The same page on your site can often be reached by several addresses. With www and without. With https and without. With a trailing slash and without. With a tracking tag on the end from a Facebook post. To you they are one page. To Google they can look like four, and it may pick the wrong one or spread the signals across all of them.
What the tag does
A canonical tag is one line in the head of the page saying this is the address I want you to treat as the real one. Google explains it in its guide to canonicalisation.
When it is right, everything a page has earned is gathered on one address. When it is wrong, or missing, that value is spread thin, and Google picks a version for you, which may not be the one you would have picked.
A worked example
Here is an illustration of how that plays out. A drainage firm's site was live on both the www and the plain version of its domain, with no redirect and no canonical tags. Both versions were indexed. Links from a local directory pointed at one, links from Facebook at the other.
Neither version ranked as well as one version would have. Setting the canonical on every page and redirecting one address to the other joined the two halves back together. No new content, no new pages, just one address instead of two.
How to check
Open a page on your site, view the page source, and search for canonical. You should see one tag, holding the full address of that page, on the domain you actually use, with https. If the tag points at a different page, or at the homepage on every single page, something is wrong.
Why it matters for your site
Check these on your own site
- One canonical tag per page, pointing at its own full address.
- Pick www or no www and stick to it everywhere, including in the sitemap.
- Redirect the version you are not using, do not leave both live.
- Tracking tags on the end of an address should not create a second indexed page.
- A canonical is a strong hint, not a command. Get the redirects right as well.
Common questions
What happens if a page has no canonical tag?
Google will choose which version of the address to index on its own. Often it picks correctly, but if the same page is reachable several ways the value can split between them.
Can a canonical tag point at another page?
Yes, and that is the point when you genuinely have duplicates. But a canonical pointing at an unrelated page, or at the homepage from every page, can keep your pages out of the results.