What a UX audit is, and what it is for
A UX audit is a structured examination of an existing website, app or system to find where it makes things hard for the people using it. It is commissioned when something is not performing: sign-ups stall, customers abandon a process, support is overloaded with how-to questions, staff avoid an internal tool. The audit explains why, with evidence, and says what to change first.
It is a diagnosis, and it should be kept apart from the cure. An audit that concludes you need the full redesign its authors happen to sell deserves a second look. Often the honest finding is that a handful of specific fixes would recover most of the lost ground, which our comparison of a UX audit and a full product redesign discusses.
The methods an audit can use
| Method | What it involves | What it reveals | Limit |
|---|---|---|---|
| Expert review | A specialist walks through key tasks against established usability principles | Obvious obstacles, inconsistencies, unclear wording | One person's judgement. Needs evidence from other methods |
| Analytics review | Funnels, drop-off points, search terms, device differences | Where people leave and how many | Shows where, not why |
| Usability testing | Real users attempt real tasks while being observed | Why people struggle, in their own words and actions | Small numbers. Takes time to arrange |
| Session recordings and heatmaps | Recorded visits and click patterns | Hesitation, rage clicks, ignored elements | Privacy must be handled, and watching takes hours |
| Support and feedback review | Tickets, chats, reviews, sales objections | Recurring confusion, in customers' terms | Skewed toward people who complain |
| Accessibility check | Automated scan plus manual testing with keyboard and screen reader | Barriers for people with disabilities, and legal exposure | Automated tools find only part of the problems |
A sound audit combines at least one method that shows what is happening with one that shows why. Ask which methods are included before you agree a price. An expert review alone is quicker and cheaper and can be the right choice, as long as it is described as that.
What you should receive
- A summary for decision makers. One or two pages: the main problems, what they are costing and what to do first.
- A findings list. Every issue, each with a location, evidence, severity and recommendation, in a form you can sort and assign, such as a spreadsheet.
- Annotated screens. The problem shown where it occurs, so that designers and developers need no interpretation.
- What works. The parts of the product that serve users well and should be protected from well-meaning change.
- A prioritized plan. Findings grouped into quick fixes, medium changes and larger pieces of work, in a suggested order.
- The raw evidence. Test recordings or notes, the analytics views used, the accessibility results. You paid for them, and your team will want to see them.
- A walkthrough. A session where the auditors present the findings to your team and answer questions.
Ask to see an anonymized sample report before commissioning an audit. Its structure will tell you more about what you will receive than any proposal.
The anatomy of a useful finding
The quality of an audit shows in its individual findings. Compare two ways of reporting the same problem. The first: the registration form is too long and should be simplified. The second names the screen, states that analytics show a large share of visitors leaving at that step on phones, quotes three test participants who stopped at the company registration number because they did not have it to hand, rates the issue as high severity because it blocks sign-up, and recommends making that field optional at registration and requesting it before the first invoice.
| Part | Question it answers |
|---|---|
| Location | Where exactly does this happen? |
| Observation | What goes wrong, described without judgement? |
| Evidence | How do we know? Data, test observations, quotes, a guideline |
| Impact | Who is affected, how often, and what does it cost them or the business? |
| Severity | How does it rank against the other findings, on a stated scale? |
| Recommendation | What specifically should change, and is there more than one option? |
| Effort | Roughly how hard is the fix, as judged with your developers? |
How to judge the audit you receive
A useful audit
- Findings tied to tasks that matter to the business
- Evidence shown for each, from more than one source
- A clear severity scale, consistently applied
- Recommendations specific enough to build
- A short list of what to do first, with reasons
A weak one
- A long list of observations, all given equal weight
- Opinions about taste presented as findings
- Best practice cited with no link to your users
- Recommendations such as improve the navigation
- A conclusion that everything must be redesigned
A worked example
An audit of an insurance quote journey
This example is invented to show an audit narrowing a problem down. An insurance broker's online quote form is started by many visitors and finished by few.
- MethodsAnalytics review of the six-step form, an expert walkthrough, six usability sessions with small business owners and an accessibility check.
- WhereAnalytics show the steepest drop at step three, where annual turnover and payroll are requested, and it is twice as steep on phones.
- WhyFive of six participants did not know their exact payroll figure and were unsure whether an estimate was allowed. Two left to look for documents and did not return. On phones, the numeric fields opened a text keyboard.
- FindingsThirty-one in all. Four rated critical, all on step three and the save-and-return function, which existed but went unnoticed. Nine high. The rest medium or low.
- PlanFirst: say that estimates are accepted, offer ranges instead of exact figures, fix the keyboards and make save-and-return prominent. Later: restructure the order of steps.
The broker had budgeted for a new quote system. The audit showed that four changes to one step addressed the largest loss, and those were made within the existing form before any larger work was considered.
Make sure the audit leads to change
- 01
Review findings with developers
Effort estimates turn a ranked list into a plan. Some high-impact fixes prove trivial, and some small-looking ones do not.
- 02
Assign an owner
One person responsible for carrying the plan through. Audits without an owner are filed.
- 03
Fix the critical items first
Before any redesign discussion. Quick results build support for the rest.
- 04
Record a baseline
Note the measures the audit used, such as completion of a key task, so that the effect of each change can be seen.
- 05
Retest
After the first round of fixes, repeat the usability tasks with a few new participants. Fixes sometimes introduce new problems.
- 06
Keep the findings alive
Move them into the team's ordinary backlog with their evidence attached, so that they are not lost when the report is closed.
UX audit deliverables checklist
UX audit deliverables checklist
0 of 11 checkedWhat to do next
Name the one task in your product that matters most to the business and find its completion rate. If you cannot find it, measuring it is the first step. If you can and it is disappointing, that task is the scope for an audit.
Our UX audit services page describes how we carry out audits and what we hand over. Tell us which journey concerns you and we will suggest which methods fit.
Common questions
How long does a UX audit take?
It depends on scope and methods. An expert review of one journey is quick. An audit that includes recruiting participants and running usability sessions needs longer, mostly for arranging and analysing the sessions. Ask for a plan that shows the time for each method. Scope limited to the journeys that matter most gives the best return.
Do we need usability testing, or is an expert review enough?
An expert review finds many obvious problems quickly and is a sensible first step on a limited budget. It cannot tell you with confidence why users behave as they do, and experts miss things that are specific to your audience. Where an important journey is underperforming, a small number of observed sessions with real users is worth adding.
Who should carry out a UX audit?
Someone with experience of research and design who was not responsible for the product being audited. Fresh eyes matter, because teams stop seeing problems they have lived with. An external specialist or a designer from another part of your organization can both work. Whoever does it should have no stake in recommending a particular solution.
Is an audit worthwhile before a redesign?
Yes, and it is the best time. An audit tells the redesign what must change and what must be preserved, and gives a baseline for judging whether the new design is better. Redesigns begun without that evidence often carry old problems into new screens and discard things that were working.
How many findings should we expect?
The number matters less than the ranking. A typical audit of a few key journeys produces dozens of findings, of which a handful are serious. Be wary of reports that boast of hundreds of issues without priorities. What you need to know is which five to fix first and what happens if you do nothing else.