UX Audit: What Actually Moves Conversion, and What Only Looks Broken

A UX audit that returns forty findings is easy to produce and hard to act on. The value is not in the list, it is in the ranking. Three findings usually explain most of the lost conversions, and they are rarely the ones people argue about in meetings.

What a Useful Audit Looks At

Start from evidence, not opinion. Before anyone comments on the design, the audit should establish where people actually drop out and what they were trying to do.

• Entry pages and their bounce behaviour, split by device.
• The path from landing page to the form or cart, step by step.
• Form completion and abandonment, field by field.
• Search and filtering, if the site has a catalogue.
• Speed and layout stability on real mobile connections.

The Findings That Usually Matter

Across company websites the same handful of problems keep appearing, and they are structural rather than decorative.

1. 1The page does not answer what the company actually does within one screen.
2. 2The primary action is not visible without scrolling on mobile.
3. 3The form asks for more than the visitor is willing to give at that stage.
4. 4Proof of competence is missing where the decision is made, not where it is convenient.
5. 5The mobile layout reflows while loading and people tap the wrong thing.

The Findings That Usually Do Not

Plenty of audit items are matters of taste dressed as usability. They are worth noting, but not worth a sprint.

• Rounded corners, shadow depth and font pairing preferences.
• Section order on pages that already convert.
• Micro-copy tweaks with no measurement attached.
• Redesigning what almost nobody sees.

If a finding cannot be tied to a step where people leave, it is a preference, not a problem.

How to Prioritise the Fix List

Score every finding on two axes: how many people it affects, and how much effort it takes to remove. Anything high-traffic and cheap goes first, regardless of how unglamorous it is.

How to Know It Worked

Agree the measure before the work starts. For most company websites it is the number of qualified inquiries per month, not time on page or bounce rate. Fix in small batches so you can still attribute the change.