Canonical URL Tag Generator

The same page is often reachable at several addresses. Add a tracking parameter, a session identifier, a trailing slash, an uppercase letter, or a print variant and you have another URL serving identical content - and a crawler with no way of knowing which one you consider authoritative. A canonical tag is how you tell it.

This generator produces that tag. Enter the URL you want treated as the definitive version and it emits a complete link element, with the URL escaped so it is safe to place inside an HTML attribute, ready to paste into your document head.

The tag is one line of markup and the decision behind it is not. Choosing the wrong canonical is worse than having none, because it actively instructs search engines to consolidate signals onto a page you did not mean.

The default correct answer for most pages is a canonical pointing at themselves. A page whose canonical is its own clean URL is protected against every parameter variant that might be appended to it later: a link shared with UTM parameters, a filter state, a session token. All of them resolve back to the one URL you nominated, and the ranking signals consolidate there instead of scattering.

The cases where a canonical points elsewhere are specific. A paginated series where you want everything credited to the first page. A product available in several colourways at separate URLs where one is the primary. Syndicated content republished elsewhere, where the syndicated copy should point back at your original. In each of these the target page must be substantially the same content - a canonical is a claim about duplication, not a redirect and not a preference.

Pointing a canonical at a page with meaningfully different content is where real damage happens. Search engines may honour it and drop your page from the index in favour of the target, or may ignore it and treat every canonical on your site with more suspicion. Both outcomes are worse than the duplicate content you were trying to fix.

Use the full absolute URL including the scheme. A relative path is technically permitted and resolves against the current document, which works until the tag is served from an unexpected path and starts resolving somewhere wrong. Absolute URLs have no such failure mode and cost you nothing.

Be consistent about the details that create distinct URLs. https and http are different. www and the bare domain are different. A trailing slash and no trailing slash are different. Uppercase and lowercase paths are different on most servers. Whichever form your site actually serves - the one your redirects settle on - is the one the canonical must name, and mixing forms across your templates produces a site where every page canonicalises to a URL that immediately redirects.

That redirect chain is the most common self-inflicted error. A canonical pointing at a URL that 301s to another URL is asking a crawler to follow an indirection to find your preferred page, which weakens the signal and sometimes breaks it. Test the canonical you generate by loading it and confirming it returns 200 directly.

Your input is trimmed and passed through an attribute-escaping step that replaces ampersands with &, double quotes with ", and less-than signs with <. Those three are the characters that would otherwise break out of the attribute or the tag, so escaping them is what makes a URL containing query parameters safe to embed. The escaped string is then wrapped in a complete self-closing link element with rel set to canonical.

Separately, the raw input is checked by attempting to construct a URL object from it. If that fails you get a warning that the format may be invalid - and the tag is still generated, deliberately, because the parser is stricter than some legitimate inputs and a warning is more useful than a refusal.

Nothing else is validated. The tool does not fetch the URL, does not check that it resolves, does not follow redirects, and does not compare the target’s content against the page you intend to put the tag on. It is a formatter with a syntax hint, running entirely in the browser.