A useful SEO audit checks whether important pages can be found, understood and used, then turns problems into a prioritised repair list. Work through indexing, site structure, page content, performance, measurement and platform maintenance. Record an affected URL and evidence for every finding; a warning from a crawler is a starting point, not a diagnosis.
1. Establish the scope and baseline
- List the domain, language versions, important page types and business actions to review.
- Export URLs from the sitemap and CMS, then compare them with a crawl. Investigate important pages missing from navigation.
- Record the date, tools and access limitations. Distinguish a public crawl from checks requiring Search Console, analytics or server access.
- Keep a baseline of organic landing pages and outcomes for a stated period. Mark migrations, campaigns and tracking changes that could affect comparisons.
2. Check access, indexing and preferred URLs
| Check | What to inspect | Evidence |
|---|---|---|
| HTTP responses | Important pages return their intended content; broken links and redirect chains are explained. | Source URL, destination and response status |
| Indexing controls | No unintended noindex directive or access block on pages intended for search. | Robots directives, headers and a Search Console inspection where available |
| Canonical choice | Preferred URLs are consistent across page markup, internal links and sitemap. | Declared canonical and Google’s selected canonical when available |
| Rendered content | Main text, links and controls are available in the rendered page. | Rendered view and the missing element, if any |
| Language versions | Language links point to the matching translated page, not an unrelated homepage. | Example language pair and return links |
Do not treat robots.txt as an indexing-removal tool: it controls crawling, and a blocked URL may still appear in results. See Google’s robots.txt guidance. Also distinguish a canonical hint from a redirect or an instruction to delete a page; Google’s canonicalisation guidance explains the available signals.
3. Review the actual answer on each page
- Does the title promise a comparison, calculator, checklist or template that the page actually supplies?
- Can a reader find the main answer without passing through repeated introductions?
- Are claims, prices, units, sources and limitations clear?
- Do related articles serve different questions, or repeat the same generic body?
- Are page titles and descriptions accurate and distinctive? Check headings for a logical reading order.
- Do meaningful internal links lead to useful next steps? Check image descriptions and broken media.
- Does structured data describe the content actually visible on the page?
Use Google’s SEO Starter Guide as a reference for the fundamentals. Improve the answer rather than adding words to meet an arbitrary length target.
4. Measure speed and usability
Check representative templates on mobile and desktop. Review real-user data where available and use lab tests to investigate individual problems. The good Core Web Vitals thresholds are LCP ≤2.5 seconds, INP ≤200 milliseconds and CLS ≤0.1, assessed at the 75th percentile. A single lab score is not that field assessment. See web.dev’s Core Web Vitals definitions.
Inspect oversized images, font loading, render-blocking resources, long-running scripts and unexpected layout movement. Also try menus, forms and any calculators. Record what failed and on which page or device before proposing a fix.
5. Check measurement and WordPress operation
Submit a test through each important conversion path and check whether the expected event is recorded once, under the intended consent state. Compare form submissions with the receiving system. A tag appearing in page source does not prove that reporting is correct.
For WordPress, review available updates, plugin responsibilities, backup restoration, caching and output from the SEO plugin. Check that two plugins are not generating conflicting metadata. Test consequential plugin or theme changes on a copy of the site and verify the affected pages after deployment.
6. Include AI-search eligibility without inventing extra requirements
For Google’s AI Overviews and AI Mode, Google says the existing SEO requirements apply; no special AI file or schema is required. Inclusion is not guaranteed. Check accessible textual answers and source support, and use Google’s AI-features documentation for the scope of that guidance. Do not assume it describes every other AI service.
Turn findings into work
Issue | Affected URLs | Evidence | Reader/business impact | Proposed fix | Owner | Verification | Status
Prioritise blocked important pages, broken conversion routes and incorrect answers before cosmetic improvements. For each repair, define the observable result: for example, a working form, a corrected canonical or a template visible on the promised page. Recheck that result after the change, then monitor search outcomes without treating a short-term fluctuation as proof of causation.
