UX Audit

Use this when you're joining an existing product and need to quickly understand where the user experience falls short -- or when it's time for a periodic reassessment. Produces a prioritized gap map with severity ratings, evidence, and recommendations.

Related skills: For WCAG accessibility evaluation (perceivable, operable, understandable, robust), use /accessibility-audit -- this skill covers Nielsen's heuristics, not accessibility standards. For evaluative testing with real users, use /usability-test-plan.

Process

Step 1: Gather inputs

Ask the user to provide:

1. Product description — what it does, who it's for, current state
2. Raw observations — notes from walking through the product yourself (screenshots, flow descriptions, pain points noticed)
3. User data — any available support tickets, NPS responses, app store reviews, analytics (drop-off rates, low-adoption features, time-on-task data)
4. User segments — who are the primary users? Different users experience different gaps.
5. Core flows to assess — which user flows matter most? (e.g., onboarding, core task, checkout, search)

The user can paste raw notes, screenshots, or data exports. If they haven't walked through the product yet, advise them to spend 2-4 hours using it before running the audit.

Step 2: Heuristic evaluation

Analyze the observations against Nielsen's 10 usability heuristics:

## UX Gap Assessment — (Product name, date)

### Heuristic Evaluation

| # | Heuristic | Rating | Key Findings |
|---|-----------|--------|-------------|
| 1 | Visibility of system status | (Good / Issues found) | (Brief summary) |
| 2 | Match between system and real world | (Good / Issues found) | (Brief summary) |
| 3 | User control and freedom | (Good / Issues found) | (Brief summary) |
| 4 | Consistency and standards | (Good / Issues found) | (Brief summary) |
| 5 | Error prevention | (Good / Issues found) | (Brief summary) |
| 6 | Recognition rather than recall | (Good / Issues found) | (Brief summary) |
| 7 | Flexibility and efficiency | (Good / Issues found) | (Brief summary) |
| 8 | Aesthetic and minimalist design | (Good / Issues found) | (Brief summary) |
| 9 | Error recovery | (Good / Issues found) | (Brief summary) |
| 10 | Help and documentation | (Good / Issues found) | (Brief summary) |
| 11 | Accessibility (WCAG 2.2 AA baseline) | (Good / Issues found) | (Brief summary) |
| 12 | Internationalization & cultural fit (if multi-locale) | (Good / Issues found) | (Brief summary) |

Heuristic 11 -- Accessibility quick screen: This is not a replacement for /accessibility-audit (which covers full WCAG 2.2 compliance). This is a baseline check that every UX audit should include:

• Perceivable -- do text and UI elements meet contrast ratios (4.5:1 text, 3:1 components)? Is information conveyed beyond color alone? Do images have alt text?
• Operable -- can all interactive elements be reached and activated via keyboard? Are focus indicators visible? Do touch targets meet minimum size (24x24px WCAG 2.2, 44x44px recommended)?
• Understandable -- are form inputs labeled? Do error messages identify the problem and suggest correction? Is language consistent?
• Robust -- does the heading hierarchy follow h1-h6 order? Are custom components using appropriate ARIA roles?

Flag issues found, note severity, and recommend /accessibility-audit for a comprehensive evaluation if significant issues surface.

Heuristic 12 -- Internationalization & cultural fit (if multi-locale): Skip this heuristic if the product serves a single locale. Include it when the product serves or plans to serve multiple languages or regions.

• Text truncation -- do translated strings overflow containers or get clipped? Check longest locales (German, Finnish).
• Date/time/number formatting -- are formats correct for each locale, or hardcoded to one convention?
• RTL layout support -- if serving Arabic, Hebrew, Farsi, or Urdu: is the layout properly mirrored? Navigation, icons, progress indicators?
• Cultural appropriateness -- do imagery, icons, colors, or metaphors carry unintended meaning in target cultures?
• Text expansion handling -- do UI layouts accommodate 30-40% longer strings without breaking?
• Locale switcher -- is the language/region selector discoverable and functional? Does it use native language names (Deutsch, not German)?
• Input methods -- does the product support IME input for CJK languages, diacritics for European languages?

For each issue found, detail:

• What's happening — describe the problem
• Severity — Critical / Major / Minor / Cosmetic
• Frequency — how often does a user encounter this?
• Evidence — observation + supporting data (if available)

Mobile and touch-specific heuristics

If the product has a mobile or responsive experience, add this evaluation:

### Mobile & Touch Evaluation

| Criterion | Rating | Key Findings |
|-----------|--------|-------------|
| Touch targets (minimum 44x44px, 48x48px preferred) | (Pass / Fail) | (Summary) |
| Thumb zone placement (primary actions in easy-reach zone) | (Good / Issues found) | (Summary) |
| Viewport adaptation (no horizontal scroll, readable without zoom) | (Good / Issues found) | (Summary) |
| Input method fit (appropriate keyboards, no hover-dependent UI) | (Good / Issues found) | (Summary) |
| Gesture discoverability (swipe, pull-to-refresh, long-press are discoverable) | (Good / Issues found) | (Summary) |
| Orientation handling (landscape/portrait transitions work) | (Good / Issues found) | (Summary) |
| Fat-finger tolerance (adequate spacing between tappable elements) | (Good / Issues found) | (Summary) |

Step 3: Cross-reference with existing research

Before layering in quantitative data, check if prior research exists that strengthens or challenges your heuristic findings. Ask the user:

• Do you have past interview transcripts, usability test results, or research syntheses for this product?
• Are there research nuggets or synthesis documents from /interview-synthesis or /research-synthesize?

If prior research is available, cross-reference:

### Prior Research Cross-Reference

| Heuristic finding | Prior research evidence | Effect on severity |
|---|---|---|
| (Gap from heuristic review) | (What interviews/tests revealed about this same area) | Increased / Confirmed / No prior data |

This step matters because heuristic evaluation is expert opinion — prior research adds real user evidence. A heuristic issue that users also struggled with in testing is higher-confidence than one based on expert judgment alone. Conversely, an issue you flagged that users navigated easily may be lower-severity than your review suggested.

If no prior research exists, note this as a gap:

• [NOTE: No prior user research available for this product. Heuristic findings are expert-judgment only. Consider follow-up usability testing to validate severity ratings.]

Step 4: Cross-reference with user data

If the user provided support data, analytics, or reviews, layer it against the heuristic findings:

### Data Cross-Reference

| Finding | Heuristic evidence | Data evidence | Confidence |
|---------|-------------------|---------------|------------|
| (Gap) | (What you observed) | (What the data shows) | High / Medium / Low |

Findings supported by both observation and data are high-confidence gaps.

Step 5: Gap map

Produce the prioritized gap map:

### Gap Map (prioritized)

| # | Gap | Severity | Frequency | User Segments | Evidence | Recommendation |
|---|-----|----------|-----------|---------------|----------|----------------|
| 1 | (What's missing or broken) | Critical | Daily | (Who) | (Sources) | (What to do) |
| 2 | ... | Major | Weekly | ... | ... | ... |

### Quick Wins (high value, low effort)
- (Improvement 1) — (why it's quick, what it unlocks)
- (Improvement 2)

### Deep Investigations Needed
- (Gap that needs user research to fully understand)

### What's Working Well
- (Strength 1 — protect this)
- (Strength 2)

Step 6: Review and refine

Ask the user:

• Do the severity ratings match your experience with the product?
• Any gaps missing from the map?
• Are the quick wins actually quick? (You know the codebase and team capacity better than the agent.)
• Is "what's working well" accurate? (Don't let the audit be entirely negative.)

Adjust as needed.

Portfolio example: B Lab (2020-2021)

Kate audited B Lab's certification platform and SDG Action Manager. Key findings: 55% of SDG Action Manager users hadn't answered a single question, 90% abandon rate, NPS of 26, 50%+ confusion-related support tickets. She identified 9 platform-wide UX problems including IA hidden behind hamburger menus, insider "B" language users didn't understand, too many distracting options post-registration, and no clear path to success. Created lo-fi dashboard mockups showing before/after: stripped irrelevant nav items and guided new users only to the actions that mattered most.

Output location

Present the assessment as formatted text in the conversation for the user to share with the team and convert into backlog items.