INDUSTRY FIELD GUIDE
Search portal canonical snapshot records for AI
A page can have a publisher-declared preferred URL and a different canonical selected by a search system.
NAICS 519290 · Web Search Portals and All Other Information Services
A page can have a publisher-declared preferred URL and a different canonical selected by a search system. Google explains canonicalization and indexing as distinct processes. A portal can test whether an assistant interprets an invented crawl and index snapshot without assuming a preference has been followed. Google Search Central canonicalization
Use fictional pages with a canonical declaration, an index observation, and a later crawl record. Make one publisher declaration change without evidence of a new index selection. Ask what can be reported at each date and which evidence would resolve the mismatch. Use generic domains and an independent answer key.
Do not include real search queries, private ranking signals, user identities, crawler infrastructure details, or pages you lack permission to reuse. Keep access records out of the dataset.
Describe a synthetic index records exercise
Original Fiund exercise
A changed canonical declaration does not establish a changed index selection
Fictional pages on example.test have separate publisher declarations, crawl observations and index snapshots. The exercise makes no statement about an actual site's rankings.
| Record | Source | What it records |
|---|---|---|
| SCN1 | Day-1 crawl | Page /a declares /b as preferred URL. Day-2 index snapshot selects /a. |
| SCN2 | Day-5 crawl | Publisher changes /a declaration to /c. Crawl confirms that markup; no day-5 index snapshot supplied. |
| SCN3 | Day-7 index snapshot | Search-system observation still selects /a. No later snapshot exists in the packet. |
Question for the model
What preferred URL did the publisher declare on day 5, and what selected canonical is actually evidenced after that change?
Show the reference answer and checks
SCN2 documents the publisher's day-5 preference for /c. SCN3 separately evidences /a as the search system's selected canonical on day 7. A crawl confirming changed markup does not prove the preference was adopted by the index. Report both dated signals and their disagreement rather than calling /c selected, indexed or higher-ranking from the declaration alone.
- Treating a publisher-declared canonical as the search system's verified selection.
- Using a successful crawl as proof of indexing or ranking improvement.
Fields to preserve
- Page and declared preference
- Captures the publisher-controlled signal.
- Crawl timestamp
- Dates the observed markup.
- Index snapshot and selected URL
- Separately records the search-system observation.
Test another case
Provide a later snapshot selecting /c but no search-impression evidence. Require the model to resolve selection without claiming traffic or ranking gains.
Use generic fictional domains and original records. Exclude real queries, private ranking signals, user data, crawler infrastructure and access logs.
Discuss a contribution
Describe the sources you control, the task they could support and what has already been checked. Use the existing-data assessment for an archive or the collection brief for a new recording project. The contributor agreement defines Fiund’s licensing authority and compensation. A guide or fictional exercise does not establish an accepted collection or a sale.