A canonical tag tells a search engine which version of a page is the real one, when several URLs show substantially the same content.
And the thing almost nobody knows: it is a hint, not a command.
Google chooses the canonical URL itself. Your tag is one signal among several, and Google frequently disagrees with it. If your internal links, your sitemap and your redirects all point somewhere else, Google will follow the weight of evidence and ignore your tag.
Which means a canonical is not a switch. It is a vote, and you need the rest of your site to vote the same way.
What a canonical actually does
Duplicate and near-duplicate URLs are everywhere and most of them are accidental.
example.com/product and example.com/product?ref=email — example.com/shoes and example.com/shoes?colour=blue&size=9 — http:// and https:// and www and non-www — /page and /page/ — a product in three categories, reachable at three URLs.
Without a canonical, the search engine has to guess which one is real. It will pick one. It may not pick the one you wanted, and it may split the signals across several, which is the actual damage.
A canonical says: this one. Consolidate the signals here.
The five ways people break it
One. Every page canonicalises to the homepage.
A genuinely common and completely destructive error, usually introduced by a plugin misconfiguration or a developer who did not understand the tag.
You have just told Google that every page on your site is a duplicate of the homepage. Your entire site can drop out of the index. We have seen it, more than once, and it takes a shocking amount of time for people to find it.
Two. Canonicalising to a URL that redirects.
The canonical points at /page. /page 301s to /page/.
You have sent a conflicting instruction: this is the real page, and also, the real page is somewhere else. Google will resolve it, badly, and unpredictably.
Three. Canonicalising to a noindex page.
"The real version of this page is that one, which you must not index."
Google may apply the noindex to both. That is not a bug. It is a reasonable reading of a contradictory instruction.
Four. Paginated pages canonicalising to page one.
Page 2 of a product listing is not a duplicate of page 1. It contains different products.
Canonicalising page 2 to page 1 tells Google not to index the products on page 2, and on a large ecommerce site this can remove thousands of products from the index and nobody will notice for months. → Ecommerce SEO
Use self-referencing canonicals on paginated pages. Page 2 canonicalises to page 2.
Five. The canonical does not match the internal links.
The tag says /shoes/. Every internal link on the site points to /shoes?category=1.
Google follows the weight of evidence. Your internal links outvote your tag, and this is the most common reason a canonical is silently ignored.
When Google ignores your canonical
And it does, constantly, and it now tells you so.
In Search Console, the URL Inspection tool reports two things: your declared canonical and Google's selected canonical.
When they differ, Google has overruled you.
Why it happens:
Your signals contradict each other. Internal links, sitemap and redirects point elsewhere. The pages are not actually duplicates. You canonicalised two pages together that are genuinely different, and Google can see that. The canonical target is weaker. You pointed a strong page at a weak one, and Google concluded you were mistaken. The target does not exist, 404s, or redirects.
Check this on your important pages. It takes thirty seconds per page and almost nobody does it. → The SEO audit
Canonical, 301 or noindex? The decision that matters
People reach for a canonical when they should be reaching for something else, and the three are not interchangeable.
| Situation | Use |
|---|---|
| The page is gone for good | 301 redirect. Not a canonical. |
| Two URLs, same content, both need to stay live | Canonical |
| The page should exist for users but not be in search | noindex |
| Filter and sort permutations nobody searches | noindex, or block the parameters |
| Same content across two domains you own | Cross-domain canonical |
| Paginated series | Self-referencing canonical on each page |
| Syndicated content on someone else's site | Ask them for a cross-domain canonical to you |
The most common mistake in this table: using a canonical when a 301 is correct. A canonical keeps both pages live and crawlable. A 301 removes one. If the page should be gone, redirect it.
→ 301 redirects and migrations
The rules
Every page has a self-referencing canonical. Including the ones you think do not need it. It is the default and it prevents an enormous number of accidental duplicate problems.
Absolute URLs. Never relative. https://example.com/page/, not /page/.
One canonical per page. Two is a contradiction and Google will ignore both.
Consistent with everything else. Your canonical, your internal links, your sitemap and your redirects must all agree. If they disagree, the canonical loses.
Only in the <head>. A canonical injected into the <body> by JavaScript is ignored.
And it must be the same for the mobile version, because Google indexes mobile-first, and a canonical present on desktop and absent on mobile does not exist. → Mobile SEO
What is a canonical tag?
A tag that tells a search engine which version of a page is the real one when several URLs show substantially the same content, so the signals consolidate on one URL rather than splitting across many.
Is a canonical tag a command?
No, and this is the thing almost nobody knows. It is a hint. Google chooses the canonical itself, and if your internal links, sitemap and redirects all point somewhere else, it will follow the weight of evidence and ignore your tag.
Why is Google ignoring my canonical?
Usually because your other signals contradict it, because the pages are not actually duplicates and Google can see that, or because you pointed a strong page at a weaker one. Search Console's URL Inspection tool shows your declared canonical and Google's selected one side by side.
Should paginated pages canonicalise to page one?
No, and this error can remove thousands of products from the index of a large ecommerce site. Page 2 is not a duplicate of page 1. It contains different products. Use a self-referencing canonical on every page.
When should I use a 301 instead of a canonical?
When the page should be gone. A canonical keeps both URLs live and crawlable. A 301 removes one. Using a canonical when a 301 is correct is the most common mistake in this area.
Can a canonical point to a redirecting URL?
It should not. You are sending a contradictory instruction: this is the real page, and also, the real page is somewhere else. Google will resolve it badly and unpredictably.
Does every page need a canonical?
Yes, a self-referencing one, including the pages you think do not need it. It is the correct default and it prevents a large number of accidental duplicate problems.
Still Have Questions on Your Mind?
Contact our support team and we’ll guide you every step of the way.