What a Canonical URL Tag Does — and What It Does Not

The one-sentence version

A canonical tag tells a search engine: of these URLs that show the same content, index this one. It is a hint, not a command, and treating it as a command is where most canonical-tag problems come from.

Why duplicate URLs exist at all

Before the tag makes sense, it helps to be clear about the problem it solves. A single page routinely has several legitimate addresses:

• With and without a trailing slash — /about and /about/
• With http:// and https://
• With www. and without
• Reachable through a ?utm_source= campaign link or a ?ref= internal link
• Reachable through a filtered view — ?page=2, ?sort=asc, ?category=guides

None of these are different content. A search engine that indexed all of them would spend crawl budget on duplicates and could split link equity across them, so consolidation matters.

The syntax

<link rel="canonical" href="https://example.com/article/real-url/" />

Three requirements, and violating any of them makes the tag unreliable:

1. Absolute URL. Include the scheme and host. A relative href is tolerated in some cases and ignored in others.
2. In the <head>. Not in the body.
3. One per page. If a page emits three canonicals pointing at different URLs, the behaviour is undefined. Pick one.

The WHATWG specification treats rel="canonical" as a link type with defined semantics rather than proprietary syntax, which is worth knowing: it is standard HTML, not a Google-specific hack.

What it does

Google describes the tag as a hint, and the phrasing is deliberate. When honoured, a canonical causes the engine to consolidate signals — links, mentions, crawl history — onto the declared URL and to pick that one for its index.

In practice, “canonical” is close to “the URL I would have chosen anyway, if you had made it obvious.”

What it does not do

This is the part worth memorising, because each of these is a common and expensive misunderstanding.

It does not redirect users. The tag is invisible to browsers. A visitor following a link to /page?variant=2 is served /page?variant=2 regardless of what the tag says. If you need visitors moved, use a redirect.

It does not override a server redirect. If /old issues a 301 to /new, the redirect determines what is served. Adding a canonical on top of that is a configuration conflict, not a refinement.

It does not create pages. A canonical pointing to a URL that does not exist, or that returns 404, is ignored — and you have lost the signal you were trying to send.

It does not stop crawling. The tag influences indexing, not discovery. Google will still crawl a non-canonical URL it finds linked, in order to see its canonical.

It does not fix content. If two pages are genuinely different, tagging one as canonical of the other is a signal to consolidate signals onto a page that does not contain the material a searcher wanted. The right fix is usually to merge the content.

Canonical or redirect?

The two mechanisms answer different questions, and using the right one is most of the battle.

Situation	Use
The old URL should send visitors to the new one	Redirect (301)
Several URLs show identical content, any may be linked	Canonical tag
A filtered view should not be indexed but should stay crawlable	Canonical tag pointing at the base view
A campaign URL should count for the landing page	Canonical tag; the utm_ params are stripped in reporting
Two pages are near-identical but not identical	Canonical tag, plus noindex on the weaker one if needed
A page moved and the old URL is gone entirely	301 redirect, plus sitemap update

A useful rule: canonical is for telling search engines which of several live URLs is authoritative. Redirect is for telling everyone, including users, that a URL has moved.

These compose. If /old redirects to /new and /new self-references, both mechanisms agree and nothing needs reconciling.

Diagnosing a tag that is being ignored

When a canonical is not honoured, check these in order. Most failures are configuration mistakes rather than anything to do with content:

1. Does the target URL return 200? A canonical pointing to a redirect or an error is dropped.
2. Is the target itself canonicalised elsewhere? Canonical chains and loops occasionally resolve in unhelpful ways.
3. Is the tag absolute, in the head, and singular? Check all three.
4. Are the pages actually duplicates? Google will sometimes decline a hint it disagrees with, particularly when the pages differ substantially.
5. Is another canonical signal winning? For URLs with many inbound links, engines sometimes prefer the most-linked version regardless of the tag.
6. Is the page noindex? If a page carries both noindex and a canonical, the noindex wins and the tag is moot. This is a common and confusing combination.

Self-referencing canonicals

Put a self-referencing canonical on every indexable page:

<link rel="canonical" href="https://example.com/this-exact-page/" />

This is unremarkable and widely recommended. It does nothing on its own, which is precisely the point — it removes ambiguity, so that when a non-canonical URL is linked, the signal is unambiguous about where it should consolidate.

It also means every URL a site publishes is explicit about its own identity, which is a useful property as a site grows into hundreds of pages and query parameters start appearing in internal links.

Anti-patterns

A short list of things that look helpful and are not:

• Canonicalising every page to the homepage. Collapses the entire site into one URL.
• Changing a canonical in response to a query string parameter. If ?page=2 and ?page=3 point at different content, they are different pages and need different URLs.
• Canonicalising to a page that redirects. Silently does nothing.
• Relying on a canonical to consolidate pages you have not merged. Searchers arriving on the consolidated page still will not find the material they were looking for.
• Adding a canonical to a noindex page and expecting it to index. Contradictory directives; the noindex is honoured.

The underlying principle

Canonical tags are a way of expressing an editorial decision about which URL represents a piece of content. The tag only helps when that decision has already been made honestly and implemented consistently across the templates, the sitemap and the internal links.

A site whose canonical tags match its real information architecture has done the hard part already. A site using the tag to paper over an inconsistent structure will find the search engine quietly declining the suggestion.