On this page8 sections
Why it matters for founders and small teams
Every page you publish can be reached at more addresses than you think — with tracking parameters, with and without www, through a syndication partner — and each copy splits the links and attention one page should get. A correct canonical tag is a one-line fix that keeps your best page credited as the source, in Google’s results and in the AI answers built on its index.
How does a canonical tag work?#
A canonical tag works by placing <link rel="canonical" href="…"> in a page’s <head> to name the URL you want indexed; when Google clusters duplicate or near-duplicate pages, it weighs that hint alongside redirects, sitemaps and other signals to choose the one version it crawls most, ranks and shows.
<!-- On every version of the page, including the canonical itself --><head> <title>Plannora pricing: plans for teams of every size</title> <link rel="canonical" href="https://www.plannora.io/pricing" /></head> <!-- For a PDF or other non-HTML file, send an HTTP header instead -->Link: <https://www.plannora.io/guides/launch-plan.pdf>; rel="canonical"Google ranks its canonicalization signals by strength: redirects (strong), rel="canonical" annotations (strong) and sitemap inclusion (weak), and they stack when they agree. The chosen canonical “will be crawled most regularly; duplicates are crawled less frequently,” it becomes “the main source to evaluate content and quality,” and links pointing at the duplicates are consolidated onto it.
- Use absolute URLs. Google supports relative paths but recommends against them.
- Put it in the
<head>. Google only accepts the element there, so keep the<head>valid HTML. - Add a self-referencing canonical to the canonical page itself, as Google recommends.
- One method, one answer. Never declare a different URL in the HTTP header, the HTML and the XML sitemap.
Why does Google ignore my canonical tag?#
Google ignores a canonical tag when other signals disagree with it — the pages aren’t really duplicates, internal links point elsewhere, or the target redirects — because a canonical is a strong hint, not a command, and Google picks whichever version it judges most useful to searchers.
Search Console shows the verdict. URL Inspection lists the user-declared and Google-selected canonical side by side, and the Page indexing report flags “Duplicate, Google chose different canonical than user” when “Google thinks another URL makes a better canonical.” That’s different from “Alternate page with proper canonical tag,” which means everything worked and needs no action.
| Likely cause | Fix |
|---|---|
| The pages aren’t truly duplicates | Make them clearly distinct, or merge them — Google clusters on the main content |
| Internal links point at another version | Link to the canonical URL everywhere, navigation included — see internal linking |
| Header, HTML and sitemap disagree | Declare the same URL in every method |
The target redirects, 404s or is noindex | Point the tag at a live, indexable page |
| HTTP and HTTPS both resolve | Redirect to HTTPS; Google prefers it unless signals conflict |
| JavaScript rewrites the tag | Set it in the server HTML and leave it unchanged |
Worked example
The URL Variant Multiplier
A way to count how many addresses one page really has before deciding where canonicals and redirects are needed. The inputs are illustrative, for the pricing page of a fictional app called Plannora — count your own variants the same way, from your server logs.
- 1
Protocol and host
httpandhttps, with and withoutwww— four combinations when the server answers on all of them.4
- 2
Trailing slash
/pricingand/pricing/both return 200 instead of one redirecting to the other.× 2 = 8
- 3
Tracking parameters
The clean URL plus three parameters that show up in the logs from newsletters and ads:
?utm_source=,?ref=and?gclid=.× 4 = 32
- 4
Syndicated copy
A partner republished Plannora’s pricing explainer on its own domain.
+ 1 = 33
- =
Result
33 crawlable URLs for one page, each able to collect links and split signals. Redirecting
httpand the non-wwwhost to one version, enforcing one slash rule and adding a self-referencing canonical collapse the 32 on-site variants into one; the partner’s copy needsnoindexor a canonical back.33 → 1
The result: Run the same count across a site and it doubles as a crawl budget estimate: every variant a crawler can reach is a fetch not spent on a new page. Fix protocol and host with redirects first, since they’re the strongest signal, then let canonicals handle the parameters you can’t avoid.
Free to use and adapt. If you cite it, link to rankbox.xyz/glossary/canonical-tag.
Canonical tag vs noindex vs redirect: which should you use?#
Use a redirect when the duplicate URL should stop existing for people too, a canonical tag when both versions must stay reachable, and noindex only when a page shouldn’t appear in search at all — Google advises against noindex for choosing a canonical within a site, because it removes the page entirely.
| Situation | Use | Why |
|---|---|---|
| An old URL replaced by a new one | 301 redirect | The strongest signal; people and bots both land on the new page |
| Tracking parameters, sort orders, print views | rel="canonical" | Every URL keeps working while signals consolidate on one |
| Your article republished by a partner | The partner adds noindex, or a canonical to you | Google now prefers the partner blocking indexing for syndication |
| Cart, account and internal search pages | noindex, or keep them out of crawling | They shouldn’t be in search at all |
| Duplicates you’d like to hide | Not robots.txt | A blocked URL can still be indexed without its content, and the crawler never sees your canonical |
Syndication is the case that matters most for AI search. Google’s troubleshooting guide says the canonical element “is not recommended for those who want to avoid duplication by syndication partners,” and that the most effective fix is for partners to block indexing of your content. Write one or the other into every syndication deal: Columbia’s Tow Center caught Perplexity citing republished copies of articles instead of the originals.
Do AI search engines respect canonical tags?#
Google’s AI features inherit its canonicalization, because AI Overviews, AI Mode and Gemini ground answers in a Search index where duplicates are already folded into one canonical. OpenAI and Perplexity haven’t documented whether they honor rel=canonical, and Brave’s discovery client, which feeds Claude’s search provider, reads it only from the HTML head.
- Google — AI Overviews, AI Mode, GeminiHelps
- Grounding retrieves from the Search index, so the Google-selected canonical is the URL that can be cited. Google’s AI optimization guide lists reducing duplicate content among its technical best practices.
- Brave — Claude's search providerHelps
- Brave’s discovery client drops URLs with more than one query parameter unless a canonical points to a clean version, and it reads canonicals from the
<head>, not HTTP headers. - OpenAI — ChatGPT searchUnconfirmed
- OpenAI doesn’t say whether
OAI-SearchBothonorsrel=canonical. Keep canonicals right for Bing, a named ChatGPT search provider. - PerplexityUnconfirmed
- Not documented, and it has been caught citing syndicated copies over originals — so control duplicates at the source.
One rule covers every engine that doesn’t run JavaScript: a canonical injected by a script doesn’t exist for it, exactly like any other head tag. Put the tag in the server HTML — see server-side rendering — so the same answer reaches Google, Bing, Brave and the AI crawlers alike.
Common mistakes with canonical tags#
The most common canonical tag mistakes are pointing every paginated page at page one, canonicalizing to URLs that redirect or carry noindex, letting a theme and a plugin each emit a conflicting tag, and using canonicals to paper over different pages that should be merged.
Myth
A canonical tag is a command.
Reality
It’s a hint. Google says it “may choose a different page as canonical than you do” — check URL Inspection rather than assuming.
Myth
Every paginated page should point to page 1.
Reality
Google says not to use the first page of a sequence as the canonical. Page 2 isn’t a duplicate of page 1; give each page its own self-referencing canonical.
Myth
A canonical fixes two articles competing for one keyword.
Reality
Different articles aren’t duplicates, so Google may ignore the hint. Merge them and redirect — see keyword cannibalization.
Myth
Two canonical tags are fine as long as one is right.
Reality
Google warns that multiple or conflicting canonicals “may lead to unexpected results.” Themes and SEO plugins often both add one — view source and count.
Related terms#
- Content & relevanceKeyword cannibalizationWhen two or more pages on the same site target the same query and intent, splitting links and relevance so search engines rotate between them or rank neither as well as one consolidated page would.Read the entry
- Technical SEOIndexingThe step in which a search engine processes a crawled page and stores it in its searchable database; only indexed pages can rank, or be retrieved for AI answers built on that index, and being crawled does not guarantee being indexed.Read the entry
- Technical SEOCrawl budgetThe number of URLs a search engine is willing and able to crawl on a site in a given period, set by how much load the server can take and how much the engine wants the content — a real constraint mainly for large or fast-changing sites.Read the entry
- Technical SEOXML sitemapA machine-readable file listing a site’s canonical URLs, optionally with last-modified dates, so search engines can discover pages and prioritize recrawling the ones that changed — a discovery aid, not a ranking factor or a guarantee of indexing.Read the entry
- Authority & off-siteInternal linkingThe practice of linking between pages on the same website, which helps crawlers find pages, moves authority from strong pages to weaker ones and shows search engines and AI systems how the topics on a site relate.Read the entry
- SSRTechnical SEOServer-side renderingThe practice of generating a page’s full HTML on the server before sending it, so crawlers that don’t run JavaScript — which includes most AI crawlers — can read the content in the first response.Read the entry
Go deeper
Sources
- 1.How to specify a canonical URL with rel="canonical" and other methodsGoogle Search Central · developers.google.com ↗
- 2.What is canonicalizationGoogle Search Central · developers.google.com ↗
- 3.Fix canonicalization issuesGoogle Search Central · developers.google.com ↗
- 4.Page indexing reportSearch Console Help · support.google.com ↗
- 5.How Brave Search discovers new pagesMERJ · merj.com ↗
- 6.We compared eight AI search engines. They're all bad at citing newsColumbia Journalism Review · cjr.org ↗
