UX/UI Designer interview questions
100 real questions with model answers and explanations for Middle candidates.
See a UX/UI Designer resume example →Practice with flashcards
Spaced repetition · Hunter Pass
Questions
Qualitative research supports contextual explanations of experience, while quantitative research estimates its measurable patterns.
- Qualitative evidence can reveal how people interpret a situation and help form explanations for behavior or problems within the studied context.
- Quantitative evidence can estimate prevalence, distributions, differences, and associations in a defined population, with uncertainty determined by the design and sample.
- Neither label alone guarantees causality or generalizability, so conclusions must remain within the sampling, measurement, and study design limits.
Why interviewers ask this: The two approaches support different kinds of inference and become complementary when the research question matches the evidence each can produce.
Generative research defines what should be understood or created, while evaluative research judges how well a concept or design works.
- Generative studies explore unmet needs, contexts, mental models, and opportunity areas before the solution space is fixed.
- Evaluative studies test an existing proposition, flow, prototype, or product against usability, comprehension, value, or other criteria.
- The distinction follows the research purpose rather than a specific method, because interviews, observation, and surveys can serve either purpose when designed accordingly.
Why interviewers ask this: Separating discovery from assessment prevents early exploration from being treated as validation and later testing from being mistaken for open-ended discovery.
Attitudinal evidence captures what people report, while behavioral evidence captures what they actually do in an observed or recorded context.
- Attitudinal data reveals beliefs, preferences, expectations, and recalled experiences through statements, ratings, or choices.
- Behavioral data reveals actions, sequences, errors, completion, and usage patterns through observation, testing, or instrumentation.
- The two can diverge because memory, social desirability, stated ideals, situational constraints, habits, and interface conditions affect reports and actions differently.
Why interviewers ask this: A divergence is evidence to investigate rather than a reason to automatically privilege either what participants say or what they do.
Together, the sampling frame, selection process, and screening criteria determine whose evidence a study can credibly represent.
- The sampling frame is the reachable set from which participants are recruited, and gaps in it exclude parts of the intended population before selection begins.
- Selection bias occurs when inclusion or participation is systematically related to relevant traits, producing a sample that distorts the phenomenon being studied.
- Screening criteria operationalize who is relevant, so they should follow the research question, avoid convenient proxies, and exclude only traits that would make the evidence inapplicable.
Why interviewers ask this: Clear documentation of these elements makes the study's coverage and limits visible instead of implying representation that recruitment did not achieve.
Qualitative sample size depends on information power or another criterion appropriate to the method, while quantitative sample size depends on statistical power or estimation precision.
- Qualitative studies may need fewer participants when cases are highly relevant, data is rich, and the scope is narrow, while saturation is suitable only for methodologies that define and use it.
- Quantitative studies need enough observations for the expected effect, variability, confidence level, margin of error, design, and planned subgroup comparisons.
- There is no universal magic number for either approach, so the sample rationale must be stated before interpreting the strength and reach of findings.
Why interviewers ask this: Sample adequacy is a property of the research aim, evidence quality, and intended inference rather than a fixed participant count.
Build themes through a traceable analytic path from observations to codes, patterns, and bounded interpretations.
- Record observations close to the source, attach codes to specific excerpts or events, and keep enough context to audit how each claim was formed.
- Group codes into themes when evidence shows a recurring or explanatory relationship, not merely because labels sound similar or one comment is memorable.
- Preserve negative cases and conflicting accounts, then refine, split, qualify, or reject a theme instead of forcing all evidence into consensus.
Why interviewers ask this: A credible theme explains a supported pattern while showing the context, variation, and counterevidence that limit it.
Triangulation compares evidence across methods, sources, researchers, or contexts to test whether an interpretation remains credible.
- Convergence among sufficiently independent evidence streams increases confidence that a finding is not an artifact of one measure or sample.
- Divergence should trigger examination of context, definitions, timing, segments, and method limitations because it can reveal a conditional pattern.
- Triangulation is not a vote count, so evidence must still be weighted by relevance, quality, independence, and the kind of claim each source can support.
Why interviewers ask this: Its value comes from exposing assumptions and boundary conditions, not from making every source produce the same result.
A strong Jobs to Be Done statement describes the progress a person seeks in a specific situation.
- It starts with a contextual situation or trigger that makes the job relevant, rather than a broad persona or demographic category.
- It states the person's motivation and desired progress, then defines the functional, emotional, or social outcome used to judge success.
- It stays independent of a particular solution, while push, pull, anxiety, and habit can be examined separately when studying a switch.
Why interviewers ask this: This structure keeps the job stable around context and progress without confusing it with a feature request or a demographic profile.
An evidence-based journey map organizes a user's end-to-end experience into stages and ties every claim to research.
- For each stage, it identifies relevant channels and records the actions people take as they move toward their goal.
- It represents supported thoughts and emotions while distinguishing direct evidence, interpretation, variation, and uncertainty.
- It locates pain points and derives opportunities from their causes, impact, and context instead of adding unsupported solution ideas.
Why interviewers ask this: The map is an analytic model of observed experience, so its stages and layers must remain traceable to evidence rather than workshop assumptions.
A journey map models the user's experience over time, while a service blueprint models the service system that produces that experience.
- The journey map centers stages, goals, actions, thoughts, emotions, channels, and pain points from the user's perspective.
- The blueprint adds frontstage interactions visible to the user, backstage employee actions hidden from view, and supporting processes, policies, systems, and dependencies.
- Linking user steps to frontstage and backstage layers helps identify operational contributors to an experience problem that still require validation.
Why interviewers ask this: The artifacts can share a timeline, but a blueprint extends beyond experience evidence to expose service dependencies and hypotheses about operational contributors.
Define observable success criteria before the study and apply them consistently.
- Binary success records whether the participant reached the required end state without disqualifying help.
- Partial success captures meaningful progress when predefined milestones are completed but the full goal is not.
- Criteria should specify the expected outcome, allowed paths, acceptable assistance, and conditions that count as failure.
Why interviewers ask this: Predefined criteria make task success comparable across participants while preserving useful distinctions between complete, partial, and failed attempts.
Use these measures together because speed alone does not define a good experience.
- Time on task measures completion duration, but interpretation depends on task complexity, user intent, and the need for careful decisions.
- Error rate tracks incorrect actions or outcomes using a predefined error definition and a consistent counting rule.
- Recovery shows whether users can recognize an error, understand what happened, and return to progress without excessive cost.
Why interviewers ask this: A faster flow can be worse if it encourages mistakes, rushed choices, or poor comprehension, so efficiency must be interpreted with accuracy and recovery.
SUS is a standardized study-level questionnaire that produces a converted usability score from 0 to 100.
- It contains 10 statements rated on a five-point agreement scale, with positively and negatively worded items alternating.
- For odd items subtract 1 from the response, for even items subtract the response from 5, then multiply the total by 2.5.
- Compare the score with relevant benchmarks and context, but account for sample size, respondent mix, product type, and the scale's limited diagnostic detail.
Why interviewers ask this: A SUS score is a converted scale score rather than a percentage, and it summarizes perceived usability without identifying the causes of problems.
Use SEQ for task-level ease and SUS for the perceived usability of the studied system as a whole.
- SEQ is usually asked immediately after each task on a seven-point scale from very difficult to very easy.
- SUS is administered after participants have experienced the relevant system or study flow and answers ten broader usability items.
- SEQ helps compare tasks and locate difficulty, while SUS supports an overall standardized summary and external benchmarking.
Why interviewers ask this: The measures complement each other because SEQ diagnoses perceived ease per task while SUS summarizes the broader experience.
Define events and funnel rules before interpreting conversion.
- Give each event a stable name, clear trigger, owner, required properties, identity rule, and versioning convention.
- State the funnel denominator, step order, allowed time window, and whether users may skip or repeat steps.
- Deduplicate retries and duplicate emissions consistently, and distinguish unique users or sessions from raw event counts.
Why interviewers ask this: A funnel is only interpretable when event meaning, eligible population, ordering, and counting rules remain explicit and consistent.
Use meaningful groups to reveal differences that an overall average can hide.
- Cohorts group users by a shared starting event or period, such as signup month, so behavior can be compared over equivalent lifecycles.
- Segments divide users by relevant attributes or behavior, such as device, experience level, plan, or acquisition source.
- Choose slices from a clear question, report sample sizes and uncertainty, and avoid conclusions from tiny or repeatedly mined subgroups.
Why interviewers ask this: Cohort and segment views can expose distinct experiences, but small or opportunistic slices create unstable estimates and false patterns.
A well-defined A/B test specifies its decision rule, population, assignment, measurement, and analysis before results are examined.
- State a falsifiable hypothesis, eligibility and exposure rules, the proposed change, expected effect, and rationale.
- Define the randomization unit, control and treatment, one primary outcome, guardrails, and the minimum detectable effect.
- Calculate sample size and duration, set stopping and analysis rules, then verify instrumentation and sample-ratio integrity before interpretation.
Why interviewers ask this: Predefining assignment, measurement, power, and analysis reduces biased stopping, ambiguous interpretation, and metric cherry-picking.
Separate evidence that an effect is distinguishable from noise from judgment about whether the effect matters.
- Statistical significance evaluates compatibility with a null hypothesis under stated assumptions and does not prove importance or truth.
- A confidence interval shows a range of effect estimates supported by the method and communicates precision better than a threshold alone.
- Effect size quantifies the magnitude of change, while practical significance weighs that magnitude against user value, costs, risks, and context.
Why interviewers ask this: A tiny effect can be statistically significant in a large sample yet too small to justify a product change.
Correlation shows that variables move together but does not isolate the reason for the relationship.
- Reverse causality may apply when the outcome influences the measured behavior rather than the behavior causing the outcome.
- Confounders such as user experience, device, acquisition source, seasonality, release timing, or task difficulty can affect both variables.
- Selection bias, survivorship bias, instrumentation changes, and concurrent product changes can also produce misleading associations.
Why interviewers ask this: A causal claim requires a design and assumptions that rule out credible alternative explanations, not merely a strong association.
Use HEART dimensions selectively, then map each chosen goal to observable signals and operational metrics.
- Happiness covers attitudes and satisfaction, Engagement covers depth or frequency of involvement, and Adoption covers uptake by new or existing users.
- Retention covers continued use over time, while Task success covers effectiveness, efficiency, and error-related performance on intended tasks.
- For each relevant dimension, state a user-centered goal, identify behaviors or attitudes that signal progress, and define metrics with populations, windows, and data sources.
Why interviewers ask this: The goals-signals-metrics mapping prevents teams from collecting convenient metrics that do not represent the user outcome they intend to improve.
Locked questions
- 21
How should taxonomy, ontology, and metadata be distinguished when designing information architecture?
designinformation-architecture - 22
What roles do global, local, contextual, and utility navigation play in an interface?
navigationtypes - 23
How do hierarchical and faceted navigation differ, and when does each improve findability?
navigation - 24
How should search and browse be supported as different finding behaviors?
- 25
What do card sorting and tree testing evaluate, and what can neither prove alone?
information-architecturedecision-makingtesting - 26
How do controlled vocabulary, labels, synonyms, and canonical terms work together?
forms - 27
What are the benefits and risks of polyhierarchy and cross-linking in information architecture?
information-architecturearchitecture - 28
How should a UI state model distinguish data, interaction, network, and permission state?
state - 29
Which asynchronous states should a UI distinguish beyond a generic loading state?
asyncgenericsstates - 30
How do optimistic and pessimistic updates differ, including rollback and duplicate-action risks?
lockingrollback - 31
How should a designer choose between error prevention and error recovery?
recoverydesign - 32
What are the core parts of a microinteraction?
- 33
How should a branching multi-step form communicate and manage its structure?
formscommunication - 34
How should validation timing differ across input, blur, submit, and asynchronous checks?
formsasyncvalidation - 35
What interaction fundamentals should a usable data table provide?
tables - 36
When should pagination, Load more, or infinite scroll be used?
pagination - 37
How are primitive, semantic, and component tokens organized in a design system?
componentsdesign-systemsystem-design - 38
How can component variants avoid a Cartesian product explosion?
variantscomponents - 39
What should design-system component documentation include?
system-designdesigncomponents - 40
When should a team consume an existing design-system component versus contribute a change?
system-designdesigncomponents - 41
How do semantic versioning, deprecation, and migration apply to design system components and tokens?
deprecationsystem-designdesign - 42
What does design-code parity mean between a Figma library and an implemented component library or Storybook?
componentsfigmastorybook - 43
What are WCAG conformance levels A, AA, and AAA, and what does conformance apply to?
wcaga11y - 44
Which concepts determine how an interface appears in the accessibility tree?
a11ytypes - 45
Why should native HTML semantics be preferred to custom ARIA patterns?
htmla11y - 46
What should an assistive technology testing matrix cover, and what are the limits of automated checks?
assistive-techtesting - 47
How should live regions communicate dynamic updates accessibly?
communication - 48
How do Figma variables, collections, modes, scopes, and aliases work together?
figma - 49
How do prototype variables, conditional actions, and interactive components support state modeling in Figma, and where are their limits?
figmacomponentsprototypes - 50
How should a pattern adapt across web, iOS, and Android while preserving the same user intent?
- 51
How would you design an invitation and permissions flow for a 200-person workspace with Admin, Editor, and Viewer roles, where invitations can be pending, expired, domain-blocked, or accepted into the wrong account, and how would you validate that users understand the access they grant?
flowsdesignvalidation - 52
How would you design a subscription plan change when an account can move from a $49 monthly plan to a $399 annual plan immediately, downgrade at renewal, or lose seats and features, while taxes and prorated credits vary by billing date?
design - 53
How would you design saved views, filters, and bulk actions for a work queue of 20,000 cases with 14 filter dimensions, mixed permissions, and records that may change while selected?
controlsdesigndata-structures - 54
How would you design notification preferences across email, mobile push, and in-app channels when security alerts are mandatory, weekly digests require email, quiet hours affect push, and users currently miss critical assignment updates?
responsivenotificationsdesign - 55
How would you diagnose and redesign a collaborative editor where two people can edit the same section, one may go offline for 15 minutes, and users report overwritten content without knowing whether autosave succeeded?
- 56
How would you design an interruptible identity verification flow with document capture, selfie matching, and manual review when 35% of users pause midway and many resume on another device?
flowsdesignidentity - 57
How would you design a new case-summary feature when support agents want a fast two-line scan, compliance reviewers need source evidence, and interviews conflict with usage data showing that both groups frequently inspect full histories?
evidencedesign - 58
How would you design a responsive master-detail workflow for triaging 300 daily tickets so desktop users can scan and act quickly while mobile users can open details, return to the same list position, and retain filters and an unsent reply?
responsiveworkflowsdesign - 59
How would you design progressive disclosure for an expert export configuration with 40 options, five interdependencies, reusable presets, and a requirement that occasional users can produce a valid export without understanding every setting?
progressive-disclosuredesignconfig - 60
How would you design team deletion when the owner must transfer ownership of active projects and billing, seven members will lose access, audit exports remain available for 30 days, and deletion becomes irreversible after a 72-hour grace period?
ownershipdesign - 61
An onboarding funnel shows a sharp drop before the first successful outcome. How would you build a mixed-method research plan to diagnose it?
onboardingfunnelflows - 62
A workflow product has buyers, administrators, and end users with different goals. How would you plan research across these roles?
workflows - 63
You need participants who recently completed a rare workflow, but the easiest recruits are enthusiastic customers. How would you recruit while controlling selection bias?
participantsbiasworkflows - 64
Interviews, usability tests, analytics, and support data point to different problems in the same flow. How would you triangulate them?
flowsusabilitytesting - 65
Research participants provide contradictory evidence about the same feature. How would you synthesize it without flattening meaningful differences?
participantssynthesisevidence - 66
How would you design a benchmark usability test that can be repeated after a redesign using task success, time, errors, and SEQ?
usabilitydesign - 67
You must concept-test several alternatives, but one is visually polished and the others are rough. How would you avoid visual-fidelity bias?
hypothesis-testing - 68
A new tool may change users' behavior over several weeks rather than immediately. How would you combine a diary study with follow-up research?
diary-study - 69
A tree test has low success on several tasks. How would you diagnose the failures and decide whether the information architecture should change?
information-architecturearchitecture - 70
Users request specific features during research. How would you turn the findings into evidence-backed opportunity statements and testable design hypotheses without treating requests as solutions?
evidencefindingsdesign - 71
A checkout funnel loses 18 percentage points between delivery selection and payment entry; how would you diagnose the drop before changing the interface?
flowscontrolstypes - 72
After a redesign, task completion rises from 68% to 82%, but median time increases by 40% and errors among completers double; how would you judge the result?
- 73
Overall onboarding completion has fallen, and a redesign has been suggested; what cohort and segment analysis would you do first?
flowsonboardingcohorts - 74
Usability testing and analytics reveal twelve issues, but the team can address only three this cycle; how would you prioritize them?
usabilityprioritizationtesting - 75
You are testing a revised account creation flow; how would you design an A/B decision that uses conversion as the primary metric without hiding quality problems?
flowsdesignmonitoring - 76
An experiment is statistically significant after analysts checked it every day, but the conversion increase is only 0.2 percentage points; what should you do?
experimentssignificanceconversion - 77
Support has hundreds of tickets about a confusing upload flow, including follow-ups and copied reports; how would you turn them into usable evidence?
evidenceflows - 78
Users report that search rarely finds the right item; how would you use behavior data and findability research to improve it?
research - 79
Users say a save action feels slow and often click it repeatedly, although the server usually responds in under one second; how would you diagnose and fix this?
- 80
An accessibility audit finds many defects across checkout; how would you decide which ones to fix first?
a11ydefectsflows - 81
How would you extend a component library for a dense, domain-specific control without creating a one-off fork?
components - 82
How would you audit local style and token drift, then migrate affected designs and code safely?
tokensdesigniac - 83
How would you structure semantic tokens for light, dark, and high-contrast themes?
contrasttokensa11y - 84
How would you propose, document, and test a new component across Figma and Storybook?
componentsstorybookfigma - 85
How would you deprecate a component variant while giving consumers a safe migration path?
variantsdeprecationmigrations - 86
How would you build an advanced Figma prototype with variables and conditions that includes explicit reset and error paths?
figmaprototypesprototyping - 87
How would you choose ProtoPie or Framer when Figma cannot model required sensor, data, or timing behavior, and how would you state the prototype's limits?
figmaprototypesprototyping - 88
How would you specify an accessible ARIA grid, and when is it justified instead of a native table?
grida11yaccessibility - 89
How would you specify an accessible ARIA tree view with selection, expansion, and typeahead?
a11ycontrols - 90
How would you specify accessible drag-and-drop with a keyboard equivalent, instructions, status updates, and meaningful targets?
keyboardreporting - 91
You are handing off a file import flow whose API can return queued, processing, completed, failed, canceled, and expired states, with polling and retry behavior. What do you provide to engineering?
concurrencydata-structuresapi - 92
A booking flow assumes a selected slot is reserved while the user enters details, but the backend can only confirm availability at final submission. How do you revise the flow without silently weakening the user outcome?
flowscontrols - 93
A Figma approval flow lets every reviewer reject a request, the API permits rejection only for administrators, and the delivery ticket is ambiguous; how would you resolve the conflict before implementation?
figmaapiflows - 94
Before releasing a three-step document export flow, how do you define analytics events and properties with the PM and engineer so the implementation is testable?
flows - 95
A component has newer variants in Figma, older examples in Storybook, and different spacing tokens in the shipped code. How do you restore parity and clarify source of truth boundaries?
componentsstorybooktokens - 96
The same identity verification workflow must ship on web, iOS, and Android. How do you keep the workflow consistent while using native platform conventions?
workflowsidentity - 97
You are handing off a responsive analytics chart, but the chosen chart library has limited labeling and keyboard support; how would you preserve comprehension and accessibility?
keyboarda11yresponsive - 98
A proposed mobile reorder interaction depends on dragging nested items while the list scrolls, but gesture conflicts and technical limits are unknown. How do you run a feasibility spike with an engineer?
responsive - 99
For a file upload dialog with progress, cancellation, errors, and completion announcements, how do you write accessibility acceptance criteria and a manual assistive technology test matrix?
assistive-techa11yresilience - 100
A release deadline threatens a document submission flow because rare error states and optional batch actions are unfinished. How do you reduce scope while preserving the core task and creating explicit follow ups?
flowsstatesbatch