A canonical tag is a single line in a page’s <head>, <link rel="canonical" href="https://example.com/page/">, that tells search engines which URL you consider the main version when the same or very similar content is reachable at more than one address. Search engines use it to pick one URL to index and show, and to merge signals such as links into that URL. Google treats it as a strong hint, not a rule: it can still pick a different canonical if your other signals disagree (Google Search Central).
Most canonical problems are not missing tags. They are wrong ones: a tag that points every paginated page back to page one, a plugin and a theme that each print a different canonical, or JavaScript that rewrites the tag after load. Each of those can quietly drop real pages out of the index. The same mess now costs you in AI answers too. Microsoft’s Bing team says LLMs group near-duplicate URLs into one cluster and pick a single page to represent it, which may not be the one you meant to highlight (Bing Webmaster Blog). Last updated: November 9, 2026.
What Is a Canonical Tag?
The canonical tag, often called rel canonical, is a link element with rel="canonical". It was formalized in RFC 6596 in 2012, which describes it as a way to designate the preferred URL “from a set of resources that return the context IRI’s content in duplicated form” (RFC 6596). In practice it looks like this:
<head>
<title>Green dresses</title>
<link rel="canonical" href="https://example.com/dresses/green-dresses/" />
</head>
You place that same tag on every duplicate (for example ?color=green&sort=price or a ?gclid= ad URL) and on the main page itself. The version on the main page is called a self-referencing canonical, and Google recommends it (Google Search Central).
Google lists four practical reasons to declare a canonical: control which URL people see in results, consolidate signals such as links from duplicates into one URL, simplify reporting for a piece of content, and stop crawlers spending time on duplicates instead of new or updated pages (Google Search Central).
How Canonical Tags Work in Google
When Google indexes a page, it works out the page’s primary content. If several URLs look the same or very close, it clusters them and picks the one it considers most complete and useful as the canonical. The canonical is crawled most often. Duplicates are crawled less frequently (Google Search Central).
Your canonical tag is one input to that choice. Google names the others: HTTPS versus HTTP, redirects, sitemap inclusion and hreflang clusters. It ranks the explicit methods by strength (Google Search Central):
- Redirects: a strong signal that the redirect target should be canonical.
- rel=”canonical” link annotations: a strong signal.
- Sitemap inclusion: a weak signal.
These signals stack. When your redirects, canonical tags, sitemap and internal links all point to the same URL, Google is far more likely to agree with you. When they disagree, Google picks for itself.
Google also says you don’t strictly need to declare a canonical. If you don’t, it will choose the version it thinks is best (Google Search Central). That is fine for a five-page site. It is risky for a store with filters, tracking parameters and multiple category paths to the same product.
Canonical Tag vs 301 Redirect vs Noindex vs Sitemap
Each tool solves a different problem. Picking the wrong one is the root of many of the mistakes later in this guide.
| Method | Use it when | What happens to the duplicate | Signal strength |
|---|---|---|---|
| 301 redirect | The duplicate should not exist for users either (old URL, http, non-www) | Users and bots are sent to the target | Strong |
| rel=”canonical” tag | The duplicate must stay reachable (filters, sort orders, tracking parameters, print views) | Page still loads; search engines are asked to index the target instead | Strong hint |
| rel=”canonical” HTTP header | The duplicate is a PDF or other non-HTML file | Same as the tag, sent in the response header | Strong hint |
| Sitemap inclusion | You want to nominate canonicals across a large site | Nothing directly; it only suggests which URLs are preferred | Weak |
| noindex | The page should not appear in search at all | Page is dropped from results entirely | Not a canonical signal |
| robots.txt disallow | You want to save crawl on paths with no search value | URL can still be indexed without its content | Not a canonical signal |
Google is explicit on the last two rows: don’t use robots.txt for canonicalization, and don’t use noindex to choose a canonical within a site, because it blocks the page from Search completely (Google Search Central). Bing’s guidance lines up: use 301s when content moves, and keep 302s for very short changes (Bing Webmaster Guidelines).
Why Duplicate URLs Also Confuse AI Crawlers
Canonical tags were built for search engines, but AI answers are now downstream of the same index decisions. Three public sources make the case.
Bing says duplicates weaken grounding. Its webmaster guidelines state that duplicate URLs “dilute signals and reduce Bing’s confidence in selecting a URL for grounding results or citations,” and recommend canonical URLs, parameter controls and consistent URL structures to fix that (Bing Webmaster Guidelines). In a December 2025 post, Bing’s Fabrice Canel and Krishna Madhavan added that many LLMs rely on data grounded in the Bing index or other search indexes, that AI systems often cluster similar pages and choose one to represent the set, and that duplicates can delay updates reaching AI-generated results (Bing Webmaster Blog).
Several AI crawlers don’t run JavaScript. Vercel and MERJ analyzed crawler traffic and found that OpenAI’s crawlers (OAI-SearchBot, ChatGPT-User, GPTBot), ClaudeBot, PerplexityBot and others did not render JavaScript, while Gemini uses Googlebot’s infrastructure and does (Vercel). If your canonical tag only exists after a script runs, those crawlers never see it. They see whatever is in the raw HTML. For more on who these bots are, see our AI crawler list and OpenAI crawlers guide.
AI crawlers waste fetches on bad URLs. The same Vercel study found ChatGPT spent 34.82% of its fetches on 404 pages and another 14.36% following redirects, compared with 8.22% and 1.49% for Googlebot (Vercel). Every duplicate or redirecting URL you expose is another place for a less efficient crawler to spend its visit instead of reading the page you want cited.
No AI company has published how its own crawler treats rel="canonical". So the safe working assumption is simple: make the preferred URL obvious in the raw HTML, in your redirects, in your sitemap and in your internal links, so every system lands on the same answer whether it reads tags or not.
Where Duplicate URLs Come From
Google and Bing list the same usual suspects (Google Search Central; Bing Webmaster Blog):
- Protocol and host variants:
httpvshttps,wwwvs non-www. - Formatting variants: trailing slash vs none, uppercase vs lowercase paths.
- Parameters: sort, filter, session IDs and tracking tags like
utm_orgclid. - Site functions: sorted and filtered category pages, printer-friendly versions.
- Device and region variants: separate mobile URLs, US and UK pages with the same English copy.
- Accidental variants: a staging or demo site left open to crawlers.
- Syndication and campaigns: the same article on partner sites, or several near-identical landing pages for one offer.
Some duplication is normal and is not a spam issue (Google Search Central). The goal is not zero duplicates. It is one clear answer about which URL counts.
How to Add a Canonical Tag
In HTML
Put one link rel="canonical" element inside <head>, using an absolute URL. Google supports relative URLs but recommends absolute ones, because relative paths can cause problems later, such as a crawlable staging site canonicalizing to itself. The tag is only accepted in the <head>, so the head must be valid HTML (Google Search Central).
In the HTTP header
For PDFs and other non-HTML files, send a Link header instead:
Link: <https://www.example.com/downloads/white-paper.pdf>; rel="canonical"
Google supports both the element and the header but recommends choosing one, since using both makes it easier to send two different URLs by mistake (Google Search Central).
In a CMS
- WordPress: core prints a canonical link for single posts and pages through
rel_canonical()(WordPress Developer Resources). SEO plugins such as Yoast or Rank Math usually take over that output and let you override it per page. - Shopify: themes expose a
canonical_urlLiquid object, normally printed intheme.liquid(Shopify Liquid docs). Check it once after every theme change. - JavaScript frameworks: render the tag on the server. Google’s advice is to set the canonical in the HTML source and make sure JavaScript doesn’t change it (Google JavaScript SEO basics).
Canonical Tag Mistakes That Hide Pages
These are the errors that take pages out of search, not just the ones that look untidy in an audit.
1. Pointing paginated pages to page one
Page 2 of a category or blog archive holds different products or posts than page 1. If page 2 canonicalizes to page 1, you are telling search engines page 2’s content is a duplicate, and the items only linked from it can lose their crawl path. Google’s pagination guidance says not to use the first page as the canonical and to give each page its own canonical URL (Google Search Central). RFC 6596 says the same (RFC 6596).
2. Canonicalizing pages that are not duplicates
A canonical should point to content that is the same as, or a superset of, the current page (RFC 6596). Pointing every category to the homepage, or every product variant with a unique spec sheet to one parent, asks search engines to drop pages that answer different queries. Google may ignore the hint, or it may follow it and you lose the pages.
3. Pointing to a redirect, a 404 or a noindexed URL
RFC 6596 advises against canonical targets that permanently redirect, that return an error, or that themselves canonicalize somewhere else. When that happens, search engines can fall back on their own heuristics and ignore your tag (RFC 6596). Pointing at a noindexed page sends two opposite instructions at once.
4. Sending conflicting signals
Google warns against declaring one URL in your sitemap and a different one in rel="canonical" for the same page, and asks you to link internally to the canonical rather than to duplicates (Google Search Central). If your navigation links to /shoes?sort=new while the canonical says /shoes/, you are voting against yourself. Our XML sitemap guide covers keeping the sitemap to canonical, 200-status URLs only.
5. Printing more than one canonical
This usually happens when a theme and an SEO plugin both output the tag, or when an HTML element and an HTTP header disagree. RFC 6596 says to specify only one canonical per resource (RFC 6596). View the source and search for rel="canonical"; you should find exactly one.
6. Letting JavaScript change the canonical
If the raw HTML says one URL and a script swaps in another, Google may see two answers, and non-rendering AI crawlers only ever see the first. Google says to set the canonical in the HTML and not let JavaScript change it. If you cannot set it in HTML, leave it out of the source and set it only with JavaScript (Google Search Central).
7. Leaking staging or http URLs into the tag
Hard-coded or relative canonicals copied from a staging environment can point live pages at a test domain. Google prefers HTTPS canonicals, but an HTTPS page that canonicalizes to its HTTP version (or redirects through HTTP) can push Google toward the HTTP URL (Google Search Central).
8. Putting the tag outside the head
A canonical inside <body>, or after an element that breaks the head early (some injected scripts and iframes do this), is not accepted. Google only reads rel="canonical" in a valid <head> (Google Search Central).
9. Mixing hreflang and canonicals across languages
On multilingual sites, each language page should canonicalize to itself (or the best page in the same language), not to the English original. Google ignores canonical annotations that carry hreflang, lang, media or type attributes, and prefers URLs that sit in reciprocal hreflang clusters (Google Search Central).
10. Relying on canonicals for syndicated copies
Bing suggests asking partners to add a canonical pointing to your original (Bing Webmaster Blog). Google is more cautious: it does not recommend relying on the canonical element for syndication because partner pages are often very different, and says the most effective fix is for partners to block indexing of the copy (Google Search Central). If you syndicate, ask for both a canonical and a noindex, or publish excerpts with a link back.
11. Missing a hacked canonical
Google notes that some attacks insert a cross-domain canonical or redirect into your HTML or headers, pointing at a spam URL. Its systems may then pick the attacker’s page as canonical (Google Search Central). A sudden drop in indexed pages with no content change is a reason to check.
How to Check Your Canonical Tags
- Read the raw HTML. Use View Source, not the browser inspector, which shows the page after JavaScript. Or fetch it from a terminal:
curl -s https://example.com/page/ | grep -i canonical. This is roughly what a non-rendering AI crawler receives. - Check headers for files.
curl -sI https://example.com/file.pdfshows anyLink: ...; rel="canonical"header. - Compare user-declared and Google-selected canonicals. Google Search Console’s URL Inspection tool shows both. If they differ, decide whether Google’s choice actually makes more sense for searchers (Google Search Central).
- Read the Page indexing report. Three statuses matter most (Search Console Help):
| Status | What it means | What to do |
|---|---|---|
| Alternate page with proper canonical tag | The page points correctly at an indexed canonical | Nothing, if that was the intent |
| Duplicate without user-selected canonical | Google sees a duplicate and picked another URL itself | Add a canonical, or make the page clearly different |
| Duplicate, Google chose different canonical than user | Google disagrees with your tag | Check for conflicting signals or near-identical content |
- Check Bing too. Bing Webmaster Tools surfaces duplication issues such as many pages with identical titles in its Recommendations tab (Bing Webmaster Blog). Our Bing Webmaster Tools guide walks through setup.
After a fix, be patient. Google says it can hold pages in a duplicate cluster for up to two weeks, and pages split out faster when the content difference is clear and significant. You can use Request Indexing for your most important URLs, within its quota (Google Search Central). On Bing and other participating engines, IndexNow can announce the change faster (Bing Webmaster Blog).
A Short Canonical Checklist
- Every indexable page has exactly one self-referencing canonical, absolute URL, in the server-rendered
<head>. - Duplicates that must stay live (filters, sort orders, tracking URLs) canonicalize to the clean URL.
- Duplicates that don’t need to exist are 301-redirected instead.
- Paginated pages each canonicalize to themselves.
- The canonical target returns 200, is indexable and does not canonicalize elsewhere.
- Sitemap, internal links,
hreflangand canonicals all name the same URL. - No JavaScript changes the canonical after load.
- Staging sites are blocked from crawling and never appear in canonical tags.
Pair this with a clean robots.txt so crawlers, human-built or AI, spend their visits on the pages you want found.
Canonical Tag FAQ
Is a canonical tag a directive or a hint?
A hint. Google calls a canonical preference “a hint, not a rule” and may choose another URL if other signals disagree. Redirects, consistent internal links and sitemaps make your hint much more likely to be followed.
Should every page have a self-referencing canonical?
Yes, for indexable pages. Google recommends adding rel=”canonical” to the canonical page itself. It protects the page from parameter and tracking variants that other sites or ads create.
What is the difference between rel canonical and a 301 redirect?
A 301 sends users and bots to another URL, so the old one effectively disappears. A canonical leaves the duplicate reachable and only asks search engines to index the preferred URL. Use 301s for retired URLs and canonicals for variants that must stay live.
Does a canonical tag pass link signals?
Google says canonicalization helps it consolidate signals, such as links, from duplicates into the preferred URL. That only works if Google accepts your canonical.
Can a canonical tag point to another domain?
Yes. RFC 6596 allows a canonical target on a different hostname or domain. For syndicated articles, though, Google prefers that partners block indexing of the copy rather than rely on a cross-domain canonical.
Do AI crawlers respect canonical tags?
No AI company has published how its crawler handles rel=”canonical”. Bing says duplicate URLs reduce its confidence in picking a URL for grounding and citations, and several AI crawlers do not run JavaScript. Keep one server-rendered canonical and consistent signals so every system lands on the same URL.
How long does it take Google to update a canonical?
Google says it may hold pages in a duplicate cluster for up to two weeks after you fix content issues. Use URL Inspection’s Request Indexing for your most important URLs once the fix is live.
Can AI engines read your site?
Check any page for free: crawler access for ChatGPT, Perplexity, Claude and Google AI Overviews, plus the fixes that matter most.


