If you publish fact-checks, do they use the standard ClaimReview structured data?
Checks that fact-check content is marked up in the format AI grounding systems and search fact panels recognise.
What this signal tests
If your site publishes fact-checks, debunkings, or claim-verification articles, this signal looks for JSON-LD with @type ClaimReview. We confirm the three required properties: claimReviewed (the claim being evaluated as a text string), itemReviewed (the CreativeWork or post that contained the claim), and reviewRating (a Rating with a numeric ratingValue indicating how true the claim is). If any of these is missing or mistyped, the signal fails.
Why it matters for your visibility in AI
ClaimReview is the standard machine-readable encoding for fact-check content. Google's fact-check panel, Bing's fact-check snippets, and LLM grounding systems used by ChatGPT and Claude all read it directly. A fact-check that lacks ClaimReview is essentially invisible to these surfaces, even if it ranks well organically. The consequence is significant for newsrooms and fact-check organisations. AI systems are increasingly conservative about repeating contested claims, and they look for verified fact-checks to determine which claims are debunked. If your debunking is not in ClaimReview format, the AI may quote the original claim without your correction, propagating misinformation. For organisations whose mission is to counter misinformation, missing ClaimReview is a mission-level failure.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| All three required properties present and well-typed. |
How we test it
We parse your page's JSON-LD blocks and look for any object with @type set to ClaimReview. For each match, we confirm the three required properties are present and well-typed: claimReviewed must be a string, itemReviewed must be a CreativeWork or one of its subtypes, and reviewRating must be a Rating with a numeric ratingValue. We do not enforce a specific rating scale, but Schema.org recommends a 1–5 scale where 1 is false and 5 is true.
Show technical detection method
JSON-LD @type ClaimReview with required properties.
If your site fails: how to fix it
- Confirm whether your site actually publishes fact-check content. ClaimReview is specifically for evaluating claims as true or false; it does not apply to general news or analysis. If you do not publish fact-checks, this signal is not relevant and you can mark it as not-applicable.
- For each fact-check article, add a JSON-LD block with @type: ClaimReview. Populate claimReviewed with the exact claim being evaluated, itemReviewed with details of the source post or article, and reviewRating with the numeric verdict.
- Use Schema.org's published example as a template. Most fact-check CMS platforms (Datawrapper, ClaimBuster, Duke Reporters' Lab tools) emit compliant ClaimReview automatically; check whether your platform offers this and turn it on.
- Validate sample pages with Google's Rich Results Test, which has a dedicated ClaimReview validator. Fix any reported errors before re-running the AI Ready Test scan.
Quick facts
| Maturity | ESTABLISHED |
|---|---|
| Weight | low |
| Category | Trust & Provenance |
Primary sources
Related signals
Frequently asked questions
Will I need IT help to fix this?
Yes, for the template work. ClaimReview has specific required fields that must come from editorial metadata, so a developer needs to wire those fields from your CMS into the JSON-LD output. After that, fact-checkers fill in the fields as part of normal writing.
What if I do not publish fact-checks?
Then this signal does not apply to your site, and you can treat it as not-applicable in your assessment. ClaimReview is specifically for evaluating individual claims as true or false. Most sites legitimately have zero ClaimReview content.
Does this affect SEO or only AI visibility?
Both. Google's fact-check rich result feature is built on ClaimReview, and a properly marked-up fact-check can appear in dedicated fact-check panels on the search results page. AI grounding systems read the same signal.
How long until the change takes effect?
Once the markup is in place, Google can pick it up on the next recrawl (hours to days). AI systems vary, but most refresh structured data within a week. There is no DNS or other propagation step.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.