Wikidata is the underrated lever in entity SEO. While everyone talks about Wikipedia, Wikidata does the actual knowledge-graph work — structured, machine-readable, ingested directly by Google's KG engine, Bing, LLMs and knowledge aggregators. A maintained Wikidata item is often more valuable for entity SEO than a weak Wikipedia stub that Google does not weight as an authority signal.
The anatomy of Wikidata
Every Wikidata entity has a Q-ID (e.g., Q2283 for Microsoft, Q42 for Douglas Adams). The ID is the stable reference that never changes. The item carries labels in many languages, descriptions for disambiguation, aliases for alternative spellings and a set of statements. A statement consists of a property (P-code) + value + references. Example: P31 (instance of) + Q4830453 (business) + reference to a commercial register entry.
The quality of a Wikidata item is directly visible in the density and quality of its statements. An item with five properties and no references is weak. An item with thirty properties, all referenced to independent sources, is strong. That density acts directly on Google's KG confidence and LLM entity resolution.
stable identifier — foundation for the @id graph and sameAs
properties as the target threshold for a strong item
required per statement — without refs, a weak signal
The property set for companies
Required properties: P31 (instance of → business / company / startup), P17 (country), P571 (inception date), P856 (official website), P1454 (legal form).
Strong properties: P112 (founded by — reference to a founder item), P169 (chief executive officer), P452 (industry), P361 (part of — for group structures), P159 (headquarters location), P1128 (employees).
Identifiers: P1320 (OpenCorporates), P2611 (LEI code), P5385 (Crunchbase), P4264 (LinkedIn), P2013 (Facebook ID), P2002 (X handle), P2003 (Instagram), P2397 (YouTube channel). This identifier chain is the single most important LLM entity-resolution lever.
The property set for people
Required: P31 (human), P106 (occupation), P569 (date of birth), P27 (country of citizenship), P21 (sex or gender).
Strong properties: P108 (employer), P69 (educated at), P463 (member of), P1416 (affiliation), P800 (notable work), P166 (awards).
Identifiers: P6379 (ORCID), P4264 (LinkedIn), P214 (VIAF), P244 (LoC ID), P245962 (GitHub username), P2003 (Instagram), P2002 (X), P2850 (Apple Music ID).
| Category | Company (property) | Person (property) | Priority |
|---|---|---|---|
| Instance type | P31 instance of → business | P31 instance of → human | Required |
| Location / origin | P17 country, P159 HQ | P27 citizenship, P19 place of birth | Required |
| Occupation / industry | P452 industry | P106 occupation | Required |
| Date | P571 inception | P569 date of birth | Required |
| Leadership | P112 founded by, P169 CEO | P108 employer | Recommended |
| Official website | P856 official website | P856 official website | Required |
P4264 company ID | P4264 personal ID | Recommended | |
| Academia / ORCID | — | P6379 ORCID | YMYL-required |
| Identifier code | P2611 LEI, P5385 Crunchbase | P214 VIAF | Strengthening |
| Education | — | P69 educated at | Recommended |
Is your Wikidata item strong enough?
30 minutes of live analysis: property count, reference density, identifier-chain quality. A clear priority list for what to add within the next 30 days.
Reference discipline
Wikidata recognizes three reference types: stated in (source is a structured database such as VIAF), reference URL (a web URL), imported from (from another project). Every substantive statement should carry at least one reference URL — ideally from an independent, authoritative source. The own website as reference for self-description is acceptable; for facts like revenue, headcount, founding date, external sources should dominate.
Maintaining Wikidata in steady state
Wikidata items are never "finished". A quarterly review across four dimensions: (1) add new publications as references, (2) reflect personnel and organizational changes, (3) update external identifiers (LinkedIn URL changes, new Crunchbase profiles), (4) drift control — identify and correct incorrect edits by third parties. Tool recommendation: a watchlist plus Wikimedia Toolforge utilities for structured reviews.
Conclusion: Wikidata is the lever beneath Wikipedia
For 80% of brands, Wikidata is the right first entity step — not Wikipedia. Lower hurdles, directly machine-readable, ingested directly by Google, Bing and LLMs. Brands that maintain and reference 20+ properties cleanly hold a knowledge-graph asset that pays off measurably across all three entity-SEO dimensions (identifiability, consistency, corroboration).