Canonical URL
rel=canonical / canonical tag / duplicate content / preferred URL
A canonical URL is the version of a page you want treated as the real one when several URLs show the same content. The standard calls it the author's preferred version. Google ranks it a strong signal, not an instruction, and says none of these methods are required because it will pick a version anyway.
The standard is unusually candid about what this is. RFC 6596 defines the canonical link relation as designating an address as preferred over resources with duplicative content, and then describes it informally as the author's preferred version of a resource. Preferred. The word is in the specification, and everything else follows from it.
Google treats it that way too, and is explicit about the pecking order. It lists the ways to indicate a canonical in order of how strongly they influence the outcome: redirects are a strong signal that the redirect target should become canonical, a rel=canonical annotation is a strong signal that the specified URL should become canonical, and inclusion in a sitemap is a weak signal. It adds that these methods stack, so using more than one increases the chance your preferred URL is the one that appears.
Then the sentence that changes how you read every canonical audit. Google says that while it encourages these methods, none of them are required, and your site will likely do fine without specifying a preference at all, because if you do not specify one, Google will identify which version it considers the best to show users. You are not instructing the index. You are voting in it.
There is one clear wrong tool, and it connects to a mistake people make constantly. Google says not to use robots.txt for canonicalization, because it may still index URLs disallowed there, without their content. Blocking a duplicate does not consolidate it. It can leave the duplicate in the index as a bare address.
In practice
The common failure is not a missing canonical but contradictory ones. Google specifically warns against naming different URLs through different techniques, such as one URL in your sitemap and a different URL in the rel=canonical annotation for the same page. Each signal is doing its job. They are voting for different candidates, and the result is decided by something other than your intent.
Not the same as
- A redirect
- That moves people and crawlers to another address. A canonical leaves both pages reachable and expresses a preference about which one represents them.
- noindex
- That asks for a page to be left out entirely. A canonical asks for credit to be gathered on one of several live versions.
Why it matters to you
Canonical work appears on a lot of invoices, and it is real work, but it is worth knowing what was bought. Correctly declared canonicals make your preference clear and consistent. They do not guarantee the outcome, they are not required, and the most common reason they fail is not absence but disagreement between the signals a site is already sending.
What to ask or check
- 01Which single URL do we actually want indexed for this page?
- 02Do our redirects, sitemap and canonical annotations all name that same URL?
- 03Is anything relying on robots.txt to solve a duplication problem?
What people get wrong
That a canonical tag tells Google which URL to index. RFC 6596 calls it the author's preferred version, and Google lists it as one signal among several, saying none of them are required.
Red flags
- A sitemap and a rel=canonical annotation naming different URLs for the same page.
- Duplicate URLs blocked in robots.txt as a way of consolidating them.
- Canonical work sold with a promised indexing outcome.
Who owns it
Whoever controls the page templates and the sitemap. Those are frequently different people, which is how the signals end up disagreeing.
Where you will see it
In the head of a page as a rel=canonical link, and in the indexing reports that tell you Google chose a different URL.