Behaviour Measured
Where users stop, by task, cohort and device.
UX Audit
UX Audit That Measures Where Users Fail and What It Costs
An expert review produces a list of things a reviewer would have done differently, which is a preference document. A UX audit should produce evidence: where users actually fail, how often, what it costs, and which two or three failures are worth fixing before the rest.
Findings Come With Evidence
A finding without a measurement cannot survive an internal argument, and most of them do not.
Users Observed
Because aggregate data locates a problem without explaining it.
Ranked by Cost
Not by section, and not thirty findings of equal weight.
Accessibility Included
Manually, not only by automated check.
Actionable Output
What to change and what to expect, not a list of principles.
What a UX audit actually inspects
An audit is only worth its cost if it produces findings somebody can act on. General impressions do not qualify. Every item below produces a specific result — a location, a step, a count or a documented failure — so each finding can be verified rather than taken on trust.
Task completionWhether core tasks can be finished, and where they break
Drop-off locationsThe specific steps where users stop, from analytics rather than opinion
Support ticket themesRecurring questions that are interface failures rather than queries
Navigation comprehensionWhether labels match what users expect to find behind them
Form frictionField necessity, validation behaviour and mobile input handling
Error handlingWhether failures explain what happened and what to do
Empty and edge statesWhat the interface shows with no data, partial data or a failure
Mobile parityWhether every task completable on desktop is completable on a phone
Accessibility conformanceContrast, focus order, keyboard operation and screen reader behaviour
Performance impactWhether load or interaction delay is itself causing abandonment
ConsistencyWhether the same action behaves the same way across the product
Content clarityWhether labels and instructions are comprehensible to a first-time user
Support tickets are the cheapest research available and the most consistently ignored. A recurring question is a design failure already documented by the people paying for the product, and three months of tickets usually identifies more real problems than a commissioned study.
UX Audit, Explained
What Does an Audit Examine?
Four sources, and the strongest evidence comes from combining them.
Request an Audit →- 1
Usage Data
What people do, where they stop, which paths they never take.
- 2
Observed Sessions
Real attempts at real tasks, watched.
- 3
Support Volume
What people cannot work out, already categorised.
- 4
Expert Review
Useful for mechanical issues and weakest on comprehension.
What a UX Audit Covers
Evidence gathered, findings ranked, changes specified.
For a website rather than a product, see website design audit. For a marketing site rather than a product, use a website design audit; the accessibility half is accessible website design.
How We Audit a Product
Gather evidence from several sources, then rank by what each failure costs.
Define the Tasks
What the product is for, expressed as things people do.
Measure Completion
How often each task succeeds.
Observe Failures
Watching what happens when it does not.
Rank by Cost
Frequency times consequence, not by severity label.
Specify Changes
Concrete enough to be implemented.
When an audit is the right first step
An audit is diagnostic, and diagnostics are worth paying for when the problem is not yet defined. Where the problem is already known, an audit is a delay before the fix.
Teams who know performance is poor but not why
The clearest case. Analytics show a problem and nobody can locate the cause. The audit converts a general concern into a list of specific, addressable failures, which is what allows work to be scoped and prioritised.
Products before a planned redesign
The most valuable timing and the most often skipped. A redesign without a diagnosis carries every existing problem into a new visual layer and risks discarding whatever was working. An audit first is what makes the redesign brief specific.
Businesses with rising support volume
Where the same questions arrive repeatedly and are being answered individually rather than designed away. Support cost is a direct operating expense, which usually makes the case for this work straightforwardly financial.
Teams inheriting a product
New leadership or a new agency starts with assumptions rather than evidence. An audit substitutes an evidenced baseline for the previous team’s account of things, which is faster and less political than reconstructing the history.
Products facing an accessibility requirement
Where a procurement request, a complaint or a legal obligation has made conformance urgent. This needs a technical conformance review alongside the usability findings, and the applicable standard should be confirmed rather than assumed — VALIDATION REQUIRED.
Method
Why Expert Review Alone Is the Weakest Evidence
A reviewer who understands the product cannot experience it as someone who does not.
What can an expert not see?
Comprehension. Once you know what a label means, you cannot un-know it, and every ambiguity that confuses real users reads as clear.
Expert review is genuinely useful for mechanical problems — inconsistent patterns, accessibility failures, states that were never designed. Those do not require a naive reader.
For anything involving understanding, five observed sessions produce more than any amount of review, because confusion has to be witnessed rather than predicted.
How should findings be ranked?
By frequency multiplied by consequence, not by how serious the problem sounds. A minor irritation on the most-used screen usually costs more than a severe failure in a feature few people reach.
Severity labels attached by the reviewer are a proxy for this and a poor one, because they reflect how bad something looks rather than how much it happens.
Ranking properly usually shortens the report considerably, which is the point — a client can act on three ranked findings and cannot act on thirty unranked ones.
Findings need severity, not just a list
An audit that returns forty observations without ranking them has moved the problem rather than solved it. The team now has to decide what matters, which is the judgement they commissioned the audit to supply.
Useful ranking combines two things that are independent of each other: how much the problem costs, and how much fixing it costs. A defect blocking task completion for every user is severe regardless of how minor it looks. A cosmetic inconsistency is not, however irritating it is to a designer. Sorting by these two axes produces an obvious first tranche — high impact, low effort — that delivers most of the value quickly.
The category that needs the most discipline is low impact and low effort. These are satisfying to fix and change nothing, and a team working through an unsorted list will spend its energy there because the items are easy. Sorting before presenting is part of the deliverable rather than an optional extra.
Expert review and user testing find different things
A heuristic review by an experienced practitioner finds violations of established principles quickly and cheaply: inconsistent patterns, missing feedback, unclear affordances, accessibility failures. It is efficient and it has a specific blind spot — it cannot reveal what users misunderstand, because the reviewer knows how interfaces work.
Watching real users attempt real tasks finds the opposite category: the label everyone misreads, the step where people expect something else entirely, the assumption the product makes that its audience does not share. It is slower and less comprehensive, and it surfaces problems no amount of expert review would predict.
Neither substitutes for the other. An audit relying only on expert review will miss comprehension failures; one relying only on testing will miss systematic issues across screens nobody happened to visit. Where budget forces a choice, the deciding question is whether the product’s problems are more likely to be inconsistency or misunderstanding.
What an audit cannot tell you
An audit describes what is wrong with the current experience. It does not establish whether the product is solving a problem people have, whether the pricing is right, or whether the market is large enough. Teams sometimes commission usability work hoping it will answer a product-market question, and it cannot.
It also cannot tell you what to build instead. Identifying that a flow fails is different from designing the replacement, and the second requires understanding what users are trying to achieve rather than only where they stop. Where the question is what to build rather than what to fix, UX research is the appropriate instrument.
Being clear about this boundary at the outset prevents the common disappointment where an audit is delivered accurately and the client wanted something else. An audit is a diagnosis of the existing thing, and it is extremely useful within that scope.
Questions, answered.
A UX audit examines an existing product to find where users fail — through usage data, observed sessions, support volume and expert review — and ranks the findings by what each failure costs.
An expert review is one person's opinion. An audit combines that with measurement and observation, because a reviewer who understands the product cannot experience it as someone who does not.
By frequency times consequence. A small irritation on the most-used screen usually costs more than a severe problem in a feature few people reach, which severity labels do not capture.
Fewer than most audits produce. Three ranked findings can be acted on; thirty unranked ones cannot, and length is a poor proxy for usefulness.
For a marketing site rather than a product, see website design audit. The methods overlap; the tasks being measured do not.
Because a reviewer who understands the product cannot experience it as someone who does not. It catches mechanical problems and systematically misses comprehension — which is usually what is doing the damage.
It depends on the size of the product and whether user testing is included. A focused audit of a specific flow is considerably faster than a full review of a large application with many roles and states. What lengthens it most is breadth — every additional role, permission level and edge state multiplies the surface to be examined.
It depends on what kind of problem is suspected. If the product is inconsistent and technically flawed, expert review finds that efficiently. If users appear to misunderstand something, only observing them will reveal what. A practical approach is expert review first, with a small number of sessions targeted at whatever the review could not explain.
Findings with locations attached so each can be verified independently, evidence for each — an analytics view, a session recording, a screenshot — and a prioritisation separating what is urgent and cheap from what is structural and expensive. A finding that says the navigation is confusing is not actionable; one that names the labels users misread and the paths they took instead is.
Only where the evidence supports it, and frequently it does not. Many products with poor metrics have specific, fixable failures rather than a fundamentally wrong design, and a report recommending a full redesign in every case is worth reading with the commercial incentive in mind. An honest audit that concludes the structure is sound and three flows need work saves the cost of a rebuild.
Partly, and the limits are worth knowing. Internal teams can gather analytics, group support tickets and check accessibility with tooling — all valuable and often not done. What is difficult internally is seeing comprehension problems, because everyone knows how the product works and cannot unknow it. A practical split is to gather the evidence internally and bring outside judgement to interpret it.
Still deciding if ux audit is right for you?
Talk to UsThe Reviewer Already Knew What It Meant
Expert review is fast, cheap and requires no participants, which is why most UX audits are built on it. A skilled reviewer walks the product and lists what is wrong.
They walk it knowing what every label means, what each screen is for, and where things live — knowledge they acquired in the first hour and cannot set aside afterwards.
Every ambiguity that stops real users reads to them as perfectly clear. It is the one category of problem the method cannot detect, and it is usually the category doing the most damage.
Free Initial Review
Find Out Where Your Product Loses People
Give us access to your product and its usage data. We will measure where tasks fail and rank what is worth fixing.
Claim Your Free Marketing Audit
Enhance Your Brand Potential At No Cost!
- Expect a response within 24 hours
- NDA available upon request
- Dedicated product specialists