Refract Group2026
Glossary
UX audit
A UX audit is a structured review of how well a product helps people finish a task: one journey worked through as each kind of user would, checked against the evidence available, and written up as findings ranked for fixing.
- Partial send says CompleteMoney at risk
- Invite opens empty accountBlocks setup
- Filters reset on refreshSlows work
- Card style out of dateCosmetic
UX audit in detail
- Also called
- Usability audit, UX review, Expert review, Product audit
- Topic
- Product
- Updated
What it is
An audit starts with one task written as an outcome: who does it, where they start and what counts as finished. The reviewer sets up test accounts in the states customers are in (new, established, restricted, interrupted) and does the task as each of them would. That includes the failures: an expired invitation, a validation error, a save that breaks halfway.
Each problem becomes a finding that someone else can reproduce. It names the step, what happened, who it affects, the evidence and how confident the reviewer is in the cause. Usage data, support themes and recordings sit beside the walkthrough, and the report keeps what was observed apart from what the reviewer thinks explains it. The result is a short ordered list, and the first few findings each get an owner and a check that shows whether the fix worked.
Why it matters
Teams that redesign before they audit tend to fix the screen people complain about and miss the step that loses them. An audit puts the problems in order by what each costs the user and what it takes to fix, so engineering time goes to the task people cannot finish before the card that looks dated. Because each finding carries an acceptance check, an engineer can estimate it and a reviewer can confirm it was fixed.
How we do it
One way to try working with us is a UX audit of one journey: one week for $2,500, which comes off further work that starts within 60 days. Before the week, on a call, we agree the task, who does it and what you need to decide, and you give us access to the product and whatever evidence you have.
We then do the task as each role would, and record every point where the next step is unclear. Each finding names the screen or step, what happened, who it affects and how sure we are of the cause. We rank the findings by what each costs the person doing the task and how hard it is to fix, then take your team through them.
An example
Say a finance manager sends an approved batch of invoices, and the product says Complete while three are unsent. As a finding, that reads: the task and role; the observation; the evidence, reproduced in a test account with the record IDs attached; the consequence, that the manager may think suppliers are paid; and the confidence, behaviour confirmed and frequency in production unknown. The acceptance check is written before any design. The manager can see the failed items and retry them without resending the rest.
Common mistakes
- Auditing the whole app at once. “Review the app” produces an endless list of observations, where “find out why new account owners cannot finish setup” gives the work a boundary.
- Presenting a hunch as evidence. A reviewer’s explanation, a drop in the analytics and a support complaint are different kinds of evidence, and each finding should say which it rests on.
- Passing a check nobody ran. A check the test environment could not reproduce is untested, and saying so tells the team where the audit has no view.
- Scoring findings to two decimal places. A severity formula built on guesses looks precise and hides its reasons, so a short ordered list with a line of reasoning for each is easier to act on.
UX audit vs usability testing
A UX audit is a reviewer working through the product with the evidence beside them, and usability testing is watching representative users attempt tasks while a facilitator observes.
| UX audit | Usability testing | |
|---|---|---|
| Who does the task | The reviewer, in test accounts | Representative users |
| Finds | What an experienced reviewer can see | What a reviewer could not have guessed |
| Needs | Product access and existing evidence | Recruited participants and a facilitator |
| Produces | Ranked findings with proposed changes | Observed problems and behaviour |
Questions
What people ask about UX audit.
A heuristic evaluation checks an interface against usability principles, most often Jakob Nielsen’s ten heuristics, with three to five evaluators working independently. A UX audit usually includes that kind of check and adds two things. It follows one task from start to finish, and it weighs what the team already knows from usage data and support.
It depends on how much of the product is in scope. One journey takes about a week, and for a trial run with us we audit one journey in a week for $2,500, credited against work that starts within 60 days. A whole product takes longer, because each journey needs its own accounts, walkthrough and findings, so we scope and price it before it starts.
A UX audit delivers a short, ordered list of findings. Each names the task and role, what was observed, the evidence, the consequence for the user, how confident the reviewer is in the cause, the proposed change and a check that shows whether the fix worked. The first few also have an owner.
Where this shows up in our work
Related terms
Working on this now?
Tell us what you’re building and where it’s stuck. We reply within 12 hours.
Most teams start with a call, then a sprint or an audit.
The people you’d work with, from the first call to launch