Design quality, made measurable.
The case for running every frame through Veris before it leaves Figma — what problem it solves, how the audit actually works, and what changes for designers, CX analysts, and design leads once it does.
Prepared for design leadership · Veris, the internal audit layer for our Figma files
The problem
Design QA today depends on who reviews, and when.
- Accessibility and usability issues are usually caught by manual review — if there's time for one, and if the reviewer happens to notice.
- What "library-compliant" means quietly drifts from designer to designer, especially as the team grows.
- Handoff often ships with judgment calls baked in — spacing eyeballed, a colour picked "close enough" to a token — that developers then have to guess at or ask about.
- Feedback lives across comments, calls, and memory, not one traceable record tied to the file.
None of this is anyone's fault. It's what manual review looks like at scale — and it doesn't scale with headcount.
The solution
One audit, six lenses, every time.
Veris is a Figma plugin that reads a selected frame or page the way a senior design reviewer would — checking accessibility, layout, typography, colour, usability heuristics, and design-library compliance in a single pass. Every finding is scored, graded A–F, and traceable to a named rule and a named standard. It runs the same way on the first file of the day and the fiftieth.
How it works
From selection to graded report — one pipeline.
Everything through step 5 runs locally, inside Figma. Nothing leaves the machine unless AI review is switched on — and then it's your key, your provider, your choice.
Grounded in named standards
Not opinion. Named, citable standards.
Every finding cites which rule fired and which standard it's grounded in — so a designer can defend a decision, and a lead can audit the audit.
WCAG 2.2
AA/AAA contrast, target size, focus visibility — the same success criteria legal and accessibility review will hold the product to.
Nielsen's 10 heuristics
The classic 1994 usability framework, checked algorithmically instead of recalled from memory in a critique.
Apple HIG & Material 3
Platform-native layout and touch-target conventions for 2024-era mobile and desktop interfaces.
11 Laws of UX
Hick's, Miller's, Fitts's and Jakob's Law, Von Restorff, goal-gradient, peak-end, and dark-pattern / slip-prevention checks.
UX writing
Vague labels, ALL-CAPS tone, jargon, and inconsistent casing — the copy layer, checked like everything else.
Your own library
Registered components, canonical sizes, and design tokens — the standard your team already agreed to.
Who it changes, and how
Built for the whole review chain — not just the designer.
Self-QA becomes something you run before you ask for review, not something you hope you remembered.
- Noticeably fewer round-trips from CX or dev review for issues that were catchable earlier
- Fixes some issues in place, on the canvas, instead of writing a comment about them
- A decision backed by a named standard is easier to defend in critique than "it feels off"
Turns a manual heuristic walkthrough into a structured report — the same six lenses, applied the same way, every time.
- Meaningfully faster to review a file end-to-end, since findings are pre-sorted by severity
- A second AI opinion plus an exportable report standardises how findings reach stakeholders
- Score history shows whether a design is actually improving across iterations — not just a memory of it
Visibility into design-system health across the whole team, not just the files a lead happens to open.
- Can set a Veris pass as a handoff gate without personally re-reviewing every file
- Library and token drift are visible before they compound across a growing roster
- Governance scales with the design org's headcount, not with a lead's review bandwidth
Design library compliance
The library isn't a suggestion — the audit checks it like a spec.
Every instance on a frame is checked against the registered library: is the component recognised? Has it been resized off its canonical dimensions? Was it detached and rebuilt as a plain frame? Are enough of the frame's instances actually pulled from the library at all?
100% library compliance is reachable, not aspirational. These checks are deterministic — they compare instances against registered component keys and canonical sizes, not a subjective read. A file with zero flags here is, by definition, fully library-compliant.
Consistency as the team grows
The bigger the design team, the more this matters.
One designer can hold the whole library in their head. A team of ten, across dozens of files a week, can't — not reliably. Every hand-picked-but-close-enough colour, every resized button, every instance that quietly got detached is a small deviation. On its own, it's nothing. Multiplied across a growing team, it's how design systems erode.
Veris makes every deviation visible at the file level, before it ships — instead of surfacing months later as "why doesn't this screen match the rest of the product."
Consistency stops depending on which reviewer happened to look at the file that week.
Developer handoff
Developers inherit the file's decisions — Veris checks they're the right ones.
A clean handoff isn't just tidy layers. It's colours that are actual library tokens, not close-enough hex values; spacing that matches the grid, not an eyeballed number; components that are still the real component, not a detached lookalike. Veris flags every one of these before the file reaches engineering — and its export (JSON, PDF, or numbered annotations placed directly on the canvas) makes the finding traceable, not a comment thread to dig back through.
100% handoff quality, with developers working from the exact design tokens. When library and token-drift checks report zero issues, every colour, spacing value, and component in the file already is the real design-system token — engineering implements from the exact value, not an approximation of it.
Accessibility
Accessibility becomes a checked property of the file, not a hope.
Contrast is evaluated against every element's real, composited background — including gradients and images, via pixel sampling where metadata alone can't tell — using the same relative-luminance math WCAG itself defines. Touch targets and focus visibility are checked the same deterministic way.
100% WCAG 2.2 AA compliance on every automatically-testable success criterion. The contrast math is exact, not estimated — a frame with zero contrast-category errors has, by definition, satisfied the criteria Veris checks.
A handful of WCAG criteria — like whether alt text is truly meaningful — still need human judgment. Veris tells you exactly where those remain, rather than claiming to cover them.Why this, why now
This is design QA infrastructure, not another plugin.
- Reduces accessibility and usability risk before it reaches customers or legal review — cheaper to catch in Figma than in production.
- Protects brand and design-system consistency as the team scales, without adding a dedicated QA headcount.
- Gives every designer a fast, private second opinion — rules first, AI optional, nothing leaves the machine unless it's switched on.
- Turns "I think this is fine" into a graded, citable, exportable answer.
The ask
Start with one team, one gate.
Pilot Veris with our design team for one review cycle: every frame gets audited before it's marked ready for CX or dev handoff. Track the score trend per file. At the end of the cycle, we'll know — not guess — whether library compliance, accessibility, and handoff quality actually improved.
The truth about your design — before you ship it, not after.