What ClaimReview Schema Actually Does
ClaimReview is a structured data type created by Schema.org and adopted jointly by Google, Bing, and a coalition of fact-checking organisations. Its job is simple: it tells machines that a specific piece of content has been reviewed, what claim was made, who reviewed it, and what the verdict was.
Unlike most schema types, ClaimReview is not about describing your business or your products. It is about asserting credibility at the content level. You are essentially attaching a machine-readable fact-check label to a page, saying: "This claim was reviewed, here is the verdict, and here is who reviewed it."
That distinction matters enormously when you consider how AI search engines like ChatGPT, Perplexity, and Gemini decide what to cite. These systems are not just looking for relevant content. They are looking for content they can trust. ClaimReview is one of the clearest signals you can give them that your content has been through a verification process.
It is worth being precise about what ClaimReview is not. It is not a general-purpose credibility badge you can slap on any page. Misuse gets flagged quickly by Google's Rich Results Test and will actively harm your standing with AI crawlers. Used correctly, though, it is one of the more powerful trust signals available in the schema vocabulary.
The Core Properties You Need to Implement
Getting ClaimReview right means understanding its required and recommended properties. Here is what each one does and why it matters.
claimReviewed
This is a short summary of the claim being assessed. Keep it under 75 characters if you can. AI systems often pull this text directly when summarising fact-check content, so precision here pays off. Write it as the claim itself, not as a description of the claim. "The UK government banned petrol cars from 2030" rather than "A claim about petrol car bans."
reviewRating
This uses a Rating object with a ratingValue, a bestRating, a worstRating, and an alternateName. The alternateName is what most AI systems and search engines actually display, so use plain English verdicts: "True", "False", "Mostly True", "Misleading", "Unverified". The numeric scale typically runs from 1 (false) to 5 (true), but you can define your own scale as long as you set bestRating and worstRating explicitly.
author
This references the organisation or person who performed the review. Use an Organization or Person object here with a name and a url. AI systems use this to evaluate the credibility of the reviewer, not just the review itself. A named organisation with a real URL carries significantly more weight than an anonymous or vague attribution.
itemReviewed
This describes the claim's origin: who made it, when, and where. At minimum, include a CreativeWork or Claim object with an author and a datePublished. This is where many implementations fall short. Leaving it vague undermines the entire signal.
url
The canonical URL of the fact-check article itself. This must be the page where the review lives, not a homepage or a category page.
A Working JSON-LD Example
Here is a clean, production-ready ClaimReview implementation you can adapt:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "ClaimReview",
"url": "https://example.com/fact-check/petrol-car-ban-2030",
"claimReviewed": "The UK government banned new petrol car sales from 2030.",
"author": {
"@type": "Organization",
"name": "Example Fact Check",
"url": "https://example.com"
},
"reviewRating": {
"@type": "Rating",
"ratingValue": 4,
"bestRating": 5,
"worstRating": 1,
"alternateName": "Mostly True"
},
"itemReviewed": {
"@type": "Claim",
"author": {
"@type": "Organization",
"name": "The Daily Tribune",
"url": "https://example-tribune.com"
},
"datePublished": "2024-03-15",
"appearance": {
"@type": "CreativeWork",
"url": "https://example-tribune.com/articles/petrol-ban-2030"
}
}
}
</script>Notice a few things. The itemReviewed block traces back to the original source of the claim with a real URL. The reviewRating uses both a numeric value and a plain-English label. The author block names the reviewing organisation with its own URL. Each of these details adds a layer of verifiability that AI crawlers can follow.
Why AI Search Engines Respond to ClaimReview
AI language models are trained on enormous amounts of web content, but at inference time, when they are generating a cited answer to a user query, they are selecting from indexed, crawled content based on signals of reliability. ClaimReview is one of the few schema types that explicitly encodes a credibility verdict in machine-readable form.
Perplexity, for example, heavily weights content from sources it considers authoritative and verifiable. When your page carries ClaimReview markup with a named reviewing organisation, a clearly sourced claim, and a structured verdict, you are giving Perplexity's indexing system exactly the kind of unambiguous metadata it wants.
ChatGPT's browsing mode and its underlying citation logic similarly favour content that appears in verified, reputable sources. ClaimReview is a signal used by Google's fact-check explorer, which feeds into the broader ecosystem of trusted sources that AI systems cross-reference.
It is not magic. A poorly written fact-check article with technically valid ClaimReview markup will not suddenly become an AI citation favourite. The schema amplifies the credibility of genuinely good content. It does not manufacture credibility from thin air.
This is a point worth sitting with. If your content is thin, vague, or derivative, no amount of schema markup will get you cited by an AI search engine with any consistency. What ClaimReview does is make sure that when your content is credible, machines can actually recognise that credibility and act on it.
Who Should Be Using ClaimReview
The obvious use case is dedicated fact-checking publishers: Full Fact, Reuters Fact Check, PolitiFact, and their equivalents. But the schema has genuine applications beyond that narrow category.
News publishers and investigative journalism outlets
If your newsroom publishes content that corrects misinformation or verifies public claims, ClaimReview is a natural fit. Even a single well-implemented fact-check article can meaningfully lift the perceived trustworthiness of your entire domain in AI indexing systems.
Health and medical information sites
Health misinformation is one of the most actively monitored areas by AI systems. A medical publisher that uses ClaimReview to label reviewed health claims signals a level of editorial rigour that AI systems are specifically trained to value.
Financial information and consumer advice brands
Sites that correct common financial misconceptions or debunk misleading product claims can use ClaimReview to distinguish themselves from the sea of unverified content in their niche. For an e-commerce brand publishing educational content, this can be a meaningful trust differentiator.
What ClaimReview is NOT suitable for
Do not use it on product pages. Do not use it on general blog posts that happen to make a factual claim. Do not use it to mark up your own brand claims as "verified." AI crawlers and Google's quality systems will identify misuse, and the consequence is not just losing the rich result. It can actively flag your domain as unreliable.
Common Implementation Mistakes to Avoid
Most ClaimReview implementations fail in predictable ways. Here are the ones to watch for.
Using vague or missing itemReviewed data
Leaving out the itemReviewed block or populating it with only a name string and no URL is the most common error. AI systems follow the chain of attribution. If you cannot point to where the original claim came from, the schema loses most of its value.
Applying ClaimReview to your own brand content
You cannot fact-check yourself. The credibility signal requires an arms-length review. The reviewing author and the source of the claim should be distinct entities. If they are the same, the schema looks self-serving and will be treated accordingly.
Inconsistent rating scales
If you use a 1-to-5 scale on some pages and a 1-to-10 scale on others, or if your alternateName labels do not correspond logically to the numeric values, crawlers will flag inconsistencies. Pick a scale and stick to it across your entire site.
Ignoring the url property
The url property must point to the specific fact-check article, not your homepage, not a category archive, not a tag page. This is a canonical pointer. Get it right.
Validating Your ClaimReview Markup
Once you have implemented ClaimReview, validate it before it goes live. Google's Rich Results Test at search.google.com/test/rich-results accepts both a live URL and a code snippet. It will tell you immediately if any required properties are missing or if there are type errors.
Schema.org's own validator at validator.schema.org is also worth running alongside. The two tools catch slightly different issues, and running both takes under two minutes.
After validation, check Google Search Console under the Enhancements section. Once your pages are indexed, you will see ClaimReview appearing there if it is implemented correctly. Errors and warnings will surface within a few days of indexing.
At FlinnSchema's free AI visibility audit, we frequently find ClaimReview implemented with technically valid markup that still fails to build trust because the surrounding content is thin or the attribution chain is broken. Technical validity is the floor, not the ceiling.
Pairing ClaimReview with Other Trust Signals
ClaimReview works best as part of a broader structured data strategy. On its own, it signals that a specific piece of content has been reviewed. Combined with other schema types, it contributes to a domain-level trust profile that AI systems can build a picture of over time.
Consider pairing it with Organization schema on your site's main pages, which establishes who you are and what you do. Add Person schema for named authors and reviewers. If you publish news content, NewsArticle schema on your editorial pages rounds out the picture. NewsArticle schema has its own specific implementation requirements for AI citation, and getting both right creates a compounding trust signal that is difficult for competitors to replicate quickly.
The underlying principle is that AI search engines build a mental model of your domain based on every structured signal they can find. ClaimReview tells them your editorial process is rigorous. Organization schema tells them who you are. Author Person schema tells them real humans with real credentials are behind the content. Together, these signals make your site significantly more likely to be cited when a relevant query comes in.
If you are working on a WordPress site and want to go deeper on the technical implementation side, adding JSON-LD to WordPress without a plugin gives you full control over your ClaimReview output without the constraints of a plugin's schema templates.
For brands serious about AI visibility, getting this layer of structured data right is increasingly the difference between being cited and being invisible. The approach FlinnSchema takes is to treat schema as a system, not a checklist, and ClaimReview is one of the more sophisticated pieces of that system when it is used in the right context.
Frequently Asked Questions
Does ClaimReview schema directly affect my Google search rankings?
ClaimReview does not directly improve organic rankings for most content types. What it does is make you eligible for fact-check rich results in Google Search, and it contributes to the trust signals that AI search engines use when deciding what content to cite. For publishers whose core business involves fact-checking or claims verification, it is one of the more valuable schema types available.
Can I use ClaimReview if I am not an official fact-checking organisation?
You do not need to be an accredited fact-checker to use ClaimReview, but you do need to be publishing genuine, editorially rigorous reviews of specific claims. If your "fact-check" article is really a marketing piece or a thin opinion post dressed up as a review, both Google and AI systems will identify that over time. The schema needs to reflect real editorial work.
How long does it take for ClaimReview to influence AI search citations?
There is no fixed timeline. Once your pages are indexed and the markup is validated, AI crawlers will begin incorporating the signals into their trust models. For established domains with strong backlink profiles, meaningful changes in citation frequency can appear within a few weeks of correct implementation. For newer domains, it tends to take longer because the schema signals are being evaluated alongside other weaker trust indicators.
What is the difference between ClaimReview and other credibility schema types like Article or NewsArticle?
Article and NewsArticle schema describe the format and metadata of a piece of content: who wrote it, when, what it is about. ClaimReview goes a step further by encoding a specific verdict about a specific claim. It is not about describing content, it is about asserting a reviewed conclusion. The two work well together, but they serve different functions in the structured data ecosystem.