Connect publisher identity with JSON-LD sameAs and Wikidata

Start with the identity you control

Name the actual publisher. Add its official URL and a description that matches the page, then show a useful about, contact or author link. A personal project can identify a Person; it does not need an invented company or a page full of social links.

The scanner gives authority-category credit for observable identity and accountability evidence. It does not measure reputation or how much an AI model trusts a brand. Supported external profiles can supply evidence, but a site can satisfy the rubric without Wikipedia or Wikidata.

• Read the identity evidence checks
• Inspect the page and live search candidate

What belongs in sameAs?

Schema.org defines sameAs for a page that unambiguously identifies the entity. Use the official profile for the organization or person described by that node. A related article, a directory search result or a different company with the same name is not the same identity.

Give the publisher a stable @id and link page publisher references to it. Keep author Person nodes separate from an Organization. A founder profile belongs to the founder; it should not be copied into the company sameAs array.

<script type="application/ld+json">
{
  "@context":"https://schema.org",
  "@type":"Organization",
  "@id":"https://example.com/#publisher",
  "name":"Actual publisher name",
  "url":"https://example.com/",
  "description":"The same description readers can verify on the site"
}
</script>
<!-- Add sameAs only after confirming an exact external identity. -->

• Schema.org sameAs definition

Search returns candidates, not a resolved identity

A live name search can return several plausible Q-IDs. The scanner derives its search name from the domain, so abbreviations, brand names and legal names can produce misses or unrelated results. It deliberately does not award verified-match points for this lookup.

Inspect the candidate description and official website property P856. Check whether the domain and entity type match the publisher. A missing website property leaves work to do; a different official domain is a reason to stop and investigate.

1. Compare the candidate name and aliases with the publisher.
2. Check the official website and distinguish parent companies, products and people.
3. Add the canonical Wikidata URL only after confirming the match. Keep the evidence for that decision.

curl -G 'https://www.wikidata.org/w/api.php' \
  --data-urlencode 'action=wbsearchentities' \
  --data-urlencode 'search=Actual publisher name' \
  --data-urlencode 'language=en' \
  --data-urlencode 'format=json'

# For a candidate, inspect its entity document. Replace QID first.
# https://www.wikidata.org/wiki/Special:EntityData/QID.json

• Wikidata official website property

What if there is no Wikidata item?

Leave the property out. A missing external identifier is better than a false identity claim. Continue with accurate local markup, authorship and contact information.

Wikidata has its own notability rules. A scanner result is not a reason to create a promotional item or commission a Wikipedia page. Those projects serve their readers and editors, not a technical SEO checklist.

• Wikidata notability requirements
• Handle an unverified candidate finding

Check the full graph after editing

Inspect every schema block in the final HTML. Themes and plugins can each publish an Organization, leaving two names or conflicting URLs for the same publisher. Correct the original source when possible rather than adding a third block to override it.

Validate the graph, open every profile link and confirm the visible byline still matches the author node. This verifies your markup. It does not establish that ChatGPT, Claude, Perplexity or Gemini has accepted the identity or will select the page as a source.

• Remove conflicting schema sources
• Structured-data implementation guide