Key Takeaways
- Check real URLs and evidence before acting on an audit warning.
- Separate technical access, indexing, rankings and conversion measurement.
- Review useful content and overlapping intent before deleting anything.
- Prioritise broken customer journeys and commercially relevant pages over arbitrary scores.
An SEO audit should produce a short list of verified problems and decisions, not a frightening spreadsheet of every notice a crawler can generate. A warning can be important, harmless in context or based on a temporary response. The first job is to establish which kind you are looking at.
This checklist is designed for a UK small business reviewing its own website or checking an agency report. Work on a copy of the findings, keep a dated change log and use staging for changes that could affect the whole site. Do not delete pages or rewrite redirects just to make an audit score look cleaner.
1. Write down the business and measurement baseline
List the services and customer types that matter commercially. Identify the main enquiry or purchase routes and the pages that support them. A business selling a specialist service should not judge all progress by the traffic to a broad informational article.
Export comparable Search Console page and query data for complete periods with the same filters. Record clicks, impressions, CTR and average position, remembering that an average can combine different queries and devices. Keep any release dates or major campaign changes alongside the export.
Check analytics separately. A recorded page view is not a lead, and a form button click is not proof of a successful submission. Reconcile important events with controlled tests and actual enquiry records before using them to judge commercial performance.
2. Confirm the preferred host and successful responses
Test the homepage, a service, an article, a project and an archive. Confirm that each intended live page returns the expected successful response and that alternate host versions reach the preferred destination. Inspect the redirect chain rather than looking only at the final page.
When a crawler reports a 404, re-fetch that exact URL. Determine whether it is linked internally, has useful external links or represents an old page with a close replacement. A fabricated URL may correctly return 404 and need no redirect.
Repair the source of a broken internal link. If an old URL has genuinely moved, use an exact permanent redirect to the relevant replacement. Do not redirect unrelated missing pages to the homepage: that hides the original problem without giving the visitor a useful destination.
3. Check crawl and indexing instructions
Inspect robots.txt, page-level robots directives and response headers. Confirm that important public pages are not accidentally blocked or marked noindex. Check that resources required to render the page are available where necessary.
Robots.txt is not access control for private information. Protect private content with appropriate authentication and authorisation. Also, blocking a URL from crawling does not reliably serve the same purpose as allowing a crawler to see a noindex directive.
Use Google's robots introduction when interpreting rules. Test the actual pattern and affected paths before changing a broad restriction. A rule that appears to protect an internal folder can also prevent access to resources needed by public pages.
4. Inspect a sample of excluded URLs
Use Search Console's Page indexing report to identify the reported reasons. Separate intentionally redirected or alternate pages from URLs you want indexed. Then inspect representative URLs individually, including commercially important ones and the largest losing pages.
Check the canonical, status, rendered content and internal links. A successful live test shows current accessibility; it does not prove that the URL is already indexed or will rank. Likewise, a sitemap entry is not a promise of inclusion.
Do not use a site: search as a complete inventory of Google's index. Keep a URL-level record of the evidence, date checked and next decision. Excluded content may need technical repair, clearer purpose, consolidation or no change, depending on the reason and its value.
5. Review the sitemap for genuine indexable destinations
Submit the canonical URLs you want indexed, excluding redirects, removed pages and future-dated content. Check that the sitemap can be fetched and that its URLs return the right status. An enormous sitemap is not an achievement if it contains unsuitable destinations.
Use meaningful modification dates. Google's sitemap guidance explains that lastmod should reflect a significant update. Regenerating every date on every request makes the field less useful.
Paginated archives can be crawlable and self-canonical without being submitted when the individual articles are already present. Review the site's actual structure rather than assuming every crawlable URL must appear in the sitemap.
6. Check titles, descriptions and headings in rendered HTML
Confirm that a page has one intended title, a useful description and a clear main heading. Look for raw HTML examples accidentally becoming real title, script or image elements inside an article. Code examples should be displayed as examples, not executed or loaded as resources.
An illustrative title element might read: <title>Service and audience | Business name</title>. The correct wording depends on the page; there is no need to use that literal template everywhere. A description should explain the content accurately and make the result understandable to a searcher.
Prioritise titles and descriptions on pages already earning meaningful impressions with weak CTR. Length warnings are prompts to review, not proof of a ranking defect. Do not mechanically lengthen clear descriptions or remove necessary context to satisfy a tool's character count.
7. Test performance and mobile tasks
Use Web Vitals to understand loading, responsiveness and visual stability. Distinguish a laboratory test from field data collected from real users. One slow crawl response does not establish a persistent performance problem.
Repeat representative tests and note the conditions. Check a phone-sized layout, menu, form and any booking or payment journey. A fast homepage score does not prove that the checkout or a long article works well.
For WordPress, use the staged speed-diagnosis guide before installing more optimisation plugins or replacing the platform. Identify the bottleneck and verify the effect of each change.
8. Review content value and overlap
For each priority page, identify the question it answers, the evidence it provides and the commercial purpose it serves. Check claims, citations, dates and examples. Remove invented results and unsupported statements rather than adding more generic copy around them.
Compare pages with similar intent. Keep genuinely different questions separate and link them sensibly. If two pages answer essentially the same question, review search performance and backlinks before merging useful material into the stronger destination.
Do not retire a page merely because it is short or old. Word count alone is not a quality diagnosis. Confirm demand, links, uniqueness and business value before making an irreversible content decision, and preserve a redirect where an appropriate replacement exists.
9. Examine internal links and real customer paths
Check whether important services and supporting articles receive contextual incoming links. A footer link makes a page discoverable, but a useful explanation in a related guide can help the reader understand why to visit it.
Use ordinary links with descriptive text. Avoid relying on JavaScript-only filters to expose all articles or products. Test pagination and make sure important content is not isolated behind an interface a crawler cannot traverse.
Follow the route from an informational article to the relevant service and then to an enquiry. Fix broken links, confusing labels and inconsistent promises. A topic cluster should support that decision journey, not just increase the number of links on each page.
10. Review local and external signals proportionately
If the business is eligible for a Business Profile, check the real name, address arrangement, categories and contact routes. Use the local SEO plan for the broader approach and the review guide for an honest feedback process.
Inspect relevant referring domains and broken legacy targets. A high backlink total dominated by one repeated footer credit is not the same as a broad set of editorial recommendations. Do not buy bulk links to compensate for weak content.
Do not disavow links merely because an automated tool labels them toxic or because the anchor text looks unfamiliar. Google's disavow guidance limits the circumstances in which the tool is appropriate. Investigate carefully and obtain specialist advice before taking action that could discard useful signals.
11. Treat AI-search notices as a separate question
An audit may flag a missing AI text file or access restriction for a non-search crawler. Check whether the notice is relevant to your actual objective. It does not automatically mean Google cannot index the website.
Read the AI search and SEO guide for the distinction between documented Google requirements and speculative optimisation. Do not open every training bot, invent structured data or publish mass content to satisfy an unverified checklist.
Turn the audit into a prioritised worklist
Record each finding with an affected URL, reproducible evidence, customer or search impact, owner and acceptance test. Fix broken access and enquiry paths first. Then address commercially important content, useful internal links and verified legacy redirects.
After implementation, retest the changed pages and shared templates. Keep the baseline and compare complete reporting periods rather than promising a ranking rebound from a deployment alone. Some notices will remain intentionally accepted; document why.
MattDarm can help with SEO services, a website audit or website redesign where evidence supports a rebuild. Send us the findings and the business priorities so we can separate urgent repairs from work that can wait.
Frequently Asked Questions
Does every audit warning need fixing?
No. Reproduce the finding and assess its context. Some notices describe intentional behaviour, temporary responses or optional features. Prioritise verified issues affecting important pages and customer journeys.
Does a sitemap guarantee indexing?
No. A sitemap helps discovery and communicates preferred URLs, but Google still evaluates whether to index them. Check URL-level status, content and links rather than repeatedly resubmitting the same sitemap.
Should short articles be deleted?
Not on length alone. Review search demand, backlinks, unique intent and commercial value. A concise useful page may deserve to stay, while a long repetitive page may need substantial improvement.
Should I fix every toxic-link warning?
Do not act on the score alone. Investigate the links and Google's guidance. Disavow is a specialised tool for limited circumstances, not routine cleanup for every unfamiliar referring domain.
What is the most useful audit deliverable?
A prioritised URL-level worklist with evidence, an owner and a clear retest for each change. A higher health score can be useful, but it does not replace functioning customer journeys or qualified enquiries.
