Canonical Graph Auditor
Use this workbench when a page-audit export makes it hard to see how canonical declarations connect. It follows URL nodes until a self-canonical, cycle, inconsistent declaration or absent target is reached. The report checks the supplied relationships; it does not identify the canonical URL selected by a search engine.
Key features
- Trace declaration paths and stopping points across up to 1,000 records
- Distinguish self-canonicals from multi-node cycles
- Flag chains, conflicting declarations, absent targets and supplied status/indexability
- Mark cross-origin links, relative references and fragments for review
- Retain source record numbers and normalized duplicate declarations
- Export per-URL CSV and full nodes/edges JSON
How to use
- Prepare CSV with url,canonical,status and optional indexable columns.
- Paste it or select a UTF-8 file and confirm the delimiter.
- Run the audit and inspect node counts, stopping-point groups and flags.
- Open path details to inspect the sequence and original declarations.
- Export CSV or JSON and perform any required live-page verification separately.
Use cases
- Review canonical chains before a site migration
- Inspect an A→B→A cycle and URLs that lead into it
- Find inconsistent canonical declarations across repeated export records
- Separate cross-origin declarations and missing audit targets for follow-up
Frequently asked questions
Does the result show Google’s or Bing’s selected canonical?
No. It links declarations in your CSV only. It does not visit URLs, compare content, inspect search-engine selections or check indexing. Verify live selection and crawl status in the relevant search-engine tools.
Is a self-referencing canonical a cycle?
A self-canonical is treated as a stopping point. A path reaching a multi-node cycle receives a loop flag. Sharing a stopping point also does not prove that the pages contain duplicate content.
How is an absent CSV target different from a 404?
An unknown target has no record in the supplied CSV; it is not inferred to be a live 404. A 404-related flag relies on a supplied status=404 declaration. Blank indexable values remain unknown, not false.
Are cross-origin or relative canonicals always wrong?
No. Cross-origin declarations are flagged so their intent can be reviewed. Relative references are resolved against the source URL and flagged. Fragments are removed from graph identity while their presence remains recorded.
What happens to repeated source URLs?
Normalized duplicates are grouped with their source records retained. Disagreement in canonical, status or indexable creates a conflict and stops traversal there. The tool does not silently select the first declaration as definitive.
What are the input and path display limits?
CSV supports 1 MiB and 1,000 records; raw and serialized URLs each support 2,048 characters. Tables use 50-row pages and path previews show the first 50 nodes with an omitted count. JSON retains all nodes, edges and total hops for reconstruction.
Privacy
CSV and file contents are processed in current-page browser memory. Input URLs are never visited or fetched, and input is not automatically sent to servers, analytics, URLs or browser storage. Files are created only on an explicit download request; changed inputs, cancellation and leaving the page discard previous work.
Comments & questions