Website UX audit: what to check before a redesign

A website UX audit is a structured review of whether people can understand a website and complete important tasks across it. The review examines information, navigation and interaction, then records specific usability findings. Its output should distinguish verified barriers from interpretations and proposed improvements, with enough evidence to guide the next decision.

An expert audit can identify plausible problems. It cannot establish how often every visitor encounters them or guarantee the commercial impact of a redesign.

Choose journeys before choosing pages

Start with what a suitable visitor needs to accomplish. “Explore the website” is too broad to guide a useful review.

For a hypothetical consultancy, a relevant task might be determining whether the provider supports a particular migration and then requesting a project conversation. The journey could involve a service page, supporting information and a form.

Name the entry point and the decision the visitor needs to make. Include the action after the first click, rather than stopping when a visible button appears.

Choose a manageable sample of important journeys. Record which devices and site areas were reviewed so the findings are not mistaken for a complete assessment of everything.

Inspect the path through the website

Review the information and interactions required at each stage.

Nielsen Norman Group's usability heuristics provide broad principles such as familiar language and visible status. Apply them to concrete observations about the journey.

“The enquiry form gives no confirmation after submission” is a finding another person can inspect. “The website feels dated” is an aesthetic judgement that needs a different argument.

Test mobile behaviour through the task

Use a real phone where practical and complete the intended action. A responsive first screen does not establish that navigation and form entry work well.

Inspect overlays and sticky elements while scrolling and entering information. Check whether the on-screen keyboard obscures the next action.

W3C's minimum target-size guidance explains the WCAG 2.2 size requirement and its exceptions. Use the applicable criteria accurately, rather than treating every small control as an automatic failure.

An audit of selected interactions is not a complete accessibility assessment. Record the checks performed and avoid a general compliance claim that the review cannot support.

Give forms their own review

Submit valid information, then examine a plausible input mistake. Check whether instructions help before submission and whether an error identifies what needs correcting.

The W3C forms tutorial recommends clear labels, relevant instructions and feedback on errors or completion. Use those principles to inspect the actual process.

Evaluate each requested input by its purpose. A qualification question may be necessary for a useful response, even when removing it would make the form shorter.

For the hypothetical consultancy, asking about the current platform might support project routing. The audit should check whether the question is understandable and needed at that point.

Measure enquiry quality as well as form completion when evaluating later changes. More submissions do not automatically mean a better commercial journey.

Distinguish observations from explanations

Record what happened before deciding why. “The form rejects a valid phone number containing spaces” is reproducible. “Visitors do not trust the business” is an interpretation that needs additional evidence.

When the uncertainty concerns visitor understanding, use a relevant task-based research session. Avoid showing participants the route you want them to follow.

Our article on session replay discusses another source of interaction evidence. Observed behaviour can refine a question, but it does not reveal motivation by itself.

Keep screenshots and test conditions with the finding. That makes the record useful to the person responsible for investigating or implementing the change.

Decide what deserves work

ConversionNest's view: address verified barriers in important tasks before preferences about visual style. Then consider exposure, confidence and implementation effort.

A broken submission route needs a correction and a functional check. A messaging proposal needs a hypothesis and an evaluation approach. A broad claim about customer priorities may need research first.

Do not let a composite audit score hide those differences. A repeated cosmetic warning can dominate the count while a single failed task has greater practical importance.

Assign each action an owner and an acceptance check. The team should know what evidence is required before marking the work complete.

When is a redesign justified?

A redesign may be justified when the current structure repeatedly prevents important tasks and focused corrections cannot address the shared cause efficiently.

It also brings cost and uncertainty. A visual refresh can consume resources while leaving the offer, missing information or form failure unresolved.

Before committing, ask whether the audit supports the scale of the proposed work. Compare a targeted correction with a broader change using the same business objective.

Start with the strongest verified finding and the appropriate next action. A useful UX audit makes that decision clearer and leaves the team able to learn from the result.