A UX audit is an expert review of a specific user flow: from arriving on the site to the target action, such as submitting an enquiry. It is built on a heuristic evaluation against Jakob Nielsen’s 10 heuristics, carried out independently by 3–5 evaluators who then rate each problem on a 0–4 severity scale. An audit finds obvious problems quickly, but it does not replace research with real users.
After this you will be able to
- Run a two-pass heuristic evaluation of one flow against Nielsen’s 10 heuristics
- Rate each problem on the 0–4 severity scale based on frequency, impact and persistence
- Back up audit findings with funnel data and session recordings
- Write a report: problem, evidence, recommendation, priority
Contents
- When you need an audit — and what it doesn’t replace
- Choosing the flow and goal: from entry to target action
- Heuristic evaluation: the 10 heuristics in practice
- The severity scale, and how not to drown in trivia
- Data: analytics, the funnel, session recordings
- The report: problem, evidence, recommendation, priority
- Example: auditing an enquiry form
When you need an audit — and what it doesn’t replace
An audit is useful when the product exists but the results disappoint: few enquiries, many abandoned carts, the same questions arriving at support. It answers “what in this flow stops people from reaching their goal” and produces a prioritised list of problems. Its method is the heuristic evaluation: evaluators check the interface against a set of usability principles.
An audit is good for
- Quickly catching obvious violations before expensive testing
- Reviewing a new design before handover to development
- Building a prioritised list of improvements
- Preparing tasks for a usability test
An audit doesn’t replace
- Tests and interviews with real users
- A technical audit: performance, code, security
- A WCAG accessibility review — that’s a separate checklist
- Decisions on product strategy and pricing
Choosing the flow and goal: from entry to target action
An audit of “the whole site” is almost always shallow. NN/g recommends narrowing the scope: one task, one section, one user group or one device type — the narrower the scope, the more detailed the evaluation. Businesses usually pick the flow that makes money: an enquiry, a booking, a purchase or a sign-up.
Pick the flow
One path with a clear outcome: “book a consultation from the homepage on a phone”.
Define entry and goal
Where the person comes from (ads, search, Instagram) and what counts as success (enquiry sent and confirmed).
Describe the user and device
Who walks the path and on what: a new customer on a phone behaves differently from a regular on a desktop.
Walk the path and capture the steps
Screenshot every screen and state: empty form, error, loading, success. This is the basis for notes and the report.
Agree on a note format
A spreadsheet with one observation per row, or sticky notes on screenshots. Evaluators don’t see each other’s notes until they finish.
- 01Ad
- 02Service page
- 03Enquiry form
- 04Confirmation
- 05Manager reply
Heuristic evaluation: the 10 heuristics in practice
Jakob Nielsen published the ten heuristics in 1994; in 2020 NN/g refreshed the explanations and examples, while the principles themselves stayed the same. They are broad rules of thumb, not strict rules. Below is what to check for each heuristic in a typical enquiry flow.
1. Visibility of system status
- What to check in the flow
- Is it clear that the enquiry is being sent and has been sent? Is there a step counter in a long form?
2. Match between the system and the real world
- What to check in the flow
- Does the interface speak the customer’s language: “Book” rather than “Create a request”?
3. User control and freedom
- What to check in the flow
- Can people go back a step without losing what they entered, and close the dialog?
4. Consistency and standards
- What to check in the flow
- Do buttons look and read the same across pages? Are the controls familiar?
5. Error prevention
- What to check in the flow
- Is there a phone mask, a date picker instead of typing, a block on past dates?
6. Recognition rather than recall
- What to check in the flow
- Are the chosen service and price visible on the contact step, or must people remember them?
7. Flexibility and efficiency of use
- What to check in the flow
- Does autofill work; is a repeat booking quicker?
8. Aesthetic and minimalist design
- What to check in the flow
- Are there extra fields, banners or copy pulling attention away from the form?
9. Help users recognize, diagnose and recover from errors
- What to check in the flow
- Is it clear which field is wrong and how to fix it — in plain words, no codes?
10. Help and documentation
- What to check in the flow
- Can people quickly get an answer — say, about price or contraindications — without leaving the page?
First pass — getting familiar
Walk the whole flow without judging: learn the product and the logic of the path.
Second pass — finding violations
Walk the path again and log each violation with the heuristic number, a screenshot and, optionally, a recommendation.
Consolidation
Evaluators merge findings, group similar ones and discuss disagreements: what hurts the user and the business goal most.
Any single evaluator, however experienced, will miss some issues. NN/g recommends that three to five people evaluate the same interface independently. In a small team that can be a designer, a developer and a project manager — after a short practice run on a simple example.
The severity scale, and how not to drown in trivia
An audit easily becomes a list of a hundred remarks where a typo sits next to a broken button. So that the team knows what to fix first, every problem is rated on Nielsen’s severity scale.
0
- Meaning
- Not a usability problem
- What to do
- Leave it off the improvement list
1
- Meaning
- Cosmetic problem only
- What to do
- Fix if extra time is available
2
- Meaning
- Minor usability problem
- What to do
- Low priority
3
- Meaning
- Major usability problem
- What to do
- High priority, important to fix
4
- Meaning
- Usability catastrophe
- What to do
- Must be fixed before release
- Frequency: is the problem common or rare?
- Impact: can people get around it easily, or do they get stuck?
- Persistence: once learned, is it no longer an issue, or does it bother people every time?
- Separately — market impact: some problems hurt sales and reputation more than their usability rating suggests
Data: analytics, the funnel, session recordings
A heuristic says what could get in the way. Data shows whether it actually does, and for how many people. If the product is live, attach at least three slices to the audit.
- The flow funnel: how many people reached each step and where the biggest drop is
- Form events: which field triggers the most validation errors and where the form is abandoned
- Session recordings and click maps: what people actually did at the problem step
- Support and messenger requests: recurring questions point to confusing places
- Device comparison: if conversion is noticeably lower on phones, review the mobile version first
Data raises and lowers priorities. A heuristic violation at the step where most people drop off moves up; a debatable remark on a page almost nobody reaches can wait. Analytics shows the place, not the cause: the “why” comes from a test.
The report: problem, evidence, recommendation, priority
A good report takes ten minutes to read and turns straight into tasks. Each row is one problem tied to a step in the flow, with evidence and a concrete fix. “Improve the form UX” isn’t a recommendation; “replace the date field with a calendar picker and block past dates” is.
Step
- What to record
- Screen or state in the flow
- Example
- Enquiry form, error state
Problem
- What to record
- What gets in the user’s way — observable and specific
- Example
- All fields are cleared after an error
Heuristic
- What to record
- Number and name of the violated principle
- Example
- 9. Recover from errors; 3. User control and freedom
Evidence
- What to record
- Screenshot, session recording, funnel figure
- Example
- Before/after screenshots, the error event in analytics
Severity 0–4
- What to record
- Mean of the evaluators’ ratings
- Example
- 3.7
Recommendation
- What to record
- A concrete change that can go into a ticket
- Example
- Keep entered data, highlight only the field with the error
- Up front: the flow, goal, device and who did the evaluation
- The three to five main problems are pulled into a short summary
- Every problem has a screenshot or other evidence
- Problems are sorted by severity, not by screen order
- Recommendations are specific and verifiable
- Items that need a user test are flagged
- Strengths are listed — so they don’t get broken during the rework
Example: auditing an enquiry form
A teaching example on a fictional clinic booking form: fields “Name”, “Phone”, “Service”, “Date” (a text field) and “Comment”, and a “Submit” button. The flow is a new customer booking from a service page on a phone. The severity ratings are illustrative and show how the scale works.
Nothing changes for 3–4 seconds after tapping “Submit”, so people tap again
- Heuristic
- 1. Visibility of system status
- Severity
- 3
- Recommendation
- Disable the button and show a sending indicator, then a confirmation screen
The date is typed as text, the format isn’t hinted and a past day is accepted
- Heuristic
- 5. Error prevention
- Severity
- 3
- Recommendation
- A calendar picker with only available days
An “Invalid input” error above the form; unclear which field is wrong
- Heuristic
- 9. Recover from errors
- Severity
- 3
- Recommendation
- A plain-language message next to the field: “Enter the number as +7…”
The whole form is cleared on error
- Heuristic
- 3. User control and freedom
- Severity
- 4
- Recommendation
- Keep the entered data
The button says “Submit” instead of “Book”
- Heuristic
- 2. Match with the real world
- Severity
- 1
- Recommendation
- A verb that describes the result: “Book a consultation”
An audit is valuable not for the number of remarks, but for how quickly it turns into tasks.
Sources
- Nielsen Norman Group — 10 Usability Heuristics for User Interface Design
- Nielsen Norman Group — Heuristic Evaluations: How to Conduct
- Nielsen Norman Group — Severity Ratings for Usability Problems
- Nielsen Norman Group — 3 Uses for Analytics in User Experience Practice
- Nielsen Norman Group — Error-Message Guidelines