Does ChatGPT Even See Your Customer Reviews?

Your customers see the stars. But does ChatGPT see them too?
In many shops we analyzed with our AI Trust Audit, the honest answer is: no. The reviews are there, they are good – and yet they remain invisible to AI systems. This is not a marginal problem, as independent trust signals are among the factors that most strongly influence AI responses.
Why reviews are so important for AI
A peer-reviewed study published at ACM Hypertext 2026 (researchers from USC, Northeastern University, NYU, and Nvidia, among others) examined which content generative search systems prefer to cite. A key finding: external trust and community signals perform significantly better than pure brand communication – by +18.2 percentage points in the experiment.
Put simply: what others say about you counts more for AI than what you say about yourself. Customer reviews are exactly such a signal – provided the AI can read them.
What our audits show
We evaluated over 4,400 shop audits from the last 60 days conducted with the AI Trust Audit. This is not a scientific study, but an evaluation of our own testing data – yet it shows a clear pattern:
- 94% of shops display visible reviews to visitors.
- Only 5.8% additionally provide a valid, server-side review markup (Schema.org
AggregateRatingorReview) in the HTML.
For the vast majority, this means: reviews are there for humans, but not for machines – at least not in a form that is clearly readable.
Why AI overlooks your reviews
Many review widgets only load stars and texts after the fact via JavaScript. Everything looks perfect in the browser. However, AI crawlers like GPTBot or PerplexityBot usually only read the HTML that the server delivers directly. Often, at the location of the widget, there is: nothing.
Even if review text is present in the HTML, the clear structure is often missing. An AI then has to guess whether "4.8" is a grade, a price, or a version number. Structured data takes this guesswork away.
How to check your own shop
The AI Trust Audit checks exactly this point. It accesses your page just like an AI crawler does – without JavaScript – and distinguishes between three cases:
- Visible and structured: Reviews are in the HTML and marked up as Schema.org data. Ideal.
- Visible, but not structured: Humans see the stars, but AI can hardly assign them reliably. This is the most common gap.
- Loaded only via JavaScript: For AI crawlers, the area is empty.
In the report, you can see which case applies to you and receive a copyable code example.
The solution: provide reviews in machine-readable format
The lever is manageable: Output your overall rating server-side as JSON-LD in the <head>. A minimal example:
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Organization",
"name": "Your Shop Name",
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.8",
"reviewCount": "1250",
"bestRating": "5"
}
}
</script>
Important: The values must be real and up-to-date and match the visible reviews. Invented or outdated numbers do more harm than good.
As a Trusted Shops member, you don't have to maintain this code yourself: the AI Visibility Snippet automatically delivers your verified reviews in a machine-readable format. In the AI Trust Audit, members see the snippet already populated with their own data.
Staying realistic
Machine-readable reviews are no guarantee that ChatGPT will recommend your shop. Which sources an AI selects depends on many factors. But they remove an avoidable hurdle: your best trust signals become readable in the first place.
Check now for free if AI systems can see your reviews – with the AI Trust Audit.
Tip for existing users: Did you perform your audit before the update? We have refined the detection of review widgets that are only reloaded via JavaScript and the classification of the visibility gap. A fresh run will show you the current status of your review visibility.