Your structured data names your areas of expertise using AI-readable topic links
Do you declare your three or more core topics as Wikipedia or Wikidata links, not vague text?
What this signal tests
We check whether your Organization or Person JSON-LD includes a knowsAbout field populated with at least three topics, each given as a Wikipedia or Wikidata URL (or as a typed Thing object with an @id), not as a free-text string. This pins your expertise to specific concepts AI systems can match against user queries.
Why it matters for your visibility in AI
When a user asks an AI assistant "who is an expert on lithium battery recycling?", the assistant searches its knowledge for entities whose declared topics overlap with the query. If your knowsAbout says the string 'batteries', that is too vague to match precisely. If it says the Wikidata URL for lithium-ion battery recycling, the match is exact. This matters most for specialist firms and authors. A law firm that lists knowsAbout as Wikidata URLs for 'intellectual property law', 'patent litigation', and 'trade secret protection' will surface in AI answers about those topics. A firm whose homepage merely says 'we do IP' in marketing copy will not. The Wikidata URL acts as a precise category tag that AI systems can route queries against.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| >=3 URL/Thing-typed values. |
How we test it
We parse the Organization or Person JSON-LD on your homepage and inspect the knowsAbout property. We count values that are either URLs matching wikidata.org/entity/Q (or /wiki/Q) or wikipedia.org/wiki/ patterns, or objects with an explicit @type and @id (such as { '@type': 'Thing', '@id': 'https://...' }). Plain text strings do not count. If fewer than three URL or Thing-typed values are present, the signal fails.
Show technical detection method
Parse knowsAbout; count URLs to wikidata.org/entity/Q or wikipedia.org, or objects with @type+@id.
If your site fails: how to fix it
- List the three to ten topics that best describe your organization or your expertise as a Person.
- For each topic, search Wikidata for the concept and copy its Q-number URL (for example, Q188 for 'machine learning'). If no Wikidata entry exists, use the Wikipedia URL instead.
- Add the URLs to your JSON-LD as an array under knowsAbout: [ "https://www.wikidata.org/wiki/Q188", "https://www.wikidata.org/wiki/Q11660", ... ].
- Prefer Wikidata over Wikipedia URLs; Wikidata Q-numbers are language-neutral and more durable.
- Avoid free-text topic strings; they are unparseable by AI and contribute nothing to the signal.
- Validate with Schema.org Validator and re-run the scan.
Quick facts
| Maturity | ESTABLISHED |
|---|---|
| Weight | medium |
| Category | Entity |
Primary sources
Related signals
Frequently asked questions
How many topics should I list?
Three is the minimum to pass. Five to ten covers most specialist firms well. Beyond ten you risk diluting your expertise signal: an entity that claims to know about 200 topics looks generalist and is less likely to surface for any specific query.
What if my niche is too specific for Wikidata?
Then use the closest parent concept and consider creating a Wikidata entry for the niche. Most professional specializations are already in Wikidata; if yours is genuinely novel, creating the entry is a one-time investment that benefits everyone in your field.
Should a Person knowsAbout differ from an Organization knowsAbout?
Yes. A Person's knowsAbout should reflect their individual expertise (often narrower than the org's); the Organization's should reflect the collective domain. A research lab might know about ten topics; its lead researcher might know about three of those deeply.
Will Google or ChatGPT actually read this?
Yes, Google's Knowledge Graph reads knowsAbout when it has Wikidata-linked values. LLM-based search tools (ChatGPT Search, Perplexity, Claude with web access) also pick it up because the linked Wikidata items are part of the same knowledge graph they query at retrieval time.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.