The source contains rel="canonical", but the element sits in the page body. Presence alone is insufficient: Google expects an HTML canonical declaration in the page’s valid head section.
Trace the insertion point
Fictional example: an article component was copied from a complete HTML template. It outputs its own canonical below the heading, while the shared layout produces the actual head.
Inspect the original page source first, then the browser’s DOM tree. Browsers may relocate incorrectly nested HTML, so the rendered view alone does not always explain what the server supplied.
Record which template generates the element. The correction usually belongs in shared metadata handling, rather than another canonical added at the bottom of the page.
Look for an existing canonical in the head. Adding a second, different value would create another fault. Check any Link response headers too.
The selected destination should match the content and preferred public address. Correct placement does not fully solve an incorrect hostname.
A screenshot of the visible article or a text search for “canonical” is insufficient for acceptance. Include the observed placement and actual destination URL in the technical report.
Remove competing output paths
A theme and an SEO plugin can both emit canonicals. Collect every element and any Link response headers. Compare their complete destinations. Even identical duplicates reveal multiple owners; different values contradict one another.
Assign one owner for metadata without blindly removing other components’ necessary functions. Afterwards, check the affected page and another page type. Google chooses the canonical URL; unambiguous declarations remove conflicts but cannot force its choice.
Resolve the hostname and base URL
href="/guides/" on preview.example.com refers to the preview host unless another base changes resolution. A base element can also affect the result. Inspect the fully resolved address and the public base-URL configuration. Google recommends absolute canonical URLs.
A canonical pointing to the main domain does not protect a confidential preview. Check access separately and reconcile canonicals, sitemaps and internal links on the public website.
Keep section anchors separate from document addresses
A content link such as /guides/delivery/#returns can navigate to a section. The canonical should identify the stable document address. Google generally does not support fragments as canonical URLs.
Check the original response and the output after navigating between sections. If metadata inherits the anchor, correct that generator. Do not blindly remove fragments from hash routes such as /#/product/red: first establish how those views can receive directly accessible page URLs.
Related check: Check a canonical redirect destination.
An identified metadata-output correction can be requested through Frontend optimisation.