Canonical missing or points off-site

On every live page we look for <link rel="canonical"> and verify one single thing: that its hostname matches the audited domain. The finding is raised when the element is missing, is not a valid URL, or points at another domain.

An important clarification: we do not check that the canonical points at this exact page. A canonical pointing at another page of the same site is a legitimate pattern: it is precisely how parameterised and duplicate variants are handled. A finding there would be a false alarm.

What the check does not do:

• It does not verify the canonical's target exists. A canonical pointing at your own 404 passes.
• It cannot see a canonical injected by JavaScript: Google will process it, we will not find it in the static HTML.
• It does not report contradictions between the canonical, the sitemap and hreflang, even though that causes trouble more often than a missing element.
• It does not check canonicals on pages that are not live content (redirects, errors).
• It does not find out which version of the address the search engine ended up choosing. A canonical is a suggestion, not an order. What Google made of it is shown by URL Inspection in Google Search Console, which prints the canonical it picked next to yours.

The finding is page-level; severity is warning. In the core set a missing canonical is handled by ACC-04, but through a different lens and as one of three conditions. SEO-04 additionally catches a canonical pointing off-site.

For visibility in AI answers we have no documented effect. For classic search, though, Google documents the canonical and is precise about it in a way worth quoting: rel="canonical" is to Google a strong signal, not a directive, and without a declared value it picks a canonical itself.

Two things follow that people confuse. A missing canonical does not mean the page drops out of the index: it means you are not steering the choice. And a declared canonical does not mean Google will obey it: when the content says otherwise, it decides for itself.

A canonical pointing at another domain is a different matter and more serious. You are saying “this page belongs elsewhere”, and Google treats that as a strong signal. It nearly always happens by accident: a template copied from another project, a deployment from staging, or a plugin remembering the domain from installation.

The practical conclusion: fix an off-domain canonical immediately: it is usually an unintended defect with a large impact. Treat a missing canonical as housekeeping.

Every indexable page should carry a canonical with an absolute URL of the preferred version:

<link rel="canonical" href="https://your-domain/path-to-page" />

• A canonical pointing at itself is correct: it is how you say “this is the right URL”.
• Absolute, not relative. Relative works but breaks quietly when the base path changes.
• With parameters, point at the clean version. Otherwise every campaign mints its own canonical page.
• One canonical per page. Two elements with different targets and Google ignores both.
• Check the domain in the generated HTML, not in the template; this is exactly where an off-site canonical is born.

And one check that catches the most common silent defect: the canonical must be consistent with your sitemap. When the sitemap lists a URL with a trailing slash and the canonical without, you are sending two contradictory signals about one page.