Website audit example
A practical website audit example: from findings to a useful backlog
A worked method for turning website evidence into a short, prioritised backlog that a real team can act on.
A useful website audit should help a team decide what to do next. It should not create a 400-line spreadsheet where a broken contact form competes with a missing image description on an old page.
This is the method I use to turn technical checks, search data, analytics and page reviews into one practical backlog. The example is deliberately simple. It shows the shape of the work without pretending that every website has the same problems.
The output in one sentence
Fix the journeys that can lose enquiries, strengthen the pages already close to demand, then improve the publishing system that caused the repeated problems.
Start with the commercial job
Before opening an audit tool, write down the website's job. For a service business it may be qualified enquiries. For a product it may be account creation, activation or a booked demo. This gives every finding a useful question: can this problem stop the website doing its job?
A short baseline might include:
- The primary conversion and the pages that lead to it
- The audiences and search problems those pages need to answer
- The current conversion rate or best available proxy
- The people who publish, approve and maintain the site
- Known launches, migrations or campaigns that change the order of work
A worked findings table
The table below is an example of how raw observations become decisions. Scores are intentionally rough. Their purpose is to force a conversation about impact, confidence and effort, not to disguise judgement as maths.
| Finding | Evidence | Commercial risk | Next action |
|---|---|---|---|
| Demo form fails after a validation error | Reproduced on mobile and visible in form error events | High | Fix and release this week, then test every field state |
| Maintenance guide ranks near page one but has no service path | Search impressions, position data and no contextual CTA | High opportunity | Add a useful calculator and link to the relevant service |
| Service page does not explain what happens after launch | Page review and repeated sales questions | Medium | Add scope, working rhythm and a realistic example |
| Three old images have empty descriptions | Crawl report | Low | Fix during the next content edit, unless the images carry meaning |
Use four evidence buckets
1. Customer journey
Walk through the important journeys on a real phone and desktop. Check navigation, forms, booking, checkout, error states and what happens after conversion. Record the step, the problem and the evidence. A screenshot alone is not a priority.
2. Search and content
Look for pages with real impressions, useful rankings or links before inventing a new content plan. Check whether each page answers the query, contains something original and leads naturally to the next step. Match intent precisely, but do not repeat a phrase simply because a tool says so.
3. Technical quality
Confirm that important pages can be crawled, rendered and indexed. Check canonicals, redirects, duplicate routes, status codes and performance where it affects the experience. A technical issue matters when it blocks discovery, damages trust or makes releases risky.
4. Publishing and ownership
Repeated errors often come from the system around the site. Review the CMS model, reusable components, approvals, analytics ownership and release process. Fixing the cause is usually more valuable than repairing the same symptom every month.
A copyable audit template
Use one row per decision and keep the columns small:
- Page or journey: the exact place where the issue occurs.
- Observation: what is happening, written without a solution.
- Evidence: analytics, search data, testing, customer feedback or a reproducible check.
- Impact: the customer or business outcome at risk.
- Confidence: high, medium or low.
- Effort: a rough delivery size, not a fake estimate.
- Owner and next action: one person and one clear step.
What not to do
- Do not order the backlog by the audit tool's severity label.
- Do not give every page the same checklist regardless of its purpose.
- Do not bury broken revenue journeys below easy technical tasks.
- Do not recommend a redesign when a smaller change can test the idea.
- Do not call a finding complete until the fix is released and checked.
The best audit is not the longest. It is the one that helps a team protect revenue, find a real opportunity and make the website easier to improve next month.
.jpg)
