12. Site audit
The site audit checks your website the way a search engine sees it. It follows the links between your pages and reports technical problems: broken links, redirects, missing titles and more.
Running an audit
Open a project and go to Site audit → Start audit. The audit runs in the background with every cron run, and the page updates itself while it works. Depending on your server and the cron interval, about 500 pages take 10 to 30 minutes. You can leave the page; the results are there when you come back. Stop cancels a running audit.
Owners, admins and members can start audits; viewers can read the results.
What is checked
| Severity | Issue |
|---|---|
| Error | Broken page (4xx), server error (5xx), page not reachable, missing title |
| Warning | Redirect chain, internal link over http://, duplicate title, missing or duplicate meta description, missing H1, response slower than 2 seconds, HTML larger than 2 MB |
| Notice | Not linked internally (in the sitemap, but no checked page links to it), redirect, title longer than 60 or shorter than 10 characters, meta description longer than 160 characters, several H1 headings, noindex, canonical pointing elsewhere, fewer than 200 words, blocked by robots.txt |
Title, description, heading and text checks apply only to pages that are meant to be indexed. The pages of a paginated listing (?page=2, ?page=3, …) usually share a title, so they count as one page for duplicate titles and descriptions. Pages with noindex, or with a canonical tag pointing to another address, are listed once as such and not checked further.
Reading the results
- Health score: 100 means that no checked page has an error or a warning. Pages with an error count fully against the score, and pages with warnings only count half. The arrow shows the change since the previous audit.
- Issues by type: how many pages have each issue, and how that changed since the previous audit. Click a type to see its pages and advice on fixing it.
- Linked from: for broken pages, redirects and blocked pages, up to three pages that link to them, so you know where to fix the link.
- All checked pages: every address with its status, title, word count, links, response time and depth (clicks from the home page; "—" when no link from the home page leads there).
- Export CSV: all issues of the audit as a spreadsheet.
The last 5 audits are kept, so you can open an earlier one from the Audits list.
Settings
Project editors can set:
- Pages per audit: 100, 250, 500 (default), 1,000 or 2,000. The audit stops when it has found this many addresses.
- Audit the site every week: a new audit starts automatically 7 days after the last one.
How the crawler behaves
- It starts with your home page and the pages in your sitemap (the sitemap listed in robots.txt, otherwise
/sitemap.xml). Sitemap pages fill at most 80 % of the page limit, so pages that are only linked stay included. - It follows paginated listings up to page 10; products further back are usually in the sitemap.
- It only visits your project's domain, with and without
www., and only over HTTPS. Links to other sites are counted, not visited. - It identifies itself as
SEO-Control-Centerand follows your robots.txt: the rules forSEO-Control-Center, otherwise the rules for*. To exclude parts of your site, add for example:User-agent: SEO-Control-Center Disallow: /shop/cart - It loads one page at a time, so it does not put your server under load.
- It does not log in, submit forms, run JavaScript or load images. Pages that only show their content after JavaScript runs may be reported as having little text.
If the audit reports that your domain does not resolve to a public address, the domain points to a private network address and cannot be checked from your server.