How to do an SEO audit
An SEO audit should leave you with a fix list, not just a score. Check whether important pages can be reached and indexed, review what they say, and measure their performance separately.
Updated 2 October 2026. For public websites in any country.
Start with the pages that matter
Choose your home page, a product or service page, a useful article and a page that recently changed. Write down what each page is supposed to help a visitor do. A missing social image is not the same priority as an accidental noindex on your main product page.
Run the free website audit for a sample of up to ten public HTML pages. It reports site-level HTTPS, robots.txt and sitemap checks alongside page signals. It requires no account or email. For a specific URL, use the single-page SEO checker.
The crawl is a starting point. It does not test backlinks, Core Web Vitals, actual indexing, JavaScript-rendered content or AI citations.
1. Separate access from indexing
A page needs to be reachable before a crawler can read its content. Review HTTP errors, redirects, login requirements and firewall challenges. Then check robots.txt for the exact crawler and path.
Check indexing separately in Google Search Console's URL Inspection tool for a site you manage. A successful fetch by our tool is not evidence that Google indexed the page. A robots.txt Allow rule is not an indexing instruction, and blocking crawling does not reliably remove a URL from search.
- Confirm that the intended public URL returns the expected page.
- Check meta robots and the X-Robots-Tag response header for noindex or restrictive directives.
- Compare the declared canonical with the canonical Google selected in URL Inspection.
- Check that robots.txt, a CDN or a login wall does not block the content you want discovered.
Our HTML tools report the generic robots meta directive, the X-Robots-Tag response header and the declared canonical link. They do not inspect every crawler-specific directive or validate Google's selected canonical. Use URL Inspection to complete this step.
2. Review the sitemap and internal links
Use your XML sitemap to list the public URLs you want discovered. Check that listed URLs return usable pages rather than errors, redirects or deliberately excluded content. A readable sitemap is not proof that every listed page is indexed.
Reach important pages through ordinary internal links. If the audit checks only one page, first review its fetch limitations and robots rules. It may also have found no usable same-site links in the returned HTML. Do not assume that a short report means the rest of the site is healthy.
A missing sitemap or canonical is not automatically an SEO failure. Review the site's size, duplicate URLs and discovery needs before changing anything. If you need a new file, follow the XML sitemap guide for generation, publishing and submission.
3. Check page meaning, not character-count targets
Read the title, main heading and opening paragraph together. Can someone tell what the page answers and whether it is relevant to them? Review duplicate titles across the sample, vague headings and descriptions that promise something the page does not deliver.
Our checker uses title and description lengths as review prompts, not ranking rules. A title outside a suggested range can still be appropriate. Google may generate a different search title or snippet. Likewise, several H1 elements are not proof of a ranking penalty.
Check the page's visible answer, language, mobile layout and social preview. For factual claims, link to sources and make the organisation or author clear. Structured data should describe content people can actually read on that page.
4. Measure performance and search results separately
Use PageSpeed Insights on the important pages. Its lab measurements and available real-user data answer different questions. No field data is not a passing result; it means that dataset is unavailable for the URL or origin.
In Search Console, review indexing reports and compare search impressions, clicks, queries and pages over a useful period. A traffic change can have several causes. Keep the dates of deployments, content edits and outages so you can compare them with the evidence.
Backlink review, accessibility testing, conversions and content relevance need their own checks. None can be inferred from title tags alone.
5. Turn the report into a fix list
This is a worked example, not a result from a customer's site. Set the priority from the page's purpose and confirmed evidence.
| Observation | Next check | Possible action | Verification |
|---|---|---|---|
| A product page declares noindex | Confirm it is intended for public search and check response headers | Remove the exclusion only if it was accidental | Recheck the live HTML and headers, then inspect the URL in Search Console |
| The same title appears on several service pages | Read each page and confirm they cover different services | Write titles that describe each page's actual topic | Fetch the pages again and compare their titles |
| The sitemap could not be read | Check the declared location and whether the fetch was blocked | Fix the confirmed URL or access problem | Fetch the XML and review Search Console's sitemap report |
| A social image is missing | Check whether link previews matter for this page | Add an accessible image and Open Graph data if needed | Preview the shared link on the relevant platform |
For each action, record the URL, evidence, owner, change date and verification step. Address confirmed access or indexing barriers on important pages before cosmetic review prompts. Rerun the relevant check after publishing. A fix is not complete just because the report was exported.
What this means for GEO and AI search
Google says there are no additional technical requirements for its AI Overviews and AI Mode beyond the underlying Search requirements. A supporting page must be indexed and eligible for a search snippet. Special AI files or a new schema type are not required.
Other providers have their own crawlers and policies. Use the AI crawler access checker to review robots.txt permissions, then the AI SEO technical checklist for the wider checks. Keep search access and model-training policy separate.
Clear answers, source links and crawl access make a page easier to evaluate. They do not guarantee citations. Measure real referrals and useful actions instead of treating a crawler visit as a new customer.
SEO audit questions
Can I do my own SEO audit?
Yes. Start with a public-page crawl, use Search Console to check indexing and search performance for a site you manage, then review important pages and test performance separately. A crawl alone is not a complete audit.
Can ChatGPT do an SEO audit?
It can help interpret an exported report or draft a fix list. Without current crawl data, Search Console access and performance measurements, it cannot verify your site's indexing, rankings or technical condition. Check its recommendations against the actual evidence.
Does passing an SEO audit guarantee AI search citations?
No. Accessible, indexable pages remove possible barriers. They do not guarantee rankings or citations. Google says its AI search features use the same foundational SEO practices and have no additional technical requirements.
Sources and next steps
- Google: AI features and your website
- Google: introduction to robots.txt
- Google: canonical URLs and duplicate pages
- Search Console: URL Inspection
This workflow helps you investigate evidence. It is not a ranking prediction or a guarantee of search traffic.