Site audit score
How the 0–100 readiness score is built from check categories, severities and penalties.
How a page is scored
- Every check belongs to one of nine categories: bot access, meta tags, headings, structured data, content quality, E-E-A-T (trust), AI signals, semantic HTML and performance. A tenth category, agent readiness, is shown in the audit but isn't scored yet.
- Each category starts at 100 and loses points for every check that fails on the page:
| Severity | Points deducted per failed check |
|---|---|
| Critical | 60 |
| Warning | 22 |
| Info | 8 |
- A category can't go below 0. Each failed check is counted once, however many times it occurs on the page.
- The page score is the average of its category scores.
A page that can't be fetched at all scores 0. Markdown and plain-text pages run a reduced set of checks, because checks for HTML-only features such as meta tags or structured data don't apply to them.
Example. A page is missing JSON-LD structured data (one critical) and has two meta-tag warnings. Structured data drops to 40, meta tags to 56, and every other category stays at 100. The page score is the average across all its categories, so one critical issue costs the page far more than a handful of small ones.
Site-wide checks
Some checks describe the whole website rather than one page — crawler access, HTTPS, the sitemap, duplicate titles and descriptions, and links to trust pages such as the privacy policy. They are evaluated once per audit: a site-wide check passes if it passed on any relevant page. They are listed once, under website-wide issues, so one site-level problem doesn't drag down every page separately.
Your audit score
The score on the Site audit page — and next to your brand on the Overview and Competitors pages — is the average of your audited pages' scores. Pages within one site usually score close to each other, so the average is stable between audits and a change in it means something actually changed.
Competitor sites are audited with exactly the same checks, so the scores are directly comparable.
Readiness levels
| Score | Level | Meaning |
|---|---|---|
| 80–100 | Excellent | AI ready |
| 70–79 | Good | Minor fixes needed |
| 60–69 | Fair | Moderate changes needed |
| 50–59 | Poor | Major fixes needed |
| Below 50 | Critical | Immediate action needed |
Severities
| Severity | Meaning | Priority |
|---|---|---|
| Critical | Blocks or severely limits whether AI systems can reach, index, understand or cite the page | Fix first |
| Warning | The page works, but an important signal is missing, weak or inconsistent, and citation confidence suffers | Main backlog |
| Info | Lower-impact polish, completeness or emerging-standard signals | When the blockers are gone |
What the score doesn't measure
- It is not an editorial review. A high score means the site is technically well prepared for AI discovery, not that the writing will win citations. To judge the content itself, use the content review in the AI Assistant.
- It doesn't include search engine indexing. The indexing check on the Site audit page shows whether search engines such as Google and Bing list your pages, but it doesn't change the score.
- It doesn't include agent readiness yet. These checks show whether AI agents can use your site: a Markdown version, AI usage preferences in robots.txt, and names on form fields and buttons. They are listed in the audit, marked "Not scored yet", and don't change any score.