Canonical Form Drift

Finds canonical URLs that disagree in form (www, scheme, trailing slash, tracking params) across the site

Rule ID	core/canonical-form-drift
Category	Core SEO
Scope	Site-wide
Severity	warning
Weight	3/10

How it works

Every canonical this rule looks at is individually valid. What is wrong is collective: half your site canonicalises to https://www.example.com/thing/ and half to https://example.com/thing, so the two halves consolidate onto different URLs and the ranking signal splits between them. No per-page rule can see this, because a page only knows its own canonical.

The rule compares four axes of URL form across every canonical in the crawl:

• Scheme: https versus http.
• Host: www versus apex.
• Trailing slash: present versus absent.
• Tracking parameters: retained versus stripped (utm_*, gclid, fbclid, ref, and friends).

For each axis it works out the site’s dominant form and reports only the pages that break it.

It stays quiet unless it has grounds to speak:

• Fewer than 10 crawled pages: no site norm exists, so the rule is skipped.
• Fewer than 10 comparable canonicals on an axis: that axis is dropped, not guessed at.
• No form reaching 70% on an axis: the site has no convention there, so nobody on that axis is a deviant. A site that genuinely mixes forms is a judgement call for a human, not an accusation from a rule.

Pages built from one template emit canonicals in one form, so they define the norm rather than deviate from it. A consistently param-bearing canonical is a form like any other: if every page does it, the site is not contradicting itself and nothing is reported.

Two refinements keep the trailing-slash axis honest, because it is the one most likely to accuse a healthy site:

• It is compared within a section (the first path segment of the canonical), not site-wide. Using a slash in /blog/how-to/ and none in /product is a normal shape, so each section is judged against itself and needs its own 10 pages and 70% agreement.
• The site root (/) and file-like paths (/feed.xml) never carry a trailing slash, so they are left out of that comparison entirely.

Validity is a different rule

This rule reports form only, so one root cause is never reported twice:

• A missing, relative or unparseable canonical is Canonical URL.
• A Link: rel="canonical" header that disagrees with the tag is Canonical Header.
• A canonical pointing at another host is Canonical URL as well, and its form follows the other site’s conventions, so it is excluded here.

Pages in any of those states are dropped before the comparison runs.

Canonical URL also notes, as info, that a page’s canonical points somewhere other than itself. That is a per-page observation with no claim about the rest of the site, and it is exactly the input this rule reads: the finding here is the site-wide pattern those individual notes add up to.

Solution

Each of these canonicals is valid on its own, but the site does not agree with itself: some pages canonicalise to one URL form and some to another. Search engines treat the two forms as different URLs, so links, crawl budget and ranking signals split between them.

Pick one form, the one your redirects already send traffic to, and emit it everywhere: one scheme (https), one host (www or apex, not both), one trailing-slash convention, and never any tracking parameters in the canonical. Canonicals are usually generated in one template or middleware, so this is normally a single fix in the place that builds the URL rather than a per-page edit.

Enable / Disable

Disable this rule

[rules]
disable = ["core/canonical-form-drift"]

Disable all Core SEO rules

[rules]
disable = ["core/*"]

Enable only this rule

[rules]
enable = ["core/canonical-form-drift"]
disable = ["*"]