Free tools / canonical-checker
FREE TOOL · RUNS ON OUR SERVER
Canonical Tag Checker
Reads a page’s rel="canonical", in the HTML and in the HTTP Link header, and follows where it points: to the page itself, to another page, to a redirect, or to a page that is not there.
Classic SEO, not AI visibility. Nothing on this page measures whether an assistant mentions or cites you.
Check a page
example output
not measured · nothing has been checked yet
| Level | About | Checked against | Finding |
|---|---|---|---|
| error | canonical | mechanical | 2 rel="canonical" links naming 2 different addresses. The page gives two answers to one question; Google’s guide does not say which one, if either, it uses. |
| warning | canonical | Google’s guide | The canonical http://example.com/trail-shoes redirects (HTTP 200) to https://example.com/trail-shoes/. Google treats a redirect as its own strong canonical signal, so this page is pointing at an address that points somewhere else; name the final address. |
| warning | canonical | Google’s guide | This https page names an http address as canonical. Google says it prefers HTTPS pages over equivalent HTTP pages as canonical. |
| note | canonical | ogp.me | og:url (https://example.com/shoes/) is not the address that answered (https://example.com/trail-shoes/?utm_source=newsletter). Shares of this page will be attributed to og:url. |
This check runs on our server, and it has to: a browser cannot read another site’s HTML or its response headers. The address you type goes to api.answermend.com, which fetches that page, the status of up to three addresses its canonical tags name, and the first 64 KB of its og:image. Nothing else is fetched, and no script on your page is run. What our server keeps of a run is set out on the privacy page.
WHAT IT CHECKS
Where the page says its real address is, and whether that address answers
A canonical tag is one line, and most of the ways it goes wrong only show when you follow it. So every address it names gets fetched, and you see what came back.
In the order it is read
- countNone, one, or several. Several naming different addresses is an error: the page gives two answers to one question. Several naming the same address is harmless, and noted.
- whereEach canonical is resolved to a full address and followed. It points to itself, to another page that answers, to a redirect (every hop is shown), or to a page answering 4xx. A target that errored or could not be reached is not measured. It is never counted as fine.
- placementA canonical outside the
<head>, one written as a relative path, one carrying a#fragment, and one carryinghreflang,lang,mediaortype. - headerThe HTTP
Linkheader, read forrel="canonical"and compared with the HTML. - og:urlogp.me defines
og:urlas the canonical URL too. When it andrel="canonical"name different addresses, the page is claiming two permanent addresses. - httpsAn https page naming an http address as its canonical.
- redirectsIf the address you typed redirects, every hop is listed, and the canonical is compared with where the redirects stop.
What it does not do, said out loud
It reads one page, not your sitemap or your internal links, and it cannot see which address Google has actually chosen. That is in Search Console’s URL Inspection, for sites you own. This tool reports what the page says, and what the addresses it names do when they are asked.
Only the first three addresses a page’s canonical tags name are fetched. Any beyond that come back as not read, with the reason.
GOOGLE’S OWN GUIDE
Every finding cites Google’s page, or says it is mechanical
Google publishes one guide to canonical URLs. Where a finding rests on it, the finding links it. Where the guide is silent, the finding is labelled mechanical and does not put words in Google’s mouth.
What the guide says
Use absolute paths rather than relative paths. The link element is only accepted in the <head>. Do not give a URL fragment as canonical. Do not name different canonicals for one page through different methods.
Canonical annotations carrying hreflang, lang, media or type are not used for canonicalization. HTTPS is preferred over HTTP. A redirect is a strong signal that its target should become canonical. The HTTP header form is documented for non-HTML files such as PDFs.
Where it is silent
The guide does not say what happens when one page carries two canonical tags naming different addresses. We flag that as an error on the plain fact that the page gives two answers, and we label it mechanical, not Google.
A canonical target that answers 404 is flagged the same way. No guide is needed to say a page that is not there consolidates nothing.
Google’s guide was read on 2026-09-13.
QUESTIONS
What people ask before they run it
My page has no canonical tag. Is that an error?
No, it is reported as a note. Google’s guide says that when you do not specify a canonical URL it identifies the best version itself. The tag is how you make that choice instead of leaving it to Google.
Why is a canonical that redirects only a warning?
The page still points somewhere real, just not directly. Google’s guide calls a redirect a strong signal that its target should become canonical, so a canonical naming an address that redirects elsewhere sends two signals about one page. Naming the final address removes the doubt.
What if the canonical address could not be fetched?
It comes back as not measured, with the reason. A timeout or a server error on our side tells you about that moment, not about your page, so it is never counted as fine and never as broken.
Is this an AI visibility check?
No. Canonical tags are classic SEO, and nothing on this page measures whether an AI assistant mentions or cites your site. That is a separate question, answered by a separate free check.
THE OTHER KIND OF VISIBILITY
A canonical tells a crawler which page.
An answer is a different question.
Whether an AI assistant names you when someone asks about your category is not something any tag on this page can show. The AI Visibility Checker asks.