Canonical tags: how to check one is working

1Check the canonical and its target 1 minute

Open DevTools (F12), go to the Console and paste:

Copy// Checks this page's canonical, then the page it points to (if it's on this domain)
const clean = u => u.split('#')[0];
const here = clean(location.href);
const res = await fetch(here, { cache: 'no-store', headers: { Accept: 'text/html' } });
const server = new DOMParser().parseFromString(await res.text(), 'text/html');
const tags = doc => [...doc.querySelectorAll('link[rel~="canonical" i]')];
const linkCanonical = h => (h || '').split(/,(?=\s*<)/)
  .map(p => p.match(/^\s*<([^>]*)>(.*)$/))
  .find(m => m && /;\s*rel\s*=\s*("(?:[^"]*\s)?canonical(?:\s[^"]*)?"|canonical\s*(?:;|$))/i.test(m[2]))?.[1] || '';
// noindex or none for all robots or Googlebot; ignores other bots and values such as max-image-preview:none
const noindex = (metas, header) => {
  const hit = text => text.split(/[\s,]+/).some(w => w === 'noindex' || w === 'none');
  const meta = metas.some(c => c.toLowerCase().split(',').some(p => !p.includes(':') && hit(p.trim())));
  let agent = '';
  const head = (header || '').toLowerCase().split(',').some(p => {
    let text = p.trim();
    const m = text.match(/^([a-z0-9_-]+)\s*:\s*(.*)$/);
    if (m && !/^(max-snippet|max-image-preview|max-video-preview|unavailable_after)$/.test(m[1])) { agent = m[1]; text = m[2]; }
    else if (m) return false;
    return (!agent || agent === 'googlebot') && hit(text);
  });
  return meta ? 'YES (meta tag)' : head ? 'YES (X-Robots-Tag)' : 'no';
};
const rendered = tags(document);
const header = linkCanonical(res.headers.get('link'));
const headerUrl = header ? clean(new URL(header, here).href) : '';
const canonical = rendered[0] ? clean(rendered[0].href) : headerUrl;
const serverUrl = tags(server)[0] ? clean(new URL(tags(server)[0].getAttribute('href') || '', here).href) : '';
const report = {
  'Canonical tags (server / rendered)': tags(server).length + ' / ' + rendered.length,
  'Changed by JavaScript': rendered[0] && serverUrl && serverUrl !== canonical ? 'YES: server HTML says ' + serverUrl : 'no',
  'Link header canonical': !headerUrl ? '(none)' : rendered[0] && headerUrl !== canonical ? headerUrl + ' (DIFFERS from the tag)' : headerUrl,
  'Canonical URL': canonical || '(none)',
  'Full URL in the tag': !rendered[0] ? '' : /^https?:\/\//i.test(rendered[0].getAttribute('href') || '') ? 'yes' : 'NO: use a full URL',
  'In the head': !rendered.length ? '' : rendered.every(l => l.closest('head')) ? 'yes' : 'NO: one is in the body',
  'Points to': !canonical ? '' : canonical === here ? 'this page'
    : !canonical.includes('?') && canonical === here.split('?')[0] ? 'this page, without the query string'
    : 'a DIFFERENT page',
};
if (canonical && canonical !== here && canonical.startsWith(location.origin + '/')) {
  const t = await fetch(canonical, { cache: 'no-store', headers: { Accept: 'text/html' } });
  const target = new DOMParser().parseFromString(await t.text(), 'text/html');
  const metas = [...target.querySelectorAll('meta[name="robots" i], meta[name="googlebot" i]')].map(m => m.getAttribute('content') || '');
  const ownRaw = target.querySelector('link[rel~="canonical" i]')?.getAttribute('href') || linkCanonical(t.headers.get('link'));
  const own = ownRaw ? clean(new URL(ownRaw, t.url).href) : '';
  report['Target status'] = t.redirected ? 'REDIRECTS to ' + t.url + ' (' + t.status + ')' : t.ok ? t.status : t.status + ': ERROR';
  report['Target noindex'] = noindex(metas, t.headers.get('x-robots-tag'));
  report["Target's own canonical"] = !own ? '(none)' : own === clean(t.url) ? 'itself' : own + ' (ANOTHER page: a chain)';
}
console.table(report);

It counts the canonical tags in the server HTML and the rendered page, flags a canonical changed by JavaScript, reads any canonical in the HTTP header and says if it differs, checks the tag uses a full URL and sits in the head, and says whether it points at this page. If it points at another page on the same domain, it opens that page and checks its status, noindex and whether its own canonical names itself.

If the canonical only drops the query string, it shows "this page, without the query string", which is usually what you want. A canonical with a different query string, such as ?page=1 on page 2, counts as a different page. Targets on another domain can't be fetched from the Console: open those yourself.

2See which canonical Google chose 2 minutes

In Google Search Console, paste the URL into URL Inspection and open Page indexing. It shows the User-declared canonical (yours) and the Google-selected canonical. Look at the indexed result, not the live test: the live test can't show which canonical Google chose. If they differ, Google isn't following your tag. Look for the conflicting signal: internal links, redirects or the sitemap using a different URL.

3Look for duplicates in the source 1 minute

Press Ctrl+U and search for canonical. There should be one rel="canonical" link, near the top of the head. Two usually means the theme and an SEO plugin both add one.