Glossary

Building all spacing and sizing from multiples of 8 (with 4 for fine adjustment). Yields rhythm, cross-platform consistency and fewer arbitrary decisions.

Randomized controlled experiment comparing design variants on a metric. Requires a hypothesis, a pre-chosen sample size and the discipline not to peek early.

Synthesis technique: individual observations on notes, clustered bottom-up into themes. Turns interview transcripts into structured, shareable insights.

A property of an object that defines what actions are possible with it — a flat plate affords pushing. Often confused with the signifier that communicates the possibility.

Text alternative for images. Describes function in context, not pixels; decorative images get empty alt so screen readers skip them.

The first number or option seen biases all later judgments. Why pricing pages lead with the expensive plan and why default values in forms are design decisions.

Attributes that add roles, names and states to custom widgets for assistive tech. First rule of ARIA: use native HTML instead when it exists.

Frost's mental model for composing UI: atoms → molecules → organisms → templates → pages. Useful vocabulary for structuring component libraries.

Participants group content items into categories that make sense to them — revealing users' mental categories before you design the IA. Open (users name groups) or closed (groups given).

Grouping information into meaningful units so working memory can hold more. The reason card numbers come in groups of four and long forms come in sections.

Total working-memory effort a task demands. Design reduces extraneous load (navigation, decoding, remembering) so capacity is spent on the task itself.

A group of users sharing a starting event (signup week, first purchase) tracked together over time. Separates product change effects from audience change effects.

A limitation that prevents wrong actions: disabled states, input masks, physical key shapes. Norman's most underrated principle — the best error message is the impossible error.

Field method: observe someone doing real work in their real environment while asking questions as an apprentice, not an interviewer. Surfaces workarounds people never think to mention.

Luminance ratio between text and background. WCAG AA requires 4.5:1 for normal text, 3:1 for large text and UI components; AAA raises text to 7:1.

The share of users who complete a target action out of those who could have. Sensitive to traffic mix — always segment before celebrating or panicking.

A visualization of one persona's experience toward a goal over time: stages, actions, thoughts, emotions and pain points. Alignment tool as much as research artifact.

Interfaces engineered to trick or pressure users into unintended actions: confirmshaming, roach motels, sneaked basket items. Increasingly illegal (EU DSA, FTC actions).

The shared language of a product's UI: components, tokens, patterns, and the usage documentation and governance around them. A living product, not a Figma file.

A named design decision — color.accent, space.200, radius.md — stored platform-agnostically and compiled to CSS/iOS/Android. The atom of a design system.

Participants log experiences over days or weeks in their own context. Captures longitudinal behavior that interviews miss and lab studies cannot reproduce.

The continuous practice of understanding problems and opportunities before and while building — ideally weekly, small touchpoints with users rather than rare heroic research phases.

The Design Council's process model: diverge/converge on the problem (discover, define), then diverge/converge on the solution (develop, deliver). A map, not a conveyor belt.

What a screen shows when there is no data: first use, cleared content, no search results, or an error. Each case needs its own design — first-use empty states are onboarding surfaces.

Research that tests an existing design — sketch, prototype or live feature — against user needs: usability testing, first-click tests, A/B tests, benchmark surveys.

Immediate, informative response to every action: the button depresses, the message sends, the save confirms. Actions without feedback get repeated, doubted or abandoned.

Measuring where people click first for a task. Users whose first click is right complete tasks far more often — making it a strong early signal for navigation and layout.

Time to reach a target grows with distance and shrinks with target size — bigger, closer targets are faster and safer. Explains edge-of-screen menus and generous touch targets.

Keeping keyboard focus inside a modal while it is open and returning it to the trigger on close. Required behavior for accessible dialogs per the ARIA Authoring Practices.

Behavior happens when motivation, ability and a prompt converge (B=MAP). Explains why lowering effort often beats raising motivation — and where nudges become manipulation.

The same fact lands differently depending on presentation: '90% success' vs '10% failure'. Applies to copy in errors, permissions and risk disclosures.

A sequence of steps toward a goal with drop-off measured at each. Tells you where users leave; research tells you why.

Research done to discover problems, needs and opportunities before solutions exist — interviews, field studies, diary studies. Its output is a clearer problem, not a validated screen.

Perception laws — proximity, similarity, closure, figure/ground, common region — describing how visual elements group. The physics under every layout decision.

Google's UX metric families — Happiness, Engagement, Adoption, Retention, Task success — each operationalized through goals → signals → metrics.

An inspection method where 3–5 evaluators independently check an interface against usability heuristics, then merge findings and rate severity. Cheap, fast, and a complement to (not replacement for) user testing.

Decision time grows with the number and complexity of choices. Argument for progressive disclosure and short menus at decision points — not for dumbing everything down.

The structural design of information: organization schemes, labeling, navigation and search. Good IA is invisible; bad IA is why 'it must be somewhere in settings'.

The cues — labels, snippets, icons — users sniff to judge whether a path leads to what they need. Weak scent produces pogo-sticking and back-button loops.

Users spend most of their time on other sites and apps, so they expect yours to work like the ones they already know. The strongest argument for platform conventions.

Framing needs as the 'job' a person hires a product to do — including functional, emotional and social dimensions. Focuses on causality of switching rather than demographics.

How easily users accomplish basic tasks on first contact with a design. One of Nielsen's five usability components; measured with first-time task success.

Comfortable reading needs roughly 45–75 characters per line. Longer lines lose the reader on the return sweep; a max-width on text columns is a usability feature.

Losses hurt roughly twice as much as equivalent gains please. Powers trial-expiry dread and 'don't lose your progress' — and is the mechanism most dark patterns exploit.

The relationship between controls and their effects — stove knobs matching burner layout. Natural mapping removes the need for labels and memory.

What a user believes about how a system works — built from experience with similar systems. Designs fail when the interface model contradicts the prevailing mental model.

The small words that do heavy lifting: buttons, labels, placeholders, errors, empty states. Written from the user's task, not the system's internals.

Working memory holds about 7±2 chunks. Correct application is chunking (phone numbers, grouped forms) — not 'menus must have seven items', a common misreading.

A UI state that blocks interaction with the rest of the app until resolved (modals, alerts). Interruption is a cost — platforms reserve it for required decisions.

The single metric that best captures the value users receive (e.g., weekly active teams, orders delivered). Aligns teams; dangerous when it drifts from real value.

Net Promoter Score: % promoters minus % detractors on the 0–10 'would you recommend' question. Widely used, widely criticized; treat as a trend line, never a diagnosis.

Showing an action as succeeded before the server confirms (likes, sends, renames), with rollback on failure. Trades perceived speed for reconciliation complexity.

People judge an experience by its most intense moment and its ending, not its average. Invest in the hardest step and the finish — confirmations, success states, offboarding.

A composite archetype of a user group built from research data — goals, behaviors, context. Criticized when built from assumptions instead of data; JTBD is the common alternative framing.

Writing the reader understands on first pass: common words, short sentences, front-loaded key information. An accessibility requirement, not a stylistic preference.

Bouncing repeatedly between a list and its items because list-level information is too thin to choose from. A symptom of weak information scent, fixed at the list, not the item.

WCAG's four principles: content must be Perceivable, Operable, Understandable and Robust. A useful audit lens even before reading a single success criterion.

UX design with accountability for business outcomes: a product designer connects user needs with product strategy and metrics, not only with usability.

Show the core by default; reveal advanced options on demand. Keeps first contact simple without amputating power — the interaction-design answer to Hick's law.

People recognize options far more easily than they retrieve them from memory — the basis for menus, autocomplete, suggestions and history, and Nielsen's heuristic #6.

What the team needs to learn, phrased for the team and linked to a decision. Never asked to a participant directly — interview questions are derived from it.

The share of users who return over time, read from cohort curves. The flattening point of the curve is the closest thing analytics has to product-market-fit evidence.

Assistive software (VoiceOver, TalkBack, NVDA, JAWS) that renders UI as speech or braille. Navigates by semantics — headings, landmarks, labels — which is why semantics are design.

A short questionnaire that filters study candidates for recent relevant behavior (not attitudes or demographics alone), without revealing the 'right' answers.

An interview with a prepared guide used as a map rather than a script: fixed goals and themes, flexible order and follow-ups depending on what the participant opens up.

One question asked right after a task — 'Overall, how difficult or easy was the task?' (7-point scale). Cheap, surprisingly reliable per-task difficulty metric.

A journey map extended below the line of visibility: frontstage interactions plus backstage processes, systems and people that produce them.

A perceivable cue — shadow, label, underline — that communicates where and how an action is possible. Norman introduced the term because 'affordance' was being used for the cue rather than the possibility.

A placeholder mimicking the layout of loading content. Beats spinners when structure is predictable and load is noticeable; worse than nothing when it lies about what's coming.

People copy others under uncertainty: reviews, ratings, 'most popular' badges. Legitimate when true; a deceptive pattern when fabricated or pressured ('3 people are viewing this').

A 10-item standardized questionnaire yielding a 0–100 usability score. Decades of benchmarks exist; 68 is the average. Blunt but comparable across products and time.

Kahneman's dual-process model: fast automatic thinking vs slow deliberate reasoning. Interfaces run on System 1 by default — design for glanceability, spend System 2 only where the decision deserves it.

The share of users completing a defined task. The most direct UX metric there is — cheap to collect in any usability test, powerful as a before/after redesign measure.

The controlled classification scheme behind categories, tags and facets. Decides whether 'Jackets' lives under 'Men' or 'Clothing' — and whether users ever find it.

A brief, auto-dismissing, non-blocking notification. Right for low-stakes confirmations; wrong for errors and anything requiring action — it disappears, and screen readers may miss it.

The tappable area of a control — larger than the visible element when needed. Minimums: 44×44pt (Apple HIG), 48×48dp (Material); WCAG 2.2 adds a 24px floor.

Users find items in a text-only version of your navigation hierarchy. Tests findability of a proposed structure without visual design — the evaluative twin of card sorting.

Combining methods or data sources (e.g., interviews + analytics) so the weaknesses of one are covered by another — especially bridging the gap between what people say and what they do.

User interface — the surface through which interaction happens: screens, components, typography, motion. A part of UX, not a synonym for it.

How easy and pleasant a thing is to use, decomposed by Nielsen into five measurable components: learnability, efficiency, memorability, errors and satisfaction.

Watching representative users attempt realistic tasks with a design, to find where and why they fail. Qualitative rounds of ~5 users per iteration surface most recurring problems.

Development driven by understanding users, evaluated with users, iterated — formalized in ISO 9241-210. The philosophical core the rest of this site assumes.

Whether a product provides the features users actually need — as distinct from usability, which asks whether those features are easy to use. Usefulness = utility × usability.

User experience — everything a person experiences with a product or service over time: discovery, onboarding, use, failure, support, leaving. Coined by Don Norman to name something broader than the interface.

Nielsen's heuristic #1: the design keeps users informed about what is going on through appropriate, timely feedback — progress indicators, confirmations, state changes.

Ordering perception by size, weight, contrast, color and position so the eye meets things in the order that serves the task. If everything shouts, nothing is heard.

Voice is the product's stable personality; tone adapts to the moment — playful in onboarding, plain and calm in errors and billing.

The item that differs from its group is the one remembered — the isolation effect. Why the primary CTA looks different from everything else, and why only one thing per screen should.

Knowing where you are, what's here and where you can go — in buildings and interfaces alike. Served by navigation state, breadcrumbs, titles and consistent placement.

Web Content Accessibility Guidelines — the W3C standard behind most a11y law. Organized by POUR principles with testable success criteria at levels A, AA (the common legal bar) and AAA.

Interrupted or incomplete tasks are remembered better than completed ones. The engine behind progress bars, profile-completion meters and 'continue watching'.