A canonical tag tells search engines which URL is the master copy of a page. Used correctly it consolidates your ranking signals; used wrong it silently removes pages from search. Here's how it works and the mistakes that deindex pages.
When the same content is reachable at more than one URL, the canonical tag names the one true version, so search engines consolidate signals onto it instead of splitting them.
A canonical tag is a single line of HTML you place in the <head> of a page. It names the URL that should be treated as the master, preferred version of that content. It looks like this:
Why is it needed at all? Because the same page is almost always reachable at several different URLs without you realising. Each of these can return the identical content, and to a search engine they look like separate, duplicate pages competing with each other:
Left unmanaged, Google has to guess which of these is the "real" one, and your ranking signals (links, content quality, click data) get split across the duplicates instead of pooling on one strong page. The canonical tag removes the guesswork: every variant declares the same master URL, and the signals consolidate onto it. For the vast majority of indexable pages, the correct master is the page itself, a self-referencing canonical.
Almost every indexable page should point its canonical at its own URL. This is the single most important canonical habit.
A self-referencing canonical explicitly confirms "this URL is the preferred version," which protects the page from a parameter or duplicate URL being chosen instead. So on the page at https://example.com/blue-widgets, the canonical in its head should be its own clean URL:
Always use the absolute URL (full https:// and domain), not a relative path, and match your site's exact trailing-slash and www convention. A relative or inconsistent canonical is one of the easiest ways to send a mixed signal.
Each of these is a real check in the audit, because each one quietly removes pages from search.
If a page's canonical points to a URL that 301-redirects somewhere else, you're sending search engines on a detour. Google often ignores a canonical that resolves through a redirect and picks its own, which may not be the URL you wanted. Point the canonical at the final destination directly.
A canonical that points to a 404 or a server error is naming a dead page as the master. Google can't honour it, so it disregards the signal and guesses. The page's ranking signals scatter. The audit flags canonicals pointing to 4XX and 5XX targets specifically.
A frequent and costly bug: paginated archive pages (page 2, 3, 4...) all set their canonical to page 1. This tells Google that pages 2+ are duplicates of page 1, so it deindexes them: along with the links and products only reachable from those pages. Paginated pages should self-reference their canonical.
A canonical pointing from HTTPS to HTTP (or vice versa), or off to a different domain, sends authority and indexing intent to the wrong place. The audit flags canonical HTTP-to-HTTPS, HTTPS-to-HTTP, and off-domain canonicals. After any HTTPS or domain migration, re-check that every canonical references the new, correct URL.
Every paginated page canonicalising to page 1:
Paginated pages point to themselves:
User-declared canonical is the URL you set, but Google-selected canonical differs, or the page lands in the "Alternate page with proper canonical tag" or "Duplicate, Google chose different canonical than user" report under Pages. If a page you want indexed shows either of those, its ranking signals are being sent elsewhere. That is the symptom this whole section is about.A short set of rules that prevents almost every canonical problem.
Self-reference by default. Every indexable page should canonicalise to itself unless it's genuinely a duplicate of another URL. Point at a live, canonical 200 URL. Never canonicalise to a redirect, an error page, or a non-canonical URL. Keep it absolute and consistent. Use the full URL with the correct protocol, domain and trailing-slash convention. Align canonical, hreflang and internal links. They should all reference the same URL for a page, if your hreflang points one way and your canonical another, you've sent a contradiction. Re-check after migrations, when canonicals most often break.
It's a strong hint, not a hard directive. Google usually honours a clear, consistent canonical, but if your signals conflict: canonical says one thing, internal links and sitemap say another, it may choose a different URL. Consistency across canonical, internal links, sitemap and hreflang is what makes the canonical reliable.
A self-referencing canonical on every indexable page is best practice. It's not strictly mandatory, but it removes ambiguity and protects against parameter and duplicate URLs being chosen instead. The audit flags non-canonical URLs that receive organic traffic, which often signals a missing or wrong canonical.
Use a canonical when two URLs are duplicates and you want to consolidate them onto one. Use noindex when you want a page kept out of search entirely. They solve different problems: canonical merges, noindex excludes. Don't combine a noindex with a canonical to another URL, the signals conflict.
If users never need to see the duplicate URL (an old path, a non-www version), use a 301 redirect, it's stronger and sends both users and crawlers to the right place. Use a canonical when the duplicate URL must stay accessible to users (a print version, a page reachable via filters or tracking parameters) but should not compete in search. Rule of thumb: redirect if the duplicate shouldn't exist; canonical if it must exist but shouldn't rank separately.
Give the page a self-referencing canonical pointing to the clean, parameter-free URL. Then every ?utm_source=... or ?ref=... variant declares the same canonical, so the analytics parameters don't spawn duplicate pages in the index. This is the standard fix for parameter-driven duplication.
Free to start. Find canonicals pointing to redirects, errors and the wrong URLs across your site.
Start my free audit