Should a canonical URL be absolute or relative?
Rubric check: relative canonical URL (check id canonical_relative)
A canonical URL should be absolute, such as https://example.com/shoes/. Google accepts a relative one like /shoes/, but recommends full URLs, because a relative link is resolved against whatever address the page was fetched from.
Why it matters for SEO
A relative canonical works most of the time. It goes wrong when the same page is reachable on another host or protocol, such as http://, a www version or a staging domain. The relative link then points at that copy instead of your main one.
Google treats the canonical as a hint, so it may still pick the right URL. An absolute URL removes the guesswork.
Does this matter for ChatGPT and AI Overviews?
The effect on ChatGPT or AI Overviews is small. AI search tools mostly pick pages from search indexes, and the index decides which URL to keep.
A clean, absolute canonical helps the index settle on one address, which is the address AI answers then link to.
How Rubric scores it
MinorMinor checks cost a quarter to a half of a standard check when they fail.
Why this tier: A relative canonical works, but it's fragile and can point at the wrong host or protocol.
It is judged on every page Rubric scores.
Every Rubric score starts at 100. When this check fails, it takes up to 3 points off. Even one failing page costs at least 1.5 points (half of the most), and the rest comes off as the share of affected pages grows. A warning costs less than a failure.
All the minor checks share one limit: together they can take off at most 25 points, however many of them fail.
This check doesn't cap your score. It only takes off its own points.
From the engine's scoring files, scoring 2026-v2.1 (engine d5c591d).
How common it is
Found on 2 of 300 sites where Rubric could judge it (0.7%). That is 2 of 287 judged sites in Rubric's 313-site test set and 0 of 13 sites audited on Rubric between 11 Nov 2026 and 17 Nov 2026 under the current check rules (scoring 2026-11k or later). Measured 17 Nov 2026 with scoring 2026-v2.1.
How accurate the check is
Lab test: passed. On test pages built with this problem, it caught 3 of 3. On clean test pages, it stayed quiet 2 of 2 times. (Engine build bbcec7c.)
Real sites: people checked 8 of its findings by hand and 0 were wrong (0%). Last checked 14 Nov 2026. That's a small sample, so treat the rate as rough.
Lab as of 16 Nov 2026; judged findings as of 15 Nov 2026.
How to fix it
WordPress
- WordPress core and SEO plugins (Yoast SEO, Rank Math, All in One SEO) output absolute canonicals by default.
- If the report shows a relative canonical, check the Canonical URL field in your SEO plugin's panel on that page (often under Advanced) and enter the full URL, starting with
https://. - If no field is filled in, look for a theme or plugin that prints its own canonical tag, and turn that off.
Shopify
Shopify themes print the canonical with {{ canonical_url }}, which is absolute. If Rubric reports a relative one:
- Go to Online Store > Themes > Edit code and open
theme.liquid. - Find
rel="canonical"and make sure thehrefis{{ canonical_url }}, not a hand-typed path. - Check that no app adds a second canonical tag.
Webflow
- Go to Site settings > SEO and set the Global canonical tag URL to your full domain, for example
https://www.example.com. - If you set a page's canonical in its Page settings, enter the full URL rather than a path.
- Remove any canonical tag added by hand in custom code. Publish.
Custom code
Use the full URL, with the protocol and host you want indexed:
<link rel="canonical" href="https://www.example.com/shoes/">Build it from a fixed site URL setting, not from the request's host header.