Design QA

On this page 7

Design QA (design quality assurance) is the step in a design workflow in which designers review the implemented product against the approved design and specifications, then report deviations in layout, typography, color, states, interactions and content before release. It closes the gap between what was designed and what was actually built.

Functional testing asks whether a feature works; design QA asks whether it looks, behaves and feels as intended. Even with careful developer handoff, small differences accumulate in spacing, font weights, hover states and responsive behavior, and design QA is the systematic check that catches them while they are still cheap to fix.

What design QA checks

Design QA covers every aspect of the interface that a designer specified and a user perceives. The review compares the running build in a real browser or device with the design file, the component library and the written annotations, not one screenshot with another.

AreaWhat is checkedTypical deviations
Layout and spacingGrid alignment, padding, margins, gapsSpacing off by a few pixels, misaligned edges
TypographyTypeface, size, weight, line height, truncationWrong weight, unexpected wrapping
Color and tokensUse of correct design tokensHard-coded near-match colors
Component statesHover, focus, active, disabled, error, loadingMissing focus style or empty state
Responsive behaviorBreakpoints, reflow, touch targetsOverflow on narrow screens
Motion and interactionTransitions, timing, feedbackAbrupt or missing animation
ContentFinal copy, labels, imagesPlaceholder text left in the build
Areas covered by a design QA review

Design QA overlaps with accessibility review in areas such as focus visibility, contrast and touch target size. Many teams include a short accessibility checklist in every design QA pass rather than treating it as a separate, later audit.

Where design QA sits in the workflow

Design QA works best as a recurring gate inside each delivery cycle, not a single review before launch. The usual position is after a feature is built in a staging or preview environment and before it is merged or released.

  1. The developer marks the feature ready for design review and shares a preview link.
  2. The designer reviews it on agreed browsers, devices and breakpoints.
  3. Each deviation is logged with a screenshot, expected value and severity.
  4. The developer fixes issues and the designer verifies the fixes.
  5. The feature passes design QA and proceeds to release or functional testing.

Preview deployments per branch make this practical, because the designer can review the exact change in isolation. Time-boxing each pass, for example 30 to 60 minutes per feature, keeps the step from becoming a bottleneck.

Design QA is one of several quality activities, and each has a different owner and question. Confusing them leads either to duplicated effort or to gaps where nobody checks the visual result.

PracticeOwnerCore question
Design QADesignerDoes the build match the intended design
Functional QAQA engineerDoes the feature work correctly
Visual regression testingEngineering, automatedDid anything change unexpectedly
Design reviewDesign teamIs the design itself good before building
Usability testingUX researcherCan users complete tasks with it
Design QA and neighboring quality practices

Visual regression tools compare screenshots between builds and flag pixel changes automatically. They are excellent at preventing unintended changes to approved screens, but they cannot judge whether a new screen matches the design in the first place, so they complement design QA rather than replace it.

Logging and prioritizing design QA issues

Useful design QA reports are specific, reproducible and prioritized. A comment such as the button looks wrong wastes time; an issue stating the expected token, the actual value, the breakpoint and the browser can be fixed in minutes.

SeverityDescriptionExampleRelease decision
BlockerBreaks use or accessibilityText unreadable, missing focus stateMust fix before release
HighClearly visible inconsistencyWrong component or colorFix before release
MediumNoticeable to a trained eyeSpacing off by 8 pxFix in current cycle
LowMinor polishSlight alignment of an iconBacklog
Severity levels for design QA issues

Shared design tokens and a component library reduce design QA effort substantially: when developers use the same tokens as designers, entire classes of color and spacing errors disappear, and QA can focus on composition, states and behavior.

Common pitfalls in design QA

Design QA fails most often because of timing, scope or tone, not because of a lack of attention to detail.

  • Pixel perfection as a goal: real browsers render fonts differently; focus on meaningful deviations, not single-pixel noise.
  • Review only at the end: a backlog of issues before launch creates pressure to ship anyway.
  • Desktop-only review: most defects appear on narrow screens and in edge-case states.
  • Undocumented design: if states were never designed, QA cannot claim a deviation.
  • Adversarial tone: design QA is a shared quality step, not a judgment of developers.

Stripped to the essentials

Design QA is the moment when designers check the finished product against their plans. It is like an architect walking through a new house before the owners move in. The architect checks that the windows, stairs and doors match the drawings.

Small differences slip in whenever a design is built. A button may be slightly too large, or a color slightly off. Design QA catches these details before customers see them.

Where it is applied

A product team at a software company ships a new account settings page every two weeks. Each feature branch produces a preview link, and the designer spends about 45 minutes per feature reviewing it on desktop, a tablet and a phone, in both light and dark modes.

On the settings page, the review finds nine deviations: a missing focus style on toggle switches, a hard-coded gray instead of the border token, an error message that wraps badly at 360 px, and several spacing differences. Each is logged with a screenshot, the expected value and a severity, and two are marked as blockers for release.

The developer fixes all blockers and high issues the same day, and the designer verifies them in under 20 minutes. The page ships on schedule, and no visual bugs are reported by users in the following month.