The canonical tag explained: setting the right address

When you need it

The same page can be reached at several addresses (with and without www, with parameters).
Shops with filters or sorting options generate many addresses for similar content.
Content is published on several pages, including other people’s sites.

What it looks like

<link rel="canonical" href="https://example.de/hauptseite/"> in the head section of the page.

Rules

Every page points to itself or to the main version.
The address is complete (with https and the domain).
The canonical is a hint, not an instruction: Google can ignore it if it does not fit.
For a genuine move, a 301 redirect is the better choice.

Which signals work together

Google decides for itself which address counts as canonical, and it weighs several hints to do so: the canonical tag, redirects, the entry in the sitemap, internal links, and which variant can be reached via https. If these hints agree, Google usually follows the tag. If they contradict each other, for example when the sitemap lists one address but the canonical points to another, Google may choose differently from what you wanted. So keep all signals consistent.

Example: a shop with filters

A category “Running shoes” can be reached at /laufschuhe/, and with filters also at /laufschuhe/?farbe=blau&groesse=42 and /laufschuhe/?sort=preis. All three show largely the same range. Each of these addresses then carries a canonical to /laufschuhe/ in its head section. The case is different if a filter produces a standalone page that people search for often, such as “Running shoes for women”. That address can then have its own content and a canonical pointing to itself, so that it can rank.

Canonical, redirect or noindex?

Canonical: Both addresses should stay available to visitors, but Google should prefer one.
301 redirect: The old address should no longer be used, for example after a move. More on this under 301 redirects.
Noindex: The page should not appear in search at all, such as internal search results.

A canonical together with noindex on the same page sends contradictory signals and should be avoided.

Other pitfalls

Relative addresses: A canonical without a domain can point to the wrong host if something goes wrong in the system. Complete addresses are safer.
Tag in the body: The canonical only works in the head section. If a plugin places it in the body of the page, it is ignored.
Chains: Page A points to B, B to C. Point directly to the target page.
Tracking parameters: Addresses with campaign parameters should point to the clean address.

Checking

In the URL Inspection tool in Search Console you can see which address Google has chosen as canonical. If it differs from what you specified, it is worth looking for the cause: contradictory signals, content that is too different, or a wrongly placed tag.

A decision aid for your own site

For every duplicate address, ask yourself: is the content really the same? If so, choose the cleanest address as the main version and point all variants to it. If the content is only similar but serves its own search interest, the page keeps its own canonical. If in doubt, take a look in Search Console: if it shows a variant as “Duplicate, Google chose different canonical than user”, it is worth reviewing the signals.

Common mistakes

A canonical pointing to a page that cannot be reached, several different canonicals on one page, or all pages pointing to the home page. Background in Duplicate content.