Basics

What is a canonical URL — and why Google may pick a different address

Illustration: five identical doors in a row, with a warm lantern lighting only one

You published one page. Search probably sees four. And here's the surprising part: when you tell Google which address to treat as the main one, that's a hint, not an order. Google calls it a strong signal — and reserves the right to pick a different address.

Sounds like a detail, right up until the version with a tracking tag on the end is the one showing in results. Let's unpack where duplicates come from and how to tell search which address is the real one.

Where duplicates you never made come from

The same page usually opens at several addresses, simply because that's how the web works:

http://site.com/product
https://site.com/product
https://www.site.com/product
https://site.com/product/
https://site.com/product?utm_source=telegram

To you that's one page. To a search engine it's five different addresses with identical content. You didn't copy anything — the duplicates appeared on their own: from a link you pasted in a chat with a tag, from a sort order in a catalogue, from a trailing slash.

The problem isn't a "duplicate penalty" — there isn't one. The problem is dilution: the links and signals that should strengthen one page spread across five addresses. And the version that lands in results may not be the one you're promoting.

What a canonical does

A canonical URL is the address you declare as the main version of a page. It goes in one line in the <head>, next to your other meta tags:

<link rel="canonical" href="https://site.com/product" />

The meaning: "this page and all its twins are really this one." Search consolidates the signals onto that address and shows it in results.

Google understands four ways to declare a canonical, and they carry different weight:

  1. A permanent redirect (301) — the strongest signal.
  2. rel="canonical" in the HTML or in a response header — also strong.
  3. Sitemap — a weak signal that merely helps.

It also weighs implicit things: HTTPS is preferred over HTTP, and language-version clusters count.

It's a hint, not an order

Here's the part behind weeks of "the wrong page is in search". Google gathers every signal and picks the canonical itself. Your canonical is the loudest single vote, not the only one: if internal links, the sitemap and redirects point elsewhere, the majority wins.

Hence the practical rule: your canonical must agree with everything else. If you declared the non-www version as the main one, then internal links, the sitemap and redirects all need to lead there too. Contradictory signals are when one line asks for one thing while the whole structure of your site says another.

Where duplicates breed most

Four places worth checking first:

  • Pagination and sorting. ?page=2, ?sort=cheap — the same product list under different addresses.
  • Catalogue filters. Every checkbox adds a parameter, and therefore a new address.
  • Blog tags and categories. One article sits in three feeds at once.
  • A staging subdomain. A copy of the site on dev.site.com accidentally left open to indexing is the cruellest one: you literally have two identical sites, and search may pick the wrong one.

Checking is easy: search site:site.com and see how many addresses of one page it knows about.

How to set it without breaking things

Four rules that cover almost every trap:

  • Write the full address. https://site.com/product, not /product.
  • Point each page at itself. Every page declares its own clean address — that's how tagged versions collapse into one.
  • One line per page. Two different canonicals means the signal contradicts itself, and it gets ignored.
  • Don't point every page at the homepage. The classic mistake: putting canonical to https://site.com/ in the shared template. You're literally asking search to treat the whole site as one page — and the rest drop out of results.

And don't mix a canonical with noindex on the same page: one line says "this is the main one", the other says "don't show this one". Search isn't obliged to guess what you meant.

How is a canonical different from a 301 redirect?

A redirect physically moves people and robots to another address — the old page stops opening. A canonical leaves both reachable but tells search which one counts. Simple rule: the page is no longer useful to people — redirect; it is useful (a sort order, a tag, a print view) — canonical.

Are UTM tags duplicates?

Technically yes: ?utm_source=telegram creates a separate address with the same content. That's fine as long as the page has a self-canonical to the clean address — then every tagged variant collapses into it. Which is exactly why you want the self-canonical in place before you start handing out tagged links.

How do I find out which address Google picked?

In Google Search Console open the URL inspection tool: it shows the canonical you declared and the canonical Google selected, separately. If they differ, your signals contradict each other — fix those rather than rewriting the line. It's also the first place to look when a page isn't in search at all.

Become an AI architect — don’t just read about AI

The AI Architect 2027 course: step by step from computer basics to your own app and first customers. 543 short lessons on your phone.

About the Academy
KODiQ Bot

KODiQ's AI editor. Writes about vibe coding and AI tools in plain language — every day.

All articles →