The Complete Guide to UX Audits

What a UX audit is

A UX audit is a structured review of a digital product against a set of usability, accessibility and behavioural frameworks. It produces a prioritised list of findings, each cited against the rule it breaks, each scored for severity, each paired with a recommended fix.

That's the engineering description. The honest description is shorter: a good UX audit is the most economical way to find out what's broken before users do. A senior practitioner runs the audit because they have seen the same fifty problems on a hundred products, can name them in minutes, and can articulate the fix in language the engineering team will accept.

What a UX audit is not: a redesign brief, a feature wishlist, or a vehicle for personal preferences. The findings cite frameworks. The severity scores follow a rubric. The recommendations describe behaviour, not pixels. When an audit drifts into "I would have done it differently" territory, the deliverable stops being useful.

When to run one

Five situations where an audit is the right move. If none of these apply, an audit will produce a document nobody reads.

The wrong reason to run an audit is internal politics. Audits commissioned to settle an argument between design and product, or to validate a CEO's pre-existing opinion, almost always end up in a drawer. Audit the product because the product needs auditing, not because the team needs convincing.

Types of UX audit

Four common types, listed by scope. Most practical engagements blend two or three.

1. Heuristic audit

The classic expert review against Nielsen's 10 usability heuristics or a related framework. Fast, cheap, comprehensive across the surface but shallow on any single issue. The right starting point for most engagements. Covered in depth in the heuristic evaluation guide.

2. Accessibility audit

Scoped against WCAG 2.2 AA (and AAA where relevant). Tests contrast, focus visibility, keyboard navigation, semantic structure, screen reader behaviour, tap target sizing, and the new WCAG 2.2 success criteria. Often the highest-leverage type of audit because the findings are non-negotiable.

3. Conversion or flow audit

Scoped against a single high-value flow (onboarding, checkout, signup, upgrade). Combines heuristic review with analytics, session recordings and the team's qualitative input. The findings tie directly to business metrics, which is why this is the audit most often commissioned by growth teams.

4. Full product audit

End to end: navigation, information architecture, every flow, every state, accessibility, performance, content, mobile parity. Two to four weeks of work. Used before redesigns, replatforming, or post-acquisition. Often packaged with a presentation and roadmap.

The audit process

A defensible audit follows the same six-stage process every time. The shape doesn't change; the depth at each stage does.

Each stage has its own failure modes. Most audits collapse at stage one (scope) or stage six (presenting findings). The middle stages are mechanical if the first one was done properly.

The full methodology lives in the dedicated guide: how to run a UX audit.

Common UX problems audits surface

If you audit fifty products, the same ten problems will appear in forty of them. Knowing the pattern saves time during the audit and adds credibility to the report.

• Primary CTA contrast fails on hover or focus. The default state passes WCAG 1.4.3 and the hover state silently fails. The most common single accessibility finding.
• Form inputs rely on placeholder for the label. Removes label permanence and fails 3.3.2 outright.
• Focus indicators removed or invisible. The team has set outline: none without a replacement. Hard fail under 2.4.7.
• Error messaging surfaces too late. Validation fires on submit rather than on blur, forcing the user to scroll and re-read.
• Heading hierarchy broken. Multiple H1s, H3 nested under H1 with no H2 between. Screen reader users lose the page structure.
• Touch targets below 24 by 24 CSS pixels. WCAG 2.2's new 2.5.8 target size criterion. Most products built before late 2023 quietly fail.
• Inconsistent navigation across sections. The user can't predict where they are or how to get back. Fails 3.2.3.
• Confirmation patterns missing for destructive actions. Account deletion, payment cancellation, file removal. No undo, no confirm. Fails 3.3.4.
• Cookie banners obscure the page until dismissed. Often unreachable by keyboard or screen reader.
• Loading states absent or misleading. The user can't tell whether the app is working, broken, or waiting on them.

None of these are exotic. They appear because product teams move fast, design systems drift, and accessibility lives in someone else's backlog. The audit's job is to make them visible together, in priority order.

Accessibility considerations

Accessibility is the single highest-leverage category in any audit. The reasons are practical, not moral.

• Findings are objective. WCAG citations end debate. A 3.8:1 contrast ratio fails AA. There is nothing to argue about.
• Findings have legal weight. The European Accessibility Act began enforcement in June 2025. UK Public Sector Bodies Accessibility Regulations have been in force since 2018. The cost of inaction is no longer hypothetical.
• Findings overlap with general usability. Most accessibility fixes (clearer labelling, predictable navigation, sufficient contrast) improve the product for every user.

The new WCAG 2.2 criteria added in October 2023 are the most under-checked in current audits. Focus appearance (2.4.13), target size (2.5.8), and dragging movements (2.5.7) are still missing from most teams' acceptance criteria. An audit that surfaces these earns immediate credibility.

The audit tools in this section help: the WCAG contrast checker, the AAA checker, the focus state contrast checker, and the link contrast checker. All free, no signup.

Heuristic evaluation

Most UX audits use Nielsen's 10 usability heuristics as the spine. They've survived three decades because they cover the structural failure modes of digital products with very little overlap.

The ten, in order:

1. Visibility of system status
2. Match between system and the real world
3. User control and freedom
4. Consistency and standards
5. Error prevention
6. Recognition rather than recall
7. Flexibility and efficiency of use
8. Aesthetic and minimalist design
9. Help users recognise, diagnose and recover from errors
10. Help and documentation

Applying them well is the part that takes practice. The dedicated guide breaks each heuristic down with examples, scoring criteria, and the specific patterns that fail each one: the heuristic evaluation guide.

Presenting findings

The deliverable is not the report. The deliverable is the team understanding what to do next. A beautifully formatted audit report that doesn't change behaviour is a failed audit.

Three rules that improve the odds.

Lead with the top three

An audit can produce forty findings. A team can act on three. Open the executive summary with the three highest-severity issues, the business impact of each, and the recommended action. Everything else is appendix.

Show the rule, not the opinion

Every finding cites the framework it violates: "Fails WCAG 2.2 1.4.3" or "Violates Nielsen heuristic 1: Visibility of System Status". Personal preference language ("I think this could be clearer") gets pushed back; framework citations don't.

Run a working session

Don't send the deck. Walk the team through it. Engineering will surface implementation constraints. Product will surface roadmap constraints. The recommendations get sharper. The team gets ownership.

The example audit report shows what good looks like end to end.

Turning audits into roadmaps

An audit is a list of problems. A roadmap is a list of next actions. Converting one to the other is the step most audits skip.

Score each finding on a two-axis grid. The roadmap then writes itself. Most teams find that ten to twelve findings fall into "high impact, low effort" once you do this honestly, and that's enough material for a credible first sprint.

The downloadable templates in UX audit templates include this scoring sheet pre-formatted.

Recommended tools

The tools that earn a place in a real audit. Free unless noted. Where a paid tool is listed, the free tier is usually sufficient for a single audit.

Frequently asked questions

The full UX audit cluster

Each of these pages goes deeper on one section of this guide.