Canonical tags and duplicate pages

The same page routinely exists at several addresses without anybody deciding it should. With and without www. Over http and https. With and without a trailing slash. With a tracking parameter attached from an email campaign. With an uppercase letter somewhere in the path.

Each of those is a distinct address as far as a browser and a search engine are concerned. Canonical tags are the line inside the page that names which of them should be treated as the real one. This page covers what canonical tags do, how duplicate pages and duplicate content happen without anyone deciding it, and how rel canonical tags get set wrongly.

Where url variants and canonical tags come from

Source	Example variant
Server configuration	With and without www
Protocol	The insecure and secure versions both answering
Trailing slash	The same path with and without one on the end
Case	An uppercase character in the path
Tracking parameters	An address with campaign tags appended
Platform features	Filters, sorting, print views, pagination
Index files	The path and the same path plus a default file name

Almost none of this is a decision. It is the accumulated default behaviour of servers, platforms and marketing tools. It is invisible from the front end, because every variant looks correct in a browser. Google’s own documentation illustrates the same duplicate-urls problem with a product URL such as https://example.com/dresses/green.html, reachable through several https://example.com/dresses/green.html?ref= tracking-parameter variants — example.com/dresses is the address, and each tracking tag creates a new duplicate of it. The underlying mechanism is identical for any small-business page.

Why duplicate pages and poor canonicalization matter

Duplication inside a site is a selection problem, not a penalty. This is duplicate content in the narrow, technical sense — the same words at several urls, not copied from elsewhere. A search engine sees several addresses with the same duplicate content, picks one, and gives the others nothing. Three consequences follow.

The wrong address can win. The chosen one might be the version with a tracking parameter attached, which is the one people then see in results and share.

Signals are split. Links from elsewhere pointing at different variants are describing different addresses. Consolidating them is the main practical benefit of settling this.

Reporting is fragmented. Analytics and search reports show several rows where there is one page, which makes the numbers wrong in a way nobody notices.

Copying content from another website is a different matter entirely. That is not a duplication question — it is a copyright question. The Australian position on ownership of website content is on copyright in website content.

What a canonical tag, canonical url or canonical link actually does, and how to specify canonical page authority

A canonical tag is a canonical declaration, not an instruction. It tells a search engine which URL the owner considers authoritative for this content, via the rel="canonical" attribute — the rel canonical tag. The engine takes the canonical declaration as a strong signal, not as a command. Where it contradicts other evidence — internal links pointing somewhere else, a redirect saying something different — it can be set aside.

The practical implication is that canonical tags are not a substitute for consistency. The site’s own links should point at the canonical URL. The server should settle the www and protocol questions with redirects rather than leaving both live at http and https. The canonical tags, or rel canonical links, then agree with everything else rather than arguing with it.

The ways a rel canonical tag is set wrongly

Every page pointing at the home page. A configuration error that tells a search engine the entire site is one page. Every page except the home page then disappears from results, and the site looks completely normal to a visitor.

Pointing at a page that redirects. The canonical tag names a URL that is not the final one. The engine then has to resolve a contradiction, and may resolve it differently from what was intended.

Missing on a site with parameter variants. Common on stores and on any site running paid campaigns with tagged links.

Pointing across domains without intending to. Occurs when a staging site is copied to production with its canonical addresses left pointing at the staging domain. This is a close relative of the staging failure described on taking a website out of staging.

Duplicate pages and urls that are not simple address variants

A second class of duplication is genuine: several pages with content substantially similar to one another. On a small business site the usual sources are service pages generated from a template, and location pages differing only in the place name.

A canonical tag does not fix that. If two pages are nearly the same, either they should be one page or they should be genuinely different pages. The structural version of that decision is on whether each service gets its own page. The location case, which is the higher risk, is on structuring location pages honestly.

Where the canonical declaration should point

Two rules cover almost every case and both are easy to state.

A page should declare itself. Most pages on most sites are their own canonical URL. A self-referencing canonical tag is the correct default rather than a redundancy.

Where a page genuinely is a variant, its canonical tag should point to the version a person would most usefully land on. That means the clean URL without parameters, on the domain the business actually uses, over a secure connection, in the form the internal links use. A canonical declaration is a recommendation to a machine, and a recommendation that disagrees with the rest of the site is one that will be overruled.

What to check on your own site: http and https address variants

Four quick tests, each done in a browser address bar.

• Load your site with www and without. One should redirect to the other.
• Load it with http. It should redirect to the secure version.
• Add a trailing slash to a page address, then remove it. One should redirect to the other, or both should carry the same canonical declaration.
• Add a nonsense parameter to a page address. The page should still declare the clean address as canonical.

If all four behave, the common cases are handled. If any of them serves two live versions, that is the finding, and it is a server configuration task rather than a content one.

What to do next

Put “one page, one address” on the launch checklist as a tested item, rather than an assumed one. Re-test after any platform or hosting change, because these settings live in the layer that migrations disturb. The wider set of technical obligations in a build is on web development services.