UX Research: A Practitioner's Guide to Methods, Process, and Impact

What is UX research?

UX research is the discipline that turns customer understanding into product and business judgment. It studies users, behavior, and context so an organization can see where value already exists in people's lives, which unmet needs can become demand, what trade-offs customers will make, and what a product has to do to earn adoption, retention, or willingness to pay.

UX research originated in the evaluation of users' experiences with products, primarily to catch issues before users did. The discipline has evolved to serve a wider variety of roles: it can tell a company which segment has a coherent job to be done, what package customers see as worth paying for, why retained users stay, or which roadmap bet has enough customer value behind it to deserve investment.

The Purpose-and-Altitude Map is our attempt to map out the entire spectrum of value UX research delivers. Purpose describes the goal of a study: foundational research understands the space before the team has a frame; generative research turns that understanding into candidate directions; evaluative research judges a candidate against a standard. Altitude names the level of decision the evidence can inform, from interface details to business strategy.

Business value changes with altitude. At interface altitude, research improves the fit between a chosen direction and the task people are trying to complete. At product-strategy altitude, it shapes prioritization, retention, and packaging. At business-strategy altitude, it informs where the company looks for demand and how it should understand the market.

"Research-Led Growth" is a phrase we use to describe what happens when knowledge about users informs decisions at higher altitudes. At that point, UX research is doing more than describing users. It is helping the business choose growth opportunities.

How UX Research Informs Decisions: The Purpose-and-Altitude Map

The Purpose-and-Altitude Map

A 3×4 grid that locates any UX research study by two axes: purpose, whether the study is foundational, generative, or evaluative, and altitude, the level of decision it is meant to inform, from interface to business strategy. Every study answers one of its twelve defining questions, whether or not the team has named which one.

Most introductions to UX research sort methods by process: discover, explore, test, listen; or generative, evaluative, summative. The most cited alternative is Christian Rohrer's landscape, which plots methods on three dimensions: attitudinal against behavioral, qualitative against quantitative, and the context of product use (Rohrer, NN/g)Rohrer, When to Use Which User-Experience Research Methods, Nielsen Norman Group, 2022.. Mapping the field on intersecting axes is itself well established. Liz Sanders and Pieter Jan Stappers plotted the design research landscape on two of their own, design-led against research-led and an expert mindset against a participatory one (Sanders & Stappers, 2014)Sanders & Stappers, Probes, toolkits and prototypes: three approaches to making in codesigning, CoDesign 10(1), 5–14, 2014..

Each schema answers a real question. None asks the question that determines the shape of the work: how consequential is the decision this evidence is supposed to inform?

Our map uses two axes.

Purpose describes the job a study is doing. Foundational research understands a space. Generative research creates possible directions inside that space. Evaluative research judges a candidate against a standard, whether that candidate is a concept, a prototype, a direction, or a shipped product.

Altitude is the level of decision the research serves. Altitude is not a measure of difficulty or rigor; it describes the size, stakes, and reversibility of the decision the evidence feeds.

Reversibility is one practical way to identify altitude. Jeff Bezos made the distinction famous at Amazon: high-altitude decisions are "consequential and irreversible or nearly irreversible" and "must be made methodically, carefully, slowly, with great deliberation and consultation"; low-altitude decisions are "changeable, reversible" and can be made fast (Bezos, 2016)Bezos, 2015 Letter to Shareholders, Amazon, 2016.. The altitude test is simple: a button placement can be changed on Thursday; entering a category can be undone only at great cost.

The three types of UX research, and why foundational is not generative

The conventional trio, generative / evaluative / summative, looks tidy. It creates two problems: it treats timing as purpose, and it leaves foundational work out of the picture.

Generative and evaluative describe what a study is for. Summative describes when it runs, after the thing exists. A post-launch benchmark study and a pre-launch usability test are often doing the same job: judging a candidate against a standard. The timing differs; the purpose doesn't. Treating them as separate top-level purposes makes the taxonomy harder to use when a team is deciding what evidence a decision needs.

Foundational and generative research are different because they sit on opposite sides of a framing decision. Foundational research asks what is true about a space before the team has decided how to think about it. It studies people, contexts, constraints, behaviors, needs, and alternatives so the team can understand the terrain. Generative research uses that understanding to create possible directions. It produces opportunity areas, concepts, value propositions, workflows, service models, or feature ideas that can later be evaluated.

The practical test is simple: if the study is trying to learn what world the team is operating in, it is foundational. If the study is trying to turn that understanding into candidate answers, it is generative. Foundational research gives the team the map; generative research draws possible routes through it.

The evidence bar rises with the decision

Evidence has to generalize further as the decision gets bigger. At interface altitude, a finding often needs to hold only for the people using that screen. Small samples can work because the task is finding problems, not estimating prevalence. Five participants can surface most usability issues in a design, which is where the familiar rule of thumb comes from (Nielsen, 2012)Nielsen, How Many Test Users in a Usability Study?, Nielsen Norman Group, 2012..

At business-strategy altitude, the question becomes what share of a population will behave a given way. Five people cannot estimate a share of anything. In the same article, Nielsen puts the threshold for statistically significant numbers at "at least 20 users", with tighter confidence intervals requiring more, and a population estimate needs a defined sampling frame on top of the count (Nielsen, 2012)Nielsen, How Many Test Users in a Usability Study?, Nielsen Norman Group, 2012.. The "five users" rule is an interface-altitude rule that gets quoted as though it applied to the whole map.

Higher-altitude decisions need more than one kind of evidence. At interface altitude, a single source can be enough: someone cannot complete the task, and the recording shows it. Higher up, every source can be wrong in a way that changes the answer. Stated preference overstates what people will actually do. Behavioral data records what happened and stays silent on why. Market sizing inherits the assumptions used to define the category. A market claim resting on one source arrives with its weakness attached, and senior decision-makers can usually name that weakness because they already work with financial models and competitive analyses whose limits they know. McKinsey's account of the practice that pays is a triangulation recipe: conjoint analysis blended with ethnographic interviews, then combined with competitor analytics, patent scans, and the finance team's concerns.

In practice, if the decision would commit the company to a market, segment, price, or strategic direction, budget for scale and triangulation.

The corner most research lives in

Plot a team's last quarter of studies on the map. In many organizations they cluster in the bottom-right: evaluative work at interface altitude. That is usability testing, and it is where the discipline started, for reasons discussed in the aside below. It is valuable work. It is also the region where decisions are usually smallest, most reversible, and least contested.

Judd Antin, who led research at Airbnb and Meta, sorts research into macro, middle-range, and micro, and says the discipline is drowning in the middle: "The biggest reason UX Research is facing this reckoning is that we do way, way too much middle-range research" (Antin, 2023)Antin, The UX Research Reckoning is Here, One Big Thought, 4 May 2023.. His labels differ from the map's, but the diagnosis points in the same direction: too much research sits below the altitude where expensive decisions get made.

The underused corner is the top-left. Foundational and generative work at business-strategy altitude addresses questions like what market, audience, or behavior space matters? and where should we place strategic bets? Evidence has more leverage there because those decisions are large, expensive, and hard to undo.

That diagnostic is the map's simplest use, and it takes about ninety seconds in a team meeting. If six studies fall in one region, that is not a methods problem. More methodological refinement inside the same region will not, by itself, change which decisions research reaches.

The rest of this page stays with the top of the map: why it remains underused (§03), what research can produce there (§04), why its findings still fail to reach a decision (§05), and what a team can forecast once research reaches those decisions (§06). §07 locates every common method on the map.

What the work looks like at altitude

Methods travel. An interview serves foundational work at any altitude; a survey can be foundational, generative, or evaluative depending on what it asks. The pairings on the map are typical, not prescriptive.

The top row deserves specificity because "research belongs at business strategy" is too vague unless it names the work. At business-strategy altitude, research can mean market and category structure work; segmentation grounded in behavior rather than demographics; Jobs to Be Done research; ethnographic and contextual work on the constraints an industry imposes; concept testing on directions not yet committed to; and choice modeling, including conjoint analysis and MaxDiff, that quantifies the trade-offs behind pricing or portfolio decisions.

So what? Method choice follows from the question, and the question follows from altitude. Start with the size and stakes of the decision, then choose evidence that can carry it.

Why UX research stays tactical

UX research stays tactical for two reasons: access and positioning. Many teams sit inside design or product, far from business-strategy decisions. And when research defends itself mainly as mistake prevention, it argues for a downstream role.

A case study of a research function moving from incubation to maturity names the access barriers clearly: stakeholder bias, reactive tasking, and insight fragmentation. It defines maturity as moving "beyond tactical execution" to "directly shape long-term business strategy" (Singh & Barsoum, 2026)Singh & Barsoum, Architecting Strategic Influence: Operationalising the UXR Point of View Framework for Research Function Maturity, arXiv preprint, 4 June 2026..

The prevention argument is real, and weaker than the numbers usually quoted for it. It rests on Boehm's cost-to-fix curve (Boehm, 1981)Boehm, Software Engineering Economics, Prentice-Hall, 1981., which Boehm himself later qualified: the 100x escalation is "often" true, and for small, noncritical systems it runs "more like 5:1 than 100:1", while good architectural practice narrows it even on large critical ones (Boehm & Basili, 2001)Boehm & Basili, Software Defect Reduction Top 10 List, IEEE Computer, 2001.. The largest modern test found no escalation at all. Across 171 projects completed between 2006 and 2014, drawn from a database covering 47 organizations, the effort to resolve an issue late "was not consistently or substantially greater than when issues were resolved soon after their introduction" (Menzies et al., 2017)Menzies, Nichols, Shull & Layman, Are Delayed Issues Harder to Resolve? Revisiting cost-to-fix of defects throughout the lifecycle, Empirical Software Engineering, 2017.. Two limits on that result: the projects were small and mid-sized, the regime where Boehm and Basili had already conceded the ratio collapses, and the data stops at delivery, so it never measures the post-release phase where the largest escalation is claimed. What it does establish is that the 100x multiplier has no retrievable dataset behind it. The late-1970s figures Boehm reported were never published for analysis, and in nine years nobody has replicated or rebutted the one modern study that went looking.

That budget story fit the discipline's origins. UX research grew out of post-war human factors and ergonomics, and spent its first two decades inside that discipline before computing pulled it away (Grudin, 2012)Grudin, A Moving Target: The Evolution of Human-Computer Interaction, in Jacko, ed., The Human-Computer Interaction Handbook, 3rd ed., CRC Press, 2012., and Nielsen's Usability Engineering (Nielsen, 1993)Nielsen, Usability Engineering, Academic Press, 1993. formalized the interface routine: observe real users, find where the interface breaks, fix it before ship. Prevention still matters there. It says less about a market nobody has entered, a segment nobody has noticed, or a pricing model nobody has tested.

It is also a fragile budget story. Open US UX researcher listings on Indeed peaked at 2,990 in February 2022 and stood at 335 by January 2024 (Pybus, 2024)Pybus, Is there pent-up demand for UX research?, The ¼″ Hole, 2024.. That decline does not prove why any one team was cut, but it shows that the prevention argument did not protect the function when companies reassessed where research belonged.

Judd Antin names the downstream pattern "user-centered performance," work done to show that a team cares about users after the decision is effectively made (Antin, 2024)Antin, User-Centered Performance and How to Stop It, One Big Thought, 25 January 2024.. He also refuses to treat the familiar grievance, "Never driving the roadmap, no seat at the table," as sufficient explanation. Without demonstrable business value, he argues, the complaint does not matter much (Antin, 2023)Antin, The UX Research Reckoning is Here, One Big Thought, 4 May 2023..

That critique is useful, but it needs one more step: position shapes attribution. A function that mostly checks other people's bets will struggle to claim upside because the upside belongs to whoever placed the bet. Where research sits determines which kinds of value it can demonstrate.

So what? A budget case built only on avoided mistakes argues for the seat research already has. To claim upside, research has to reach decisions before the bet is placed.

Research-Led Growth: the business value of UX research

Research-Led Growth is not the claim that research causes revenue by itself. It is the claim that user knowledge creates business value when it shapes the decisions that decide where growth can happen: which market to enter, which segment to serve, which package to sell, which users to retain, and which capabilities to build next.

McKinsey's Business Value of Design study tracked the design practices of 300 publicly listed companies over five years and found that companies in the top quartile of its Design Index delivered 32 percentage points higher revenue growth and 56 points higher growth in total returns to shareholders than their industry counterparts over 2013–18. McKinsey argues the best results come from constantly blending quantitative user research such as conjoint analysis with qualitative work such as ethnographic interviews, then combining that with market-analytics reports on competitors, patent scans that track emerging technologies, and business concerns flagged by the finance team (Sheppard et al., 2018)Sheppard, Sarrazin, Kouyoumjian & Dore, The Business Value of Design, McKinsey & Company, 2018.. On the map, that is research operating in the top two rows. Over 40% of the 300 companies they surveyed still weren't talking to their end users during development at all.

The same study found "a strong correlation between successful companies and companies that resisted the temptation to cut spending on research, prototyping, or concept generation at the first sign of trouble."

HBR calls this kind of capability "privileged insights," deep knowledge of customers competitors cannot buy off the shelf because they are not producing it (Kahn et al., 2022)Kahn, Leinwand & Mani, How to Gain a Competitive Advantage on Customer Insights, Harvard Business Review, 2022.. Teams need a deeper grasp of what customers are trying to accomplish than of their own product catalog, because value is created every time a customer achieves that purpose (Cornfield, 2021)Cornfield, Recognizing Your Customer's Purpose Is Key to Growth, Harvard Business Review, 2021..

A small number of organizations are already working this way, and the number is moving. In Maze's vendor-run survey of nearly 500 researchers, designers, and product professionals, the share of organizations where research is essential to all levels of business strategy and operations rose from 8% in the 2025 wave to 22% in 2026 (Maze, 2026)The Future of User Research Report 2026, Maze, 11 March 2026. Vendor survey, n≈500.. The figure worth watching is the other 78%.

We call the capability Research-Led Growth: treating user knowledge as a driver of business opportunity rather than a check on existing bets. The name borrows from product-led growth, but the structure is different. Product-led growth asks how the product itself creates growth. Research-Led Growth asks how what the organization knows about users changes the decisions that create growth.

The five levers are the top two rows

The Purpose-and-Altitude Map is the framework. Research-Led Growth names the business value created when research is attached to product-strategy and business-strategy decisions. Each lever occupies a region of the map, so the claim can be tested against real work rather than admired as a slogan.

Product-market fit (business strategy, foundational and evaluative). Product-market fit is not only a post-launch metric. Sean Ellis popularized a single-question benchmark: in his words, "it becomes possible to sustainably grow a product when it reaches around 40% of users who try it that would be ‘very disappointed’ if they could no longer use it" (Ellis, 2019)Ellis, Using Product/Market Fit to Drive Sustainable Growth, Growth Hackers on Medium, 2019.. That question requires a product in the market. Before launch, foundational research, concept tests, and prototype studies can still gather evidence about whether a product will fit into prospective users' lives.

New market and segment discovery (business strategy, foundational). Deep user research can surface audiences a company did not know existed, with unmet needs adjacent to what the product already does. Identifying the jobs those users are struggling to get done is a more disciplined route to growth than starting from a product idea and hoping it fits (Christensen et al., 2016)Christensen, Hall, Dillon & Duncan, Know Your Customers' Jobs to Be Done, Harvard Business Review, 2016.. The decision it changes is where the company looks for demand before the roadmap hardens around the audience it already knows.

Pricing and packaging (business strategy, generative and evaluative). Pricing is not only a finance exercise. Willingness to pay, value perception, and which features people would pay more for versus take for granted are all research questions. When pricing gets set on competitive benchmarking and instinct because there was no time or no method to do better, research can close that gap with evidence. The decision it changes is what belongs in each package, what the market treats as table stakes, and where price becomes a signal rather than a barrier.

Retention (product strategy, foundational). Research can identify why users leave before they leave. Understanding the value gap, what the product fails to deliver for a specific cohort, gives the team a chance to intervene before it becomes a churn number. The decision it changes is where the team invests before a lagging metric confirms the loss.

Feature prioritization (product strategy and product, generative). Prioritization is not the same thing as sorting feature requests. Satisfaction scores rank features by how well they already work, which is a different question from which capabilities make users upgrade, refer, and stay. Research can measure the gap between how much a capability matters and how well it currently performs, which is where expansion revenue and advocacy tend to sit. The decision it changes is which capability earns the next unit of build time.

The three conditions

The levers matter only when three conditions hold. Research-Led Growth is an operating capability, not an automatic consequence of hiring researchers.

A translation mechanism. A way to carry findings into the rooms where decisions get made. This condition often fails, and the next section is about why.

Leadership willing to act. A decision-maker who will move when research changes the expected value of a choice.

Runway. Time to do foundational and generative work before a direction is locked. Research summoned to confirm a choice already made is far more likely to ratify an opportunity than locate one.

Why UX research findings don't reach decisions

Research often loses more force in the handoff than in the study. The method can be sound, the sample adequate, the finding well supported, and the decision still made by someone who never encountered it. Improving the method does not touch that failure, because the method already did its job. The path to the decision did not.

The gap is measurable, and it is wide. In a vendor survey of 309 researchers, research leaders, and colleagues who consume research, 94% said their leadership believes research is important, while 27% said research is referenced in almost every major decision (HeyMarvin, 2026)The State of Modern Research 2026: Closing the Listening Gap, HeyMarvin, April 2026. Vendor survey, n=309.. HeyMarvin's release says 84.5% of respondents were either primary tool decision-makers or strong influencers on those decisions, so the sample was unusually close to the tool-buying process. That bias should have narrowed the gap. It is still 67 points wide.

Four links often break between a finding and a decision:

Timing. The study is commissioned after the direction is set, so the likely outcomes are confirmation or inconvenience. The test is whether evidence can still change the choice. If the answer is no, the study is not informing the decision; it is documenting the decision's risk.

Form. A report is a useful container for evidence, but it is rarely the form a decision needs. A decision-maker choosing between two roadmaps needs a recommendation, the assumptions behind it, and an honest statement of how much confidence the evidence supports. A document that presents findings and leaves the inference to the reader has handed the hardest part of the job back without saying so.

Ownership. In many teams the researcher's responsibility ends at delivery and no one is accountable for the next step. The finding enters a shared drive and stops having an owner. Ownership means someone is responsible for carrying the evidence until a decision changes, absorbs it, or rejects it explicitly.

Access. Trade-offs get made in rooms, and researchers are frequently not in them. A finding relayed secondhand loses the qualifications that made it usable: what it does and doesn't cover, where it's fragile, what would change the conclusion. Jared Spool's prescription moves the decision-makers to the evidence instead. Every team member watches real users for at least two hours every six weeks, executives and business stakeholders included, because the teams that exempted them did not see the same gains (Spool, 2011)Spool, Fast Path to a Great UX: Increased Exposure Hours, UIE, 19 December 2011. Reported as an observed floor; no effect size published.. Spool reports the two hours as an observed floor and publishes no effect size with it.

The map shows why those links get weaker with altitude.

A finding at interface altitude has a short path to a decision. The person who needs it is often a designer or product partner on the same team, possibly in the same session. Ownership and access are usually straightforward; the finding can become a design change without anyone naming a process.

A finding at business-strategy altitude has a long path and no default owner. There is often no standing meeting where user evidence is an expected input to a pricing decision or a market-entry call. The people who make those calls are further away, and the evidence competes with financial modeling and competitive analysis, which already have routes into the room.

Upper-region research can fail even when the study itself is strong. Method quality is not enough. At higher altitude, the translation mechanism has to exist before the study starts. A team that plans a business-strategy study without first establishing who will receive it, in what form, and at what moment in the decision cycle is asking the findings to find their own way into the room.

One researcher cannot be the mechanism. "As researchers, we can't talk to every customer," says Behzod Sirjani, who ran research operations at Slack. "If I help my teammates have similar conversations, we collectively get closer to that actual lived experience" (Sirjani, 2021)Sirjani, quoted in In the Loop Conversations: Democratizing Research, Maze, 15 June 2021.. At interface altitude, one person can often carry a finding to whoever acts on it. Higher up, the rooms multiply. A mechanism that depends on one researcher being present fails whenever that researcher is not in the room.

In practice, before scoping any study above product altitude, name the decision, the decision-maker, and the moment when evidence can still change the choice. If any of the three is unknown, solve that before polishing the research plan.

So what? A common cause of wasted research is a finding with no route to a decision. The route gets longer the higher up the map you go.

What UX research can forecast

The standard shorthand for UX research, "generating insights that inform product decisions," undersells what research can do at its best. An insight explains something. A forecast helps a team decide what is likely to happen if it acts. Strong research turns uncertainty into a claim with evidence, named assumptions, and a visible margin for being wrong. At the top of the map, that forecasting value matters more than the insight label.

When qualitative and quantitative streams converge on the same finding from different angles, the result is more than a stronger insight. User interviews can surface why a segment behaves a certain way. Analytics can check whether the behavior is widespread. Concept testing can measure whether a proposed solution resonates. Contextual inquiry can reveal whether the environment can support adoption. Together, those signals let a team forecast whether a group of people is likely to adopt a solution, where that forecast is fragile, and what would have to be true for it to hold.

The useful question is whether the evidence changes expectations about the future. At that level, research can answer questions like these:

• Will there be demand for this product? Foundational research identifies the unmet needs; concept testing measures response to a proposed solution; contextual inquiry reveals whether the market is structured to support adoption.
• What is likely to happen if we build this feature? Evaluative research on analogous features, combined with attitudinal data on user priorities, produces a useful forecast with named assumptions.
• What will happen if we don't? Research that maps churn to specific value gaps gives a concrete picture of what inaction costs.
• Is there a structural feature of this industry that will make this product succeed or fail? Contextual and ethnographic research can surface organizational and industry dynamics that product analytics often misses.

Here the qualitative/quantitative divide becomes less useful. Analytics tells you what happened. Qualitative work can explain why. Survey and choice data can estimate how widely the pattern may hold. Combined well, those sources can tell a team what is likely to happen next on a stronger basis than any single source provides. The forecasts that change a roadmap are often the ones nobody requested. Anticipatory UX research is the practice of producing them on purpose.

Net Promoter Score shows the limit of simple intention measures. NPS measures stated likelihood rather than behavior, and intentions are a weak proxy for what people actually do: across experiments that successfully shifted intentions, a medium-sized change in intention produced only a small-to-medium change in behavior, d+ = 0.64 against d+ = 0.41 in the most recent independent replication (Conner & Norman, 2026)Conner & Norman, Attitudes, Intentions, and Behavior Change, Annual Review of Psychology 77, 311–337, 2026.. When C Space surveyed over 2,000 consumers, they compared stated likelihood to recommend with whether customers had actually recommended a brand. The categories came apart: 50% of customers scored as promoters, while 69% had actually recommended a brand, and 16% scored as detractors, while only 4% had ever told anyone to avoid one (Stahlkopf, 2019)Stahlkopf, Where Net Promoter Score Goes Wrong, Harvard Business Review, 2019.. That gap is a construct problem: the instrument measured stated likelihood while the decision depended on actual recommendation. Catching that kind of mismatch before it reaches a dashboard is the work of quantitative UX research.

Forecasting improves when researchers work with a broader evidence stack: observed behavior, stated priorities, task data, longitudinal patterns, and context. Researchers with survey-methodology training of the kind we teach in our survey methodology course can define the constructs that matter, specify exactly how each one is measured, and help build models that anticipate when a user is likely to be retained or to churn, well before that shows up in a satisfaction score.

So what? Strategic research does not promise certainty. Its value is a better forecast: clearer assumptions, earlier evidence, and a sharper account of where a product decision may succeed or fail.

UX research methods, located on the map

Methods do not belong permanently to one cell. A method has a typical home, but it moves when the question changes. The table below gives each method's common purpose, typical altitude, and business value when it is matched to the right decision.

The clustering is visible in the middle column: the methods most teams run most often are anchored at interface altitude.

The measurement-heavy methods here, including surveys, A/B testing, tree testing, choice modeling, and analytics, share a set of problems the individual method guides do not cover: what a sample lets you claim, which test fits the question, and how big a difference has to be before it changes a decision. Our guide to quantitative UX research treats those together.

For structured training in these methods, see our UX research courses.

UX research vs user research: what's the difference

UX research and user research are often used interchangeably, and most teams will understand either phrase. The practical difference is scope. UX research usually sits inside product, design, or digital experience work. It studies how people encounter a product or service, where it fits into their lives, and what decisions would make that experience more useful, usable, valuable, or adoptable.

User research is the broader phrase. It can include UX research, market research, customer insights, service design research, policy research, and any other discipline that studies people who use or might use a product, service, system, or institution. A user researcher might study a banking app, a hospital intake process, a public-benefits form, a SaaS workflow, or a buying committee.

Use the decision as the test. If the decision concerns a product, service, interface, workflow, or experience strategy, UX research is the natural label. If the decision crosses market, policy, service, customer insight, or operations, user research may be the broader label. Above product strategy, the distinction matters less because the questions often converge.

What is a UX researcher?

A UX researcher is the person responsible for making customer knowledge usable in product and business decisions. The role creates business value when evidence helps the organization choose which market to enter, which segment to serve, which product bet to fund, or which retention problem to solve first.

Job descriptions often reduce the role to methods: interviews, usability tests, surveys, diary studies, workshops. Method lists describe activity while omitting the accountability and potential impact underneath. Researchers are responsible for the warrant behind a user claim: who the evidence comes from, what it can support, what it cannot support, and which decision it should change.

What does a UX researcher do?

A UX researcher uses an array of qualitative and quantitative research methods to create timely, relevant knowledge about users that informs decisions about a product or a business.

Methods, scope, and stakes may vary, but the process a UX researcher employs is always the same: translate a business or product question into a researchable question, gather data of the right kind from the right people, analyze the data, and formulate insights in a form a decision-maker can use.

A UX researcher progresses in their career when they take on questions that are higher in altitude, informing decisions that have to do with product and business strategy. The Purpose-and-Altitude Map captures this progression: the lower rung is interface-altitude evaluation, where the question is whether people can use a prototype and the decision can still change this sprint. Higher-trust work reaches product and business-strategy questions: which segment is worth serving, which unmet need can support a product line, which concept is worth funding, or which pricing trade-off customers will accept.

Researchers move up the map by showing how evidence changes business decisions. Method range helps, but a longer method list does not by itself make a researcher more senior. The higher rows demand a strong connection between business problem and research question, more scientific sampling, data triangulation, and earlier access to the people making the decision. Moving up the map requires claims that can survive beyond the product team: claims about markets, segments, behavior change, willingness to pay, and investment.

Research-Led Growth depends on decision access. User evidence has to reach product-strategy and business-strategy decisions while the bet can still be shaped. After the bet is placed, research can improve execution; growth value comes from influencing the choice itself.

Entry is currently narrowest at the beginning of the ladder. In a June-July 2026 sample of 1,593 UX research job postings, Brian S. Utesch found that two in three roles targeted senior levels or above, while 10% were early-career and 5% were internships (Utesch, 2026)Utesch, UX Research Job Market 2026, 2026. Sample of 1,593 postings collected June–July 2026..

UX researcher and UX designer

UX researchers and UX designers overlap in skills but differ in accountability. Designers are accountable for the proposed experience: flows, interactions, information architecture, prototypes, and the final shape of what gets built. Researchers are accountable for what the team is entitled to believe before and after that design work: what people need, where they get stuck, what they value, and how far a finding can travel. When those accountabilities work together, design becomes a stronger business bet because the proposed experience is attached to evidence about adoption, retention, and willingness to pay. Blended titles become costly when production deadlines shrink every research question until only interface-altitude evaluation remains.

So what? Track your own studies by altitude for a quarter. If none of them touch market, pricing, retention, or prioritization decisions, method expansion alone leaves the business-value problem intact; earlier access to those decisions changes it.

For the routes into the field, see our guide to UX research training courses and certifications. For the methods themselves, see our UX research courses.

Two forces moving UX research up the map

Two forces are changing both the questions UX research is asked to answer and the way evidence is produced. Market-research convergence pulls researchers up the altitude axis. AI changes what a single research team can process, and it raises the burden of judging what counts as evidence.

UX research meets market research

Moving up the map is also a methodological shift. As UX researchers are pulled toward business-strategy decisions, they are increasingly asked questions market research has historically owned. How large is this segment? What would customers pay? Which features drive the buying decision? Which trade-offs should a roadmap honor? Product managers and business leaders making large bets need those answers, and UX research can help answer them when it has the right methods and enough access to the decision.

Answering them well means adding methods from the market research tradition to the UX toolkit. Choice modeling techniques such as conjoint analysis and MaxDiff quantify the trade-offs people make and the value they place on specific features, which is the kind of evidence pricing, bundling, and prioritization decisions often require. Segmentation, willingness-to-pay work, and demand estimation do related jobs. The point is not to turn UX research into market research. The point is to add quantitative methods that let qualitative insight scale into a defensible claim about a market.

The convergence is already visible in how the field talks about training. Learners' Research Week 2026 included an Advanced Market Research program framed around knowing what entire markets want, and a Growth UXR program that asked how research can drive the user and revenue growth organizations need (Research Week 2026)Learners, Research Week 2026.. For a practitioner, the boundary between UX research and market research is often a legacy of org charts. A researcher who can move responsibly from a diary study to a conjoint study stays useful as the decision altitude rises. For a worked example of the choice-modeling side, see our course on Conjoint Analysis and MaxDiff.

The test is whether the method can support the decision's claim. Conjoint and MaxDiff are powerful and easy to misapply, with attribute design, realistic choice sets, and sample structure doing most of the work. Borrowing the technique without the methodology can produce a confident number that happens to be wrong, which is worse than no number at all.

Is UX research being replaced by AI?

AI is reshaping the field quickly, but the replacement question is too blunt to be useful. The better question is whether AI is being used as an assistant or as a substitute. AI adds value when a researcher points it at real evidence and judges the output. It becomes dangerous when teams use it as a substitute for users. Pointed at real evidence, it can help a small team code and synthesize volumes that were previously infeasible and surface patterns across studies that no one could hold in memory. Used as a stand-in for users, it manufactures false precision. Synthetic respondents can collapse toward the average and misread subgroups. In one benchmark against real survey data, 48% of an LLM's regression coefficients differed significantly from the estimates derived from the human survey, and among those, the sign of the effect flipped a third of the time (Bisbee et al., 2024)Bisbee, Clinton, Dorff, Kenkel & Larson, Synthetic Replacements for Human Survey Data? The Perils of Large Language Models, Political Analysis, 2024..

The source of truth is the test. If AI is working on real research material, it can raise the ceiling for insight depth and quality when researchers have the methodology to direct it. If AI is replacing users, context, or judgment, it gives the team plausible outputs with no evidentiary footing.

How to start doing UX research at higher altitude

The next step comes straight off the map: plot the last six studies your team ran.

Two things are worth knowing from that plot. The first is which region you operate in, and whether that is a choice anyone made or just where the requests came from. The second is what would need to change in translation mechanism, leadership appetite, and runway for the next six to include one study from the top two rows.

The smallest meaningful move is often one foundational or generative study at product-strategy altitude, scoped with a named decision, a named decision-maker, and a moment when evidence can still change the choice. That shift does more than declare research strategic. It attaches research to a decision large enough to show what strategic research can change.

The harder version of the same move is a study nobody asked for: a question the organization has not framed, aimed at a need it has not yet recognized. That is the premise of anticipatory UX research, and it is where the translation mechanism matters most, because there is no standing request for the findings to attach to.

UX research FAQ

What are the types of UX research?
At the purpose level, the three types are foundational, generative, and evaluative. Foundational research establishes the terrain before the team has a frame. Generative research turns that understanding into possible directions. Evaluative research judges a candidate against a standard, whether that candidate is a concept, a prototype, a direction, or a shipped product. Summative remains useful as a label for post-launch measurement, but it describes timing rather than purpose.

What is the Purpose-and-Altitude Map?
A 3×4 grid that locates any UX research study by purpose, foundational, generative, or evaluative, and by altitude, the level of decision it is meant to inform, from interface up to business strategy. It has twelve cells, each holding one defining research question. Its most common use is diagnostic: plot a team's recent studies and the clustering shows which decisions their research is and isn't reaching.

Is UX research being replaced by AI?
No. The better distinction is assistant versus substitute. AI is useful when it works on real research material: processing large volumes of evidence, identifying patterns, and helping researchers see across studies. It is weak when it replaces users, context, or methodological judgment. In a world where competitors have access to the same AI tools, the organizations with deeper human understanding gain an advantage.

Are UX researchers in high demand?
Dedicated UX research hiring is more selective than the 2021 and 2022 market made it look. Open US UX researcher listings on Indeed peaked at 2,990 in February 2022 and stood at 335 by January 2024 (Pybus, 2024)Pybus, Is there pent-up demand for UX research?, The ¼″ Hole, 2024. Author’s own repeated Indeed searches, US only.. In a June-July 2026 sample of 1,593 UX research postings, two in three roles targeted senior levels or above, while 10% were early-career and 5% were internships (Utesch, 2026)Utesch, UX Research Job Market 2026, 2026.. Posting samples are not employment headcounts, but they point to the same practical answer: demand exists, and employers are concentrating it in roles trusted with larger decisions.