This guide is for product owners, product managers and design or front-end leads who release interface work and keep finding the defects after users do. It sits under Netbase's software product engineering work and complements our UI/UX design service, where design QA is a review of the built screens before each release.
In this guide
- What is design QA, and how is it different from functional testing?
- Which visual and interaction defects should a design QA checklist cover?
- Which design defects should block a release?
- Which design QA tooling fits your team?
- A design QA pass in seven steps
- Where AI fits
- What delivery record exists, and what does not
- Alternatives: other ways to stop interface defects reaching users
- Limits of this guide
- Frequently asked questions
- Next step
What is design QA, and how is it different from functional testing?
Design QA asks one question: does the built product look and behave the way the approved design says it should? Functional testing asks whether the feature works. A checkout can pass every functional test and still ship with a hidden focus ring, a button label cut off at a narrow width, or an error message in the wrong colour.
The two reviews need different evidence. Functional QA uses acceptance criteria and test data; design QA needs the design file, the component library and the built screen side by side. Most teams already run the first. The second often lives in a designer's head, which is why defects reach production.
Design QA is not a usability test either. A small usability round asks whether people can complete a task; design QA assumes the design is right and checks that the build matches it.
Which visual and interaction defects should a design QA checklist cover?
Group the checklist by defect class, so each reviewer knows what evidence to collect. The classes below cover most of what reaches production.
-
Layout and spacing
- What to check
- Margins, padding, alignment, grid columns
- Evidence the check needs
- Design file with measurements beside the built screen
-
Typography
- What to check
- Font family, size, weight, line height, fallback font if the web font fails
- Evidence the check needs
- Type scale of the design system
-
Colour and tokens
- What to check
- Raw colour values where a token belongs, contrast of text and controls
- Evidence the check needs
- Token list and a contrast checker
-
Component states
- What to check
- Hover, focus, active, disabled, loading, empty, error, success
- Evidence the check needs
- A state list per component, written before the build
-
Responsive widths
- What to check
- Overflow, wrapping and navigation at each agreed breakpoint and at 320 CSS pixels
- Evidence the check needs
- The agreed width list
-
Content stress
- What to check
- Long names, missing images, empty lists, translated labels, large numbers
- Evidence the check needs
- A set of test records with extreme values
-
Motion and focus
- What to check
- Animation timing, reduced-motion behaviour, focus order and a visible focus indicator
- Evidence the check needs
- Keyboard walk-through and the motion spec
Two checks carry outside standards. WCAG 2.2 asks that content reflows without two-way scrolling at a width of 320 CSS pixels, as the W3C Understanding Reflow note explains, and that a focused control is not entirely hidden by sticky headers or banners. Pointer targets should be at least 24 by 24 CSS pixels unless an exception applies.
Write the state list before the build. A reviewer who has no list of required states can only report what looks odd, and missing states are the most common defect a design file never showed.
Which design defects should block a release?
Not every pixel difference deserves a blocked release. Rate each defect with four questions, in this order, and stop at the first yes.
- Does it stop a task? A hidden button, a field that cannot be reached by keyboard, a state with no way out. Block the release.
- Does it touch money, data entry or consent? A misaligned price column, an error message that does not show, a checkbox state that reads the wrong way. Block the release.
- Does it break a shared component or a brand rule? A token replaced with a raw value, a button variant nobody designed. Fix before release if the fix is small; otherwise log it with an owner and a date.
- Is it visible drift only? A spacing difference within tolerance, a slightly different shadow. Log it in the backlog.
Agree tolerances in writing before the first review: for example, how many pixels of spacing difference count as a match, and which colour differences are allowed when a value comes from a token. Without written tolerances, every review turns into an argument about taste.
Which design QA tooling fits your team?
| Criterion | Manual side-by-side review | Screenshot comparison in CI | Component visual tests |
|---|---|---|---|
| What it catches | Everything a trained eye sees, including missing states | Unintended change on pages already approved | Unintended change in one component and its states |
| What it misses | Regressions on pages nobody re-opened | New defects on a page with a wrong baseline | Defects that only appear when components combine on a page |
| Upkeep | Reviewer time each release | Baselines per browser and operating system | One story per component state |
| Choose it when | The design is still changing, or the team is small | Releases are frequent and key pages are stable | A design system with coded components exists |
Screenshot comparison is a built-in option in some test runners. Playwright's visual comparison guide shows how a first run writes a baseline image, how later runs compare against it with a pixel tolerance, and why baselines must be produced in the same environment, because rendering varies by operating system, browser and hardware. Storybook's visual tests turn each component story into a test case compared with an approved snapshot.
A fourth option sits beside all three: a lint rule that flags raw colour, spacing and font values in code where a design token belongs. It is cheap and catches drift before anything renders, but it says nothing about how the screen looks.
When not to automate: while a product is still being redesigned, every intended change breaks the baselines, and the team spends its time approving differences instead of finding defects. Start with manual review and a state list, then add screenshot comparison to the pages that have stopped changing.
A design QA pass in seven steps
-
Freeze the reference
Name the design file version and the component library release the build must match.
-
Write the state and width lists
Every component state and every agreed breakpoint, attached to the story or ticket.
-
Prepare stress data
Long names, empty lists, missing images and translated labels in the test environment.
-
Review on a production-like build
Minified assets, real fonts and real images; a local development build hides some defects.
-
Walk the keyboard and the widths
Tab through each flow, then resize through the width list, including 320 CSS pixels.
-
Log each defect with evidence
A screenshot, the expected value from the design and the severity from the four questions.
-
Re-check and record
Confirm fixes on the next build, and feed confirmed defects into the release readiness gate.
Where AI fits
AI tools can take some clerical work out of design QA: pre-sorting screenshot differences into likely noise and likely defects, drafting a defect note from a screenshot, and listing states a ticket does not mention. A person still compares each flagged difference with the design and decides severity, because a model cannot know which drift the team agreed to accept. Netbase works with the major commercial and open-source AI tools and models, chosen per project, under human control.
- In delivered work
Recommendation engine
A recommendation engine delivered for 4over4's online printing store; it is not a design QA tool.
- Available capability
Defect triage assistants
Screenshot difference sorting and defect drafts reviewed by people; not yet tied to a published case.
What delivery record exists, and what does not
Netbase JSC was founded in 2012 and has served 500+ clients. It designs, builds, integrates and runs software across technologies and industries. Its delivery teams combine business analysts, project managers, solution architects, developers, QA and UI/UX designers, so the designer who drew a screen and the tester who checks it work on the same build.
Netbase's delivery lifecycle runs agile execution with outcome-based milestones, which is where a design QA pass before each release fits. In custom development, the client owns the IP created for it, including the design files and component code. One registered record shows a build made against a client's own design: for a large-format printing company, Netbase redeveloped the website's UX, UI and most of the system from the client's workflow, design and front-end specifications, including a pop-up that warns shoppers about low-resolution artwork uploads.
Netbase does not publish a defect count, a defect escape rate or a review duration. This guide explains a method; it is not a case study.
Alternatives: other ways to stop interface defects reaching users
- Design review of pull requests. A designer approves interface changes in code review. Fits small teams with one designer; it does not scale past a few changes a day.
- Stronger design system governance. Fewer one-off components means fewer places for drift. Our design system adoption guide covers how to get engineers to use one.
- Functional regression suites. They protect behaviour, not appearance; they belong beside design QA, as on our quality engineering and testing service.
If the product itself is not yet scoped, start with product discovery and roadmapping. Most Netbase projects are delivered on fixed-price contracts; milestone-based and monthly team retainer models are also offered.
Limits of this guide
- It covers web and mobile interfaces built from a design file; it does not cover print output or video.
- Accessibility checks here are a design QA subset, not a full WCAG 2.2 audit.
- Tool behaviour is described from public documentation at the access date and can change.
Plan the next step with a Netbase consultant
Frequently asked questions
The designer who owns the screen should review it, because only they know which details were intended. A tester can run the state and width lists and log the evidence, so the designer's time goes to judging differences rather than finding them.
Run it on each story as soon as it reaches a production-like test build, not in one pass at the end. Defects found while the developer still has the story open are faster to fix than defects found the day before release.
Usually not at first. A small team with a changing design spends more time approving new baselines than finding defects, so a manual review with a state list pays back sooner; add screenshot comparison once key pages stop changing.
No. It covers the visible checks a reviewer can make on each build, such as focus visibility, contrast and reflow, but a full accessibility audit also tests assistive technology, semantics and content.
Next step
If built screens keep drifting from the design, tell us about the product and your release rhythm and we will reply with a proposal for a design QA pass.
Related services and solutions
AI-enabled software product engineering from discovery to operation Netbase provides software product engineering for companies that need a web, SaaS, mobile or custom platform built and kept running, with AI features and AI-assisted engineering where they pay off. We cover the full lifecycle, from discovery to operation, and the same team stays accountable for architecture, delivery and handover, so nothing gets lost between phases.
Learn More
AI-assisted QA and software testing that run inside delivery Netbase provides software testing and QA services for product and commerce teams to release often without breaking what customers use. QA runs inside delivery, not after it: acceptance criteria, automated regression with AI-generated test drafts, performance budgets and release readiness checks. Page speed shows the result: USticker's page load fell 21%, and PrintLeo's improved 35%.
Learn More
UI/UX design services: research, interfaces and design systems built in code Netbase designs the interfaces of web apps, SaaS products, mobile apps and online stores: user research, flows, wireframes, clickable prototypes and visual design, plus a design system shared by designers and developers. Because UI/UX designers sit inside the delivery team, the same people who design a screen see it built, tested and released.
Learn More
Digital product accelerators for SaaS, mobile and AI products Digital product accelerators are Netbase-built modules and delivery patterns that help founders and product teams launch SaaS, mobile and AI products faster. Instead of rebuilding accounts, billing, workflows and integrations from scratch, you reuse proven components, spend engineering time on what makes your product different, including the AI features buyers now expect, and own the custom code built for you.
Learn More
Work with Netbase remotely
Every Netbase engagement is delivered remotely, so clients in any country and any industry can work with us; Netbase has worked with clients in 80 countries. Contracts and invoices are issued by Netbase JSC, and payment is made by bank transfer to Netbase JSC's company account or PayPal Business. Talk to the CEO, Huy Nguyen (David), on WhatsApp or by email.
Discuss a project
Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.project@netbasejsc.com
+84 937 869 689
91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam
Get in touch
Tell us what you want to build, modernize, or operate.