beyond what a debrief can repair.\nResearchers must guard against extractive practices — mining users for ideas\nwithout regard for their wellbeing — and against research that exists to justify\ndark patterns or manipulative engagement loops. There is a duty to report\ninconvenient findings honestly rather than soften them for stakeholders, and to\nrepresent participants — especially marginalized or excluded ones — fairly in\nrooms where they aren't present. The researcher is often the only person in the\nroom accountable to the human on the other side of the screen.","html":"Research puts a person's attention, data, and sometimes vulnerability in the\nresearcher's hands. Core duties: obtain genuine informed consent, including how\nrecordings will be used and stored; protect privacy and anonymize by default;\ncompensate fairly for time; and never deceive beyond what a debrief can repair.\nResearchers must guard against extractive practices — mining users for ideas\nwithout regard for their wellbeing — and against research that exists to justify\ndark patterns or manipulative engagement loops. There is a duty to report\ninconvenient findings honestly rather than soften them for stakeholders, and to\nrepresent participants — especially marginalized or excluded ones — fairly in\nrooms where they aren't present. The researcher is often the only person in the\nroom accountable to the human on the other side of the screen.
\n","wordCount":126},{"heading":"Scenarios","id":"scenarios","markdown":"**A PM demands a survey to settle a feature debate.** Two teams disagree about\nwhether users want a new dashboard; the PM wants a survey by Friday to \"let the\ndata decide.\" The expert pushes back: a survey reports what people *say* they'd\nwant, famously unreliable for unbuilt features — users over-claim interest in\neverything. The real question is whether the current workflow has a painful gap.\nThe researcher reframes it as a generative question, runs six 45-minute\ninterviews watching people do the actual task, and finds the dashboard solves a\nproblem only one segment has. The survey would have produced a tidy 68% \"yes\" and\nshipped the wrong thing. Method chosen by question, not by deadline.\n\n**A usability test where everyone \"succeeds\" but the product is failing.** Logs\nshow users abandoning checkout, yet a prior test reported high satisfaction. The\nresearcher re-runs it with two changes: drop the moderator's habit of nudging\nstuck users (\"try the top right\") and add think-aloud. Now three of five users\nhesitate 20+ seconds at the shipping step, mutter \"wait, is this final?\", and one\nnearly closes the tab. The earlier test had a leading moderator who rescued\npeople, masking the real friction. The fix — a clearer confirmation step — comes\nfrom neutral observation, not the satisfaction score that lied.\n\n**Synthesizing contradictory signals before launch.** Interviews say users love a\nfeature; analytics show almost nobody uses it. Rather than pick the flattering\nstory, the researcher triangulates: the interviewees were recruited from power\nusers (a sampling bias), while analytics covers everyone. Both are \"true\" for\ndifferent populations. The honest finding — \"beloved by a small, vocal segment;\nirrelevant to the mainstream\" — reframes the decision from \"launch or kill\" to\n\"keep as an advanced option, don't put it in onboarding.\" The value was in\nresisting the clean narrative.","html":"A PM demands a survey to settle a feature debate. Two teams disagree about\nwhether users want a new dashboard; the PM wants a survey by Friday to "let the\ndata decide." The expert pushes back: a survey reports what people say they'd\nwant, famously unreliable for unbuilt features — users over-claim interest in\neverything. The real question is whether the current workflow has a painful gap.\nThe researcher reframes it as a generative question, runs six 45-minute\ninterviews watching people do the actual task, and finds the dashboard solves a\nproblem only one segment has. The survey would have produced a tidy 68% "yes" and\nshipped the wrong thing. Method chosen by question, not by deadline.
\nA usability test where everyone "succeeds" but the product is failing. Logs\nshow users abandoning checkout, yet a prior test reported high satisfaction. The\nresearcher re-runs it with two changes: drop the moderator's habit of nudging\nstuck users ("try the top right") and add think-aloud. Now three of five users\nhesitate 20+ seconds at the shipping step, mutter "wait, is this final?", and one\nnearly closes the tab. The earlier test had a leading moderator who rescued\npeople, masking the real friction. The fix — a clearer confirmation step — comes\nfrom neutral observation, not the satisfaction score that lied.
\nSynthesizing contradictory signals before launch. Interviews say users love a\nfeature; analytics show almost nobody uses it. Rather than pick the flattering\nstory, the researcher triangulates: the interviewees were recruited from power\nusers (a sampling bias), while analytics covers everyone. Both are "true" for\ndifferent populations. The honest finding — "beloved by a small, vocal segment;\nirrelevant to the mainstream" — reframes the decision from "launch or kill" to\n"keep as an advanced option, don't put it in onboarding." The value was in\nresisting the clean narrative.
\n","wordCount":303},{"heading":"Related Occupations","id":"related-occupations","markdown":"A UX researcher works closest to the UX designer, who turns insight into form;\nthe two are constant partners but think differently — one converges on truth, the\nother diverges into solutions. Product managers share the obsession with user\nvalue but trade off against business and delivery constraints. Data scientists\nare the quantitative twin — same questions, statistical tools instead of\nobservational ones, the ideal triangulation partner. Graphic designers and\ntechnical writers consume research to shape communication; marketing managers\nhold an adjacent, sales-angled view of the same customers.","html":"A UX researcher works closest to the UX designer, who turns insight into form;\nthe two are constant partners but think differently — one converges on truth, the\nother diverges into solutions. Product managers share the obsession with user\nvalue but trade off against business and delivery constraints. Data scientists\nare the quantitative twin — same questions, statistical tools instead of\nobservational ones, the ideal triangulation partner. Graphic designers and\ntechnical writers consume research to shape communication; marketing managers\nhold an adjacent, sales-angled view of the same customers.
\n","wordCount":87},{"heading":"References","id":"references","markdown":"- *Just Enough Research* — Erika Hall\n- *Rocket Surgery Made Easy* — Steve Krug\n- *Interviewing Users* — Steve Portigal\n- *Observing the User Experience* — Goodman, Kuniavsky & Moed\n- *Quantifying the User Experience* — Sauro & Lewis\n- Nielsen Norman Group articles — nngroup.com","html":"