Canonical Tags: How Google Picks One URL Among Duplicates, and How to Make It Pick Yours

Why one page ends up with many addresses

Google defines canonicalization as "the process of selecting the representative" URL of a piece of content. It groups pages that are the same or nearly the same, picks the one it thinks most useful to searchers, and shows that one. The rest become duplicates, crawled less often and rarely shown.

Nobody sets out to build duplicates. They grow out of ordinary site features, and Google's own documentation lists the usual sources.

• Protocol and host variants: the http and https versions, with and without www.
• Site functions: category pages sorted and filtered a dozen ways, each with its own parameter string.
• Device variants: a separate mobile version of a page alongside the desktop one.
• Region variants: near-identical pages for the US and the UK in the same language.
• Accidents: a staging or demo copy of the site left open to crawlers.

Add tracking parameters from ads and email, trailing slashes, uppercase letters and printer versions, and a fifty-page site can expose several hundred URLs. Google says plainly that some duplication within a site is normal and does not break its spam policies. The cost is not a penalty. It is Google showing the wrong version, splitting links between copies, and muddying your reports. The technical SEO guide puts this in the context of crawling and indexing as a whole.

The signals Google weighs, and how much your vote counts

Google's canonical documentation, updated in July 2026, ranks the methods you control by influence. It also says the methods stack, so they work better together than alone, and that none of them is required: a site will likely do fine without declaring a preference at all.

Google's HTTPS preference has exceptions: an invalid certificate, insecure dependencies on the page, or a redirect from HTTPS back to HTTP. When signals disagree, Google's canonicalization page says your preference "is a hint, not a rule," and it decides. Most surprises come from that sentence. Redirects get their own guide, 301 redirects and SEO, because they are the strongest lever here.

Writing rel=canonical correctly: the tag, the header and the platforms

The HTML version is one line in the head of each duplicate, pointing at the preferred page: <link rel="canonical" href="https://example.com/work-boots">. For a PDF, send it as a response header instead: Link: <https://example.com/guide.pdf>; rel="canonical". Google's rules for the tag are short and strict.

1. 1.It counts only in the head. A canonical in the body is ignored.
2. 2.Use absolute URLs, scheme included. A relative path can resolve somewhere you did not intend.
3. 3.Put a self-referencing canonical on the preferred page itself.
4. 4.Choose one method per page, the tag or the header, because using both is more error prone.
5. 5.Do not point the same page at different canonicals through different methods.
6. 6.Put the tag in the HTML source and do not let JavaScript change it after load.
7. 7.Never use a URL fragment (the part after #) as the canonical.

Whatever the platform, view the page source and search for rel="canonical". There should be exactly one, it should be absolute, and it should match the URL you want in results. The meta tags guide covers the other lines that share the head with it.

Canonical mistakes that quietly remove good pages from Google

A canonical tag is a request to hide one page in favor of another. Point it the wrong way and Google obliges. Several of these come from a Google Search Central post on rel=canonical mistakes, published in 2013 and still on Google's site; the rest from its current documentation.

• Every page canonical to the home page. A theme setting or a hand-edited header applies one URL site-wide, and Google is asked to treat every service page as a copy of the home page.
• Pages 2, 3 and 4 canonical to page 1. Google's pagination guidance says not to do this; give each page in a series its own canonical, or the products on later pages may never be indexed.
• A category page canonical to a featured article. The category can drop out of results in favor of the post.
• A canonical pointing at a page that fails. The target should load with real content, not a 404, a soft 404, a redirect or a noindex. Google's 2013 post says to check that the target is not noindexed.
• Two canonicals pointing at different URLs. Google says it will likely ignore all of them.
• Using noindex or robots.txt to pick the winner between two versions. Google advises against both for canonicalization; noindex removes the page from Search entirely, and a blocked page's tag is never read. Blocking whole filter patterns, below, is a different job: keeping crawlers out, not choosing a version.
• Canonicals on syndicated articles. Google now says partners who republish your content should block it from indexing instead.

Hacked sites add a twist. Google warns that an attack can insert a cross-domain canonical or redirect pointing at spam, and its systems may then choose the malicious URL. If Search Console shows a canonical on a domain you do not own, treat it as a security problem first.

Filters, sort orders and product variants on an online store

Stores create duplicates faster than any other kind of site, because every filter combination is a new URL. Google's guidance on faceted navigation gives two routes, and the right one depends on whether filtered pages deserve to rank.

For variants, Google's product variant guidance says a single-page product with color and size options should expose one canonical URL for the whole product group, while each variant stays reachable through its own URL parameters. Paginated category pages each keep their own canonical. Google no longer uses rel=next and rel=prev, though other engines may. The ecommerce SEO section covers store structure, and the ecommerce SEO audit turns these checks into a list. Building a store from scratch? Our ecommerce development work sets canonicals in code from day one.

Tag, redirect, noindex or nothing: picking the right tool

Most canonical trouble starts with the wrong tool. Use this as the tiebreaker.

That last row deserves emphasis for local businesses. Town pages that differ only in the city name are the duplicates Google clusters most readily, and a canonical cannot fix pages that should not exist in the first place. Our guide to service area pages shows how to make each one earn its place.

Checking which URL Google actually chose

1. 1

   Inspect a key URL

   Run it through URL Inspection in Search Console. Compare the user-declared canonical, which is what your tag says, with the Google-selected canonical. If they match, you are done with that page.

2. 2

   Do not rely on the live test for this

   Google says the live test cannot predict whether a page will be chosen as canonical, because that is decided at indexing time. Read the indexed result instead.

3. 3

   Read the Page indexing reasons

   "Alternate page with proper canonical tag" means Google honored your tag; Google says there is nothing to do. "Duplicate without user-selected canonical" means you gave no preference. "Duplicate, Google chose different canonical than user" means it overruled you.

4. 4

   Find the conflicting signal

   When overruled, check whether internal links, the sitemap, redirects or hreflang point elsewhere, and whether the two pages are really different enough to stand alone.

5. 5

   Fix, then wait

   Google says it may hold pages in a duplicate cluster for up to two weeks before re-evaluating. Request indexing for your most important URLs only, since requests are limited.

On a large site, do this by crawler rather than by hand. The free site audit checks SEO and content issues and emails a PDF, and the website health check catches page-level faults in seconds. Pages missing from Google entirely, rather than replaced by a twin, are a different problem covered in website not indexed by Google. If you would rather hand canonicals, redirects and the rest to someone who does this weekly, our SEO programs include it, priced on the SEO pricing page. Start from the top of the SEO guide to see where this fits.