Note · 4 min read

What a UX audit should leave you with

The Monday test: can your team open the report, choose the first fix, and explain why it comes first?

Published · Updated

In this note

An audit can identify twenty problems and still leave a team unsure what to do. The missing piece is the judgement between observation and action: what breaks, who it affects, what to change, and why that work comes first.

Before commissioning one, ask the Monday question: could your team open the report and start a useful fix without the author in the room? Agree the pages, journeys and test methods first, then judge the report against that scope.

Evidence, not taste

A finding needs a location and a way to reproduce it. “The hierarchy feels cluttered” gives the team little to investigate. Consider a more useful example: “The Join button opens a login form with no route to create an account.” You can follow that journey, confirm the missing route, and agree the repair. Whether it has reduced sign-ups is a separate question for measurement.

Record the URL, build or review date, device and relevant account state. A screenshot can locate a problem; it cannot establish how a control behaves. Contrast measurements, keyboard checks and screen-reader tests need their own methods and results. Mark checks that were not performed rather than presenting them as failures or passes.

A framework you can interrogate

A published framework gives a finding something firmer than preference. Nielsen's ten usability heuristics provide principles for reviewing interaction. WCAG provides testable accessibility criteria. These do different jobs: a heuristic concern is not automatically a WCAG failure, and a sampled review is not a claim of whole-site conformance.

Ask which version, level and checks the review covers. The method should help the team understand each finding, with enough detail to challenge or repeat it. A long explanation of the framework cannot substitute for that evidence.

The conversion path, mapped

Ask for a map of the calls to action in the agreed journeys: their labels, actual destinations, required account state, and what happens next. A button can look right while leading somewhere its label never prepared the reader for. A contact link can point to an anchor that no longer exists.

Follow those paths in the implementation, including logged-out and failure states. A design file shows the intended destination; a journey check shows where the person actually lands. Record a broken path as a confirmed defect. Estimate its commercial importance from the available evidence rather than assigning an invented lost-revenue figure.

Severity and effort, kept separate

Twenty findings with equal visual weight leave prioritisation to the reader. Ask the reviewer to explain severity, likely reach, confidence and repair effort. Keep these judgements visible rather than hiding them in a single score. Each finding should carry:

  • A severity: the task blocked or made harder, and who may be affected.
  • An effort estimate with assumptions, to confirm with the implementing team.
  • A recommended fix, specific enough to become a ticket without translation.
  • A stated confidence: confirmed on the live site, or needs verification.

A low-effort fix can be a good first task, but effort does not make a serious barrier less serious. Show dependencies and identify anything that needs specialist review. A useful priority order explains both what to do next and what must not disappear into a later backlog.

A summary someone else can forward

The full report is for the team doing the work. The people approving the work need a short decision: what is wrong, what is known about its consequence, and what to fix first. Keep uncertain impact separate from measured results. The forwardable summary is also the practical test of independence: it should stand on its own whether or not the auditor is hired to fix anything.

Make that independence explicit before you commission. Ask whether your own team, or another supplier, can use the report. Findings should remain useful even if you decide against the reviewer's proposal to implement them.

The Monday test

Evidence per finding, a clear review scope, mapped journeys, separate judgements of severity and effort, and a forwardable recommendation. Together they answer the Monday question: what should the team fix first, and why?

This is the shape of our own fixed-fee UX audit: a one-week review of agreed pages and journeys, with evidence and a prioritised action plan. Send us the URL and tell us which task matters most, so we can confirm what the review should cover.

Have a product decision to make?

Tell us what needs deciding and what constrains it. We can discuss whether definition, a build or a reset fits.