18 August 2026
Canonical tags without the confusion
Most sites have too many canonicals, or a chain from A to B to C. One URL per piece of content is enough.
A canonical is one sentence: this is the URL of this document. Google may ignore it if the HTML disagrees. If the pages are not the same, the tag is a lie. The HTML wins. I see shops pointing every filter URL at the category while the content is actually different. That is not hygiene. That is lying.
Unique page: tag to itself, absolute HTTPS. UTM and filters: to the clean URL. Pagination: usually to itself. Pointing everything at page 1 is only fine when the content is truly identical, and it almost never is. Relative canonicals and http canonicals on an https site are extra noise.
www and non-www belong in a 301, not a canonical. I still see A → B → C chains, or a tag aimed at a 404. And the cross-domain “link juice” trick. Rarely works. Often confuses. Two plugins both writing a canonical: then chance wins.
After a migration this is the first thing that goes crooked. One crawl, pull the odd targets, done. Look at the sitemap too. If the sitemap lists URL A and the canonical says B, pick one. Otherwise Google guesses, and not in your favour. Print both lists once. The mismatch is almost never subtle. It is the same twenty URLs, pointed two ways.
Want this on your site today?
First report free. Then auto-scan, a PDF in your inbox and alerts when a score drops.
Start a free report