Most websites do not fail dramatically. They leak. A contact form that confuses a few visitors, a product menu that hides the category people want, a mobile layout where the main button sits below three screens of text. Each issue is small, but together they quietly cost you enquiries and sales. A UX audit is the structured way to find those leaks, rank them and fix the ones that matter. This guide walks through the process we use on business websites, so you can run one yourself or know what to expect from a partner.
What a UX audit is (and is not)
A UX audit is a systematic review of how well your website helps real people complete the tasks they came to do. It combines evidence from analytics, expert review and user observation, then turns that evidence into a prioritized list of improvements.
It is not:
- A visual redesign. Many audit findings are about wording, order and missing information, not colours or fonts.
- A list of opinions. Every finding should point to evidence: a data pattern, a recognized usability principle, or something a user actually struggled with.
- A one-off report to file away. The output is a backlog your team can work through.
Step 1: Define goals and key journeys
Before looking at a single page, agree on what the website is supposed to achieve. For most companies this boils down to a few business goals, such as qualified leads, online orders, distributor enquiries or support deflection.
Then list the key journeys that serve those goals. A journey is a realistic task with a starting point and an end point. For example:
- A procurement manager arrives from search on a product page and needs a datasheet and a quote.
- A returning customer wants to reorder a consumable item on mobile.
- A job seeker wants to find open roles and apply.
Keep it to three to six journeys. An audit that tries to cover everything ends up shallow everywhere.
Step 2: Gather the quantitative evidence
Analytics tells you where people struggle, even if it cannot tell you why. Pull the following for each key journey, ideally covering at least a few months so seasonal swings do not mislead you:
- Entry pages and traffic sources for the journey
- Drop-off between each step (product page to quote form, cart to checkout, and so on)
- Device split, and how the journey performs on mobile versus desktop
- Internal site search terms, especially searches that return no results
- Form abandonment by field, if your tooling tracks it
- Page speed and Core Web Vitals for the pages involved
If your analytics is not set up to show these things, that is your first finding. Many sites track page views but not the events that actually matter, such as form submissions, downloads or clicks on phone numbers. Our analytics and growth service often starts by fixing exactly this, and privacy-friendly options like Dideban Analytics can capture these events without sending visitor data to third parties.
Step 3: Run a heuristic review
A heuristic review is an expert walkthrough against established usability principles. The most widely used set comes from Jakob Nielsen's ten usability heuristics, which cover ideas like visibility of system status, consistency, error prevention and recognition over recall.
Walk through each key journey yourself, on both desktop and a real phone, and note every point of friction. A simple scoring table keeps the review consistent:
| Area | What to check | Example finding |
|---|---|---|
| Navigation | Can users predict where a link goes? | "Solutions" and "Products" menus overlap; users cannot tell which holds the catalogue |
| Content | Does the page answer the visitor's real questions? | Product pages list features but no dimensions or certifications |
| Calls to action | Is the next step obvious and specific? | Three equally styled buttons compete on the homepage |
| Forms | Are fields necessary, labelled and forgiving? | Phone field rejects numbers with spaces or country codes |
| Feedback | Does the site confirm what happened? | Quote form submits with no confirmation message or email |
| Trust | Are credibility signals visible where decisions happen? | No address, certifications or client sectors near the enquiry form |
| Mobile | Does the layout work at thumb reach? | Sticky header covers a third of the screen on small phones |
Rate each issue by severity, for example: blocks the task, causes significant delay or doubt, or minor annoyance.
Do not skip accessibility
Accessibility problems are usability problems for a larger group of people than most teams expect. Include a basic check of colour contrast, keyboard navigation, focus states, image alternative text and form labels. Our guide to accessible web design and WCAG basics covers what to look for.
Step 4: Watch real people use the site
Expert review catches a lot, but nothing replaces watching someone outside your company try to complete a task. You do not need a lab. Five to eight participants who resemble your audience will usually surface the major problems.
A lightweight approach:
- Write task scenarios in plain language, without using the labels on your site ("You need a replacement filter for your unit. Find the right one and ask for a price.").
- Ask participants to think aloud while they work, on their own device if possible.
- Record the screen and audio, with permission.
- Do not help. When someone gets stuck, ask what they expected to happen.
- Note where people hesitate, backtrack, misread or give up.
Session recordings and heatmaps from live traffic are a useful supplement. They show behaviour at scale, although without the "why" that a think-aloud session gives you. Treat recordings of real visitors with care and make sure your privacy notice and consent approach cover them.
Step 5: Review content and microcopy
A surprising share of UX problems are writing problems. Look at:
- Headings that describe your company rather than the visitor's problem
- Jargon that your team uses internally but customers do not
- Button labels like "Submit" or "Click here" instead of "Request a quote" or "Download datasheet"
- Error messages that say something went wrong without saying how to fix it
- Missing answers to questions your sales team hears every week
Ask your sales or support staff for the five questions they answer most often. If the website does not answer them clearly, it is creating work for your team and doubt for your visitors.
Step 6: Prioritize findings
By now you will have dozens of findings. The value of the audit lies in ranking them. A simple impact-versus-effort matrix works well:
- High impact, low effort: fix now. Wording changes, broken links, form validation rules, button hierarchy.
- High impact, high effort: plan as projects. Navigation restructuring, new product page templates, checkout redesign.
- Low impact, low effort: batch them into routine maintenance.
- Low impact, high effort: park them, and revisit if evidence changes.
For each item, record the evidence, the affected journey, the proposed change and how you will know it worked. That last column is what keeps the audit honest.
Key takeaway: An audit is only as useful as its prioritized backlog. Tie every finding to a journey, a piece of evidence and a measurable outcome, then fix the high-impact, low-effort items first.
Step 7: Fix, measure and repeat
Ship fixes in small batches so you can see what moved the numbers. For larger changes, an A/B test or a before-and-after comparison over a comparable period gives you confidence the change helped. Our guide to form design that gets completed is a good companion here, since forms are where many audit findings end up.
Once the major issues are resolved, schedule a lighter review every quarter. Websites drift: new pages get added, campaigns introduce one-off layouts, and plugins or features change behaviour. A short recurring check prevents the slow build-up that made the first audit necessary.
Common mistakes to avoid
- Auditing only the homepage. Most visitors land on inner pages from search or ads. Start where they start.
- Testing only on desktop. If half your traffic is mobile, half your review should be too.
- Letting stakeholders' preferences override evidence. "I don't like that colour" is not a finding.
- Producing a long report and no backlog. If the output cannot be turned into tickets, it will not get done.
- Ignoring performance. A slow page is a UX problem, whatever the design looks like.
Next steps
If you want a second pair of eyes on your site, the DigiVort UI/UX design team runs audits that combine analytics, heuristic review and user testing, and ends with a prioritized backlog your developers can act on. You can also describe your goals through our project wizard and we will suggest a sensible scope for your website.