Setting canonical tags correctly and finding errors

What is a canonical?

Google calls canonicalization the process of selecting one representative URL from several identical or very similar pages. This canonical URL appears in the search results, and the others count as duplicates. With the link element rel="canonical" you tell Google which URL you prefer. The link relation is defined in RFC 6596 from 2012. According to the RFC, the target must contain the same content or a superset of it.

Google names four reasons for a canonical. You decide which URL appears in search, consolidate signals from similar pages, simplify tracking and save crawl time on duplicates.

How strong is the signal?

Google also considers whether a page is served over HTTPS and recommends linking internally to the canonical URL. If you combine several methods with the same target, Google says the chance that your preferred URL appears increases.

Setting the canonical correctly

The element goes in the <head> of the page. Google only accepts it there. The preferred page points to itself, and the variants point to it.

For files without HTML, such as PDFs, you set the canonical as an HTTP header.

• Use absolute URLs with protocol and host.
• Choose one method, HTML element or header. Using both at once makes errors more likely.
• Only one canonical per page, as RFC 6596 requires.
• No fragment such as #section as the target.
• Link internally to the canonical URL.

When do you use a canonical?

The same language for different countries, such as de-DE and de-AT with almost identical text, may be treated by Google as duplicates. For this, Google recommends canonical and hreflang together. In practice, each country version points to itself via canonical and to the others via hreflang. That way Google can show the right URL in each country. How this fits together is explained below and in the hreflang guide.

Canonical via JavaScript

Google recommends delivering the canonical in the HTML. JavaScript should not change it to a different URL. If it can only be set via JavaScript, set it there exclusively and not twice.

In practice, this means that the raw source code and the rendered HTML should carry the same canonical. If they differ, it is unclear which one Google uses.

Canonical and hreflang

A canonical to another URL says that this page should not be in the index. An hreflang says that it should be in the index as a language version. Both on the same URL contradict each other. Three rules follow from this.

1. Each language version points to itself via canonical.
2. Pages with a canonical to another URL get no hreflang.
3. hreflang only points to canonical URLs.

When choosing the canonical URL, Google prefers pages that are part of a reciprocal hreflang cluster. Examples with variants and country versions are in the article hreflang and canonical.

Which canonical does Google pick?

The URL Inspection tool in Search Console shows two fields, “User-declared canonical” and “Google-selected canonical”. If they do not match, the “Page indexing” report gives the reason “Duplicate, Google chose different canonical than user”. Then check whether the pages really are the same and whether internal links, sitemap and redirects point to the same URL. For all pages at once, you need a crawl that compares the canonical of every URL with its status.

Common mistakes

• The canonical points to a URL with a redirect, 404 or noindex.
• Relative URLs or the wrong protocol, such as http instead of https.
• Two canonicals on one page, for example from the CMS and from a plugin.
• All pages of a template point to the homepage.
• Page 2 of a category points to page 1.
• The sitemap contains different URLs from the canonicals.
• noindex and a canonical to another URL on the same page. Google advises against using noindex for canonicalization.
• robots.txt blocks duplicates. Google then cannot read the canonical.

Frequently asked questions