A page names more than one canonical address

Why it matters

A canonical tells search engines which address you prefer for a page. Google recommends choosing one method, and warns that using both the HTML tag and the HTTP header invites naming one URL in one and another URL in the other; the canonical link standard asks for only one canonical per resource.

What to do

Look at the finding: it lists every canonical address the page declares.
Decide which address is the main one, and remove the other declarations, whether they come from a second tag, a plugin or a server header.
If the header and the tag must both stay, make them name exactly the same address.

What we read

We read <link rel="canonical"> tags in the head of each crawled page's HTML, without running JavaScript, and any canonical in its Link header. Tags with hreflang, lang, media or type do not count, and neither do tags outside the head.

Before comparing, we ignore differences such as letter case in the host or a #fragment; a trailing slash, www., http against https, the query and the case of the path still make addresses different. A page is reported when two or more different addresses remain; the same address named twice counts once. Pages marked noindex are checked too.

Example & evidence

Illustrative example — one page with a tag and a header, not a live audit.

https://bakery.example/cakes/ answers 200 with this header:

Link: <https://bakery.example/cakes>; rel="canonical"

and this tag in its head:

<link rel="canonical" href="https://bakery.example/cakes/">

The two differ by a trailing slash, so the page declares two canonicals and is reported with both. The page is in the sitemap, so the finding is High.

A possible correction: remove the header, or make it name https://bakery.example/cakes/ exactly; the two then count as one.

Confirm & check again

Check for yourself: view the page source and search for rel="canonical", then look for a Link header in the response headers, for example in your browser's developer tools.

Check again: Full audit — run a new audit of the site. The finding closes when the page is crawled again as a complete HTML page, not only a frame for scripts, with at most one canonical. It can also close when the page answers directly with 200 and a file that is not HTML, or starts with a permanent redirect within your site. A page that answers 404 or 410 cannot be confirmed.

Limits & sources

The finding lists the addresses but not which came from the HTML and which from the header. It does not say which of them Google will use, and canonicals added by JavaScript are not seen.

Confidence: High because the declarations are read directly from the page and its headers.

Sources

Google, How to specify a canonical URL with rel="canonical" and other methods — recommends one canonicalization method, since using the header and the tag together invites naming different URLs in each. Checked 25 September 2026.
IETF, The Canonical Link Relation (RFC 6596) — asks for only one canonical link relation per resource. Checked 25 September 2026.

Editorial review: 29 September 2026

A page names a different page as the main one — one canonical that points to another page, rather than competing ones.
A page names a page on another site as the main one — a single canonical on another host.
Two pages name each other as the main one — two pages naming each other, rather than one page naming two addresses.

A page names more than one main address