Search • Oct 31, 2026 • 7 min read

Canonical Tags on Multilingual Websites: How to Avoid Duplicate Content

On a multilingual website, each language version should have a self-referencing canonical tag, and hreflang should connect the versions to each other. Pointing every translation's canonical at the English page is the most common mistake: Google says canonical annotations that point to alternate versions are ignored for canonicalization, so the tag does not do what the author intended.

Exporters add Russian, Malay or Spanish pages to reach new buyers, then discover that Search Console reports duplicates, or that the wrong language version ranks. Most of these problems come from canonical tags, redirects and URLs that disagree with each other. This guide explains what Google's documentation says and gives a checklist you can hand to your developer.

It is a companion to our hreflang guide for multilingual websites: hreflang tells search engines which versions exist, and canonical tells them which URL represents each page.

01What is a canonical URL, and why does it matter for exporters?

Documented

Google describes canonicalization as the process of selecting the representative URL for a piece of content. It groups duplicate pages and shows one version in results. Among the causes of duplicates it lists region variants, HTTP and HTTPS versions, sorting and filtering functions, and accidental variants such as a demo site left open to crawlers. It also states that some duplication is normal and does not violate its spam policies.

For an exporter, the common duplicates are predictable:

http:// and https:// versions of the same page
www and non-www hostnames
product pages reachable through filter or sort parameters
a staging or test copy of the site that was crawled
English pages for different regions, such as /us/ and /my/

Analysis

None of these is a penalty. They cause two practical problems: Google may pick a URL you did not intend, and it may crawl duplicates less efficiently than it crawls your key pages. Canonical signals reduce both risks.

02Are translated pages duplicate content?

Usually not. [Documented] Google's canonicalization page says different language versions are duplicates only when the primary content is in the same language. If only the header, footer and other non-critical text are translated while the body stays the same, the pages count as duplicates.

Analysis

This matters for exporters who "translate" a site by changing the navigation and leaving product descriptions in English. A Russian menu above an English product description is a duplicate of the English page, not a Russian page. A full translation, reviewed by a native speaker, is a separate page that deserves its own canonical.

Same-language regional variants are a different case. [Documented] For versions such as US and UK English, Google recommends using both canonicalization and hreflang so it can decide which regional URL to show.

03Which canonical signals does Google use, and how strong are they?

Documented

Google's guide to specifying a canonical lists these methods, from stronger to weaker:

Method	Strength per Google	When an exporter uses it
Redirect	"A strong signal" that the target should become canonical	Retiring a duplicate URL, moving HTTP to HTTPS
`rel="canonical"` link element or HTTP `Link` header	"A strong signal"	Pages that stay live but duplicate another page
Sitemap inclusion	"A weak signal"	Listing only the URLs you want indexed

Google adds that the methods can be combined to strengthen the preferred URL, and that a canonical is "a hint, not a rule": it may choose a different URL than the one you indicate.

The HTTP header form also works for non-HTML files such as PDFs, which is useful for datasheets that exist in several places. Google advises picking one form, link element or header, since using both is more error-prone.

04How should you set canonical tags on language versions?

The rule has three parts.

Each language version canonicalizes to itself. [Documented] Google says to include a rel="canonical" link on the canonical page itself.
Do not point a translation's canonical at another language. [Documented] Canonical annotations that point to alternate versions, such as those with hreflang, are ignored for canonicalization; rel="alternate" is for those relationships. Google also says that for hreflang you should specify a canonical page in the same language, or the best substitute if none exists.
Connect the versions with hreflang. Every version lists all alternates, including itself.

A correct setup for one product page looks like this:

html

<!-- On https://www.example.com/en/compressor-x/ -->
<link rel="canonical" href="https://www.example.com/en/compressor-x/" />
<link rel="alternate" hreflang="en" href="https://www.example.com/en/compressor-x/" />
<link rel="alternate" hreflang="ru" href="https://www.example.com/ru/compressor-x/" />

<!-- On https://www.example.com/ru/compressor-x/ -->
<link rel="canonical" href="https://www.example.com/ru/compressor-x/" />
<link rel="alternate" hreflang="en" href="https://www.example.com/en/compressor-x/" />
<link rel="alternate" hreflang="ru" href="https://www.example.com/ru/compressor-x/" />

Analysis

The wrong pattern, which we see often, puts href="https://www.example.com/en/compressor-x/" in the canonical tag of the Russian page. The site owner meant "English is the master copy". Google reads it as an annotation that points to a different version, and says such annotations are ignored for canonicalization. The Russian page then relies on other signals instead of the one you thought you were sending.

05What other canonical mistakes should you avoid?

Documented

Google's guidance includes several rules that are easy to break on a template-driven site:

The rel="canonical" link element is only accepted in the <head> of the HTML.
Use absolute URLs rather than relative paths, because relative paths can cause problems if a testing site gets crawled.
Do not specify different canonical URLs for the same page using different techniques.
Do not use robots.txt for canonicalization; disallowed URLs may still be indexed without their content.
Do not use the URL removal tool for canonicalization.
Do not use noindex to choose a canonical within one site; rel="canonical" is the preferred solution.
Link internally to the canonical URL, not to duplicates.
Put the canonical in the HTML source where possible, and make sure JavaScript does not change it.
Google prefers HTTPS by default. Avoid bad certificates and HTTPS-to-HTTP redirects, and do not list HTTP URLs in sitemaps or hreflang.

Analysis

The JavaScript point deserves attention for exporters using page builders or single-page apps. If the canonical in the raw HTML says one thing and a script rewrites it, you are sending conflicting signals. Check the raw HTML as well as the rendered page.

06How do you audit canonicals on a multilingual site?

Run this audit on one page from each template: home, category, product, article, contact.

View the raw HTML source and find the rel="canonical" tag. Confirm there is exactly one, in the <head>, with an absolute URL.
Confirm that the URL matches the language of the page you are on.
Confirm every hreflang alternate returns 200 OK and links back.
Check that the sitemap lists only canonical URLs, with HTTPS.
[Analysis] In Search Console's URL Inspection tool (our working practice, not taken from the two Google pages above), compare the user-declared canonical with the Google-selected canonical. If they differ, look for conflicting redirects, internal links or content that is too similar.
Search for staging or test hostnames that are crawlable and remove or protect them.

Hypothesis

On sites with thin translations, a Google-selected canonical that differs from yours is often a sign the language versions are too similar. We have not found a Google statement that quantifies this, so we treat it as a prompt to review the translation quality, not as a rule.

07Canonical and hreflang checklist for exporters

Each language page has one self-referencing canonical in the <head>
No translation points its canonical at a different language
hreflang lists all versions, including the page itself
Canonical and hreflang URLs are absolute, HTTPS and return 200 OK
Sitemap contains only canonical URLs
Internal links point to canonical URLs
Filter and sort parameters do not create indexable duplicates
Body content, not just the menu, is translated
Staging sites are blocked from public crawling
JavaScript does not rewrite the canonical tag

Our website development service builds multilingual sites with this structure from the start, and if you already have a site, contact us for a canonical and hreflang review. For the content side, see our guide to localizing a website for a new market.

08Frequently asked questions

Should my Russian page canonicalize to the English page?

No. [Documented] Google says canonical annotations that point to alternate versions are ignored for canonicalization, and that you should specify a canonical in the same language. Give each language its own self-referencing canonical and connect the versions with hreflang.

Is a canonical tag a command?

No. [Documented] Google calls it "a hint, not a rule", although it describes rel="canonical" as a strong signal. Google may choose a different canonical than the one you specify.

Can I use robots.txt or noindex instead?

Not for canonicalization. [Documented] Google says not to use robots.txt for this purpose, and that noindex is not recommended for choosing a canonical within a single site.

Do I need canonical tags on PDF datasheets?

Possibly. [Documented] Google says the HTTP Link header form works for non-HTML files such as PDFs. If the same datasheet is available at several URLs, it is worth considering.

Will fixing canonicals improve my rankings?

We cannot promise that. [Analysis] Correct canonicals remove one source of confusion and help Google show the version you intend, but they are one factor among many, and rankings depend on the quality and relevance of the content as well.

Source: What is canonicalization, Google Search Central; How to specify a canonical with rel="canonical" and other methods, Google Search Central.