Build a sameAs mesh that covers several kinds of surface
What the scan checks: it sorts your Organization schema’s sameAs entries into tiers and wants at least three different tiers covered, or two if one of them is Wikidata or Wikipedia. The tiers are:
| Tier | Examples |
|---|---|
| Knowledge graph | Wikidata, Wikipedia |
| Professional | LinkedIn, Crunchbase, GitHub |
| Citation surfaces | YouTube, Reddit |
| Review platforms | G2, Capterra, Trustpilot, Yelp, Clutch |
| Public registries | OpenCorporates, Companies House, SEC |
Anything else (X, Facebook, Instagram, TikTok, and the long tail) counts as a social profile. Socials are fine to list, they just do not move this check.
Why the rule changed on 30 July 2026. It used to be a bare count: five entries passed. That scored eight links to eight social networks identically to Wikidata plus Wikipedia plus LinkedIn plus Crunchbase plus GitHub, which inverts the whole point of the property. Five copies of the same weak signal are not five corroborations. If your score moved on a re-scan, this is why. The count is still shown in the evidence.
Only external profiles count. A sameAs entry pointing back at your own @id or your own homepage is excluded from the total, and the scan names the ones it excluded. The property means “this entity is also the thing at that URL” — it exists to connect your local declaration to surfaces elsewhere that describe you. Pointed at yourself, it disambiguates you from nobody and just pads the count.
Dead profiles are named. Each unique target gets one lightweight request. A target that answers 404 or 410 is reported as dead: a profile that moved or was deleted still looks like a link in your markup but corroborates nothing (this is the classic post-migration failure). LinkedIn is marked “not verifiable” rather than dead, because it answers a server request with a login wall whether or not the page exists, and anything other than a 404 or 410 is left alone: it is as likely to be about us as about the profile.
Why it matters for AI search visibility
Section titled “Why it matters for AI search visibility”sameAs is the list of authoritative profiles that are the same entity as your brand. Each one is a corroborating vote: when your homepage, your Wikidata item, your LinkedIn, and your Crunchbase all point at each other, an engine can confidently resolve every mention to one entity instead of guessing whether “Acme” and “Acme Blog” are the same company. This is what keeps your brand from fragmenting across the labs’ memory. A single link tells the engine little; a mesh of five or more high-authority profiles is what makes the entity solid and citable by name.
How to fix it
Section titled “How to fix it”Extend the sameAs array in priority order, strongest surfaces first:
"sameAs": [ "https://www.wikidata.org/entity/Q12345678", "https://en.wikipedia.org/wiki/Example_Inc", "https://www.linkedin.com/company/example", "https://www.crunchbase.com/organization/example", "https://www.g2.com/products/example", "https://github.com/example", "https://twitter.com/example"]- Wikidata first (it propagates through the Knowledge Graph and into training data), then Wikipedia if you have an article, then high-authority directories (Crunchbase, G2 or Capterra, LinkedIn), then the long tail of socials and industry directories.
- Only list profiles that genuinely exist and point back where the platform allows, so the corroboration is bidirectional.
- Use the same array everywhere. The homepage Organization schema is the canonical source; keep it consistent across surfaces.
Add a different KIND, not another of the same kind. If you already have LinkedIn, Crunchbase and GitHub, a fourth professional profile does not move this check; a G2 listing or a Wikidata item does. Aim past the minimum: the manual’s target list is Wikipedia, Wikidata, Crunchbase, LinkedIn, X, YouTube, G2 or Capterra, GitHub if relevant, plus the industry directories that matter in your vertical. Pair this with the Wikidata work in E3 and the base Organization schema in E1.