Free tool

Can AI read the reviews on your website?

Are your reviews in the raw HTML? Can ChatGPT, Claude, Perplexity and Google read the page? Check any page on your site for free, no sign-up.

What the check looks at

The checker fetches a single page the way a crawler would, without running any JavaScript. It also reads the site's robots.txt. Then it checks four things:

Are the reviews in the raw HTML? If you paste a review excerpt, the tool looks for that exact sentence in the page text. Otherwise it looks for clues: the words “reviews” or “avis”, stars, ratings out of 5, review markup.
Which widget shows the reviews? Elfsight, Trustindex, EmbedSocial, SociableKit, Taggbox, Shapo, Trustpilot TrustBox, a Google Maps iframe or a Revys widget: the tool tells apart widgets that inject reviews with JavaScript or inside an iframe from those that write them into the HTML.
Are crawlers allowed to read the page? robots.txt is parsed according to RFC 9309 (groups per crawler, Allow and Disallow, longest match, * and $ wildcards) for nine crawlers: OAI-SearchBot, ChatGPT-User, GPTBot, Claude-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, Googlebot and Bingbot. The meta robots tag and the X-Robots-Tag header are read too (noindex).
What does the structured data say? JSON-LD blocks are listed by schema.org type, with a warning if a rating or reviews sit on your own business.

You get a one-line verdict (readable, partly readable, not readable) followed by concrete fixes, most important first.

Why raw HTML matters

Googlebot can run JavaScript, on a delay. AI crawlers, in practice, read what the server sends back: a Vercel analysis of hundreds of millions of requests found that none of the major AI crawlers it observed (OpenAI, Anthropic, Meta, ByteDance, Perplexity) render JavaScript. A widget that loads your reviews with a script therefore leaves a page with no reviews in their eyes. An iframe has a different problem: its content belongs to the iframe's address (the widget vendor's site), not to your page.

The fix is to write the reviews into the HTML on the server. That's what Revys does with its WordPress plugin, its server-side include and its API: the reviews are part of your page, like any other paragraph.

AI crawlers and robots.txt

Each vendor runs several crawlers with different jobs:

OpenAI: OAI-SearchBot for ChatGPT search (sites that block it aren't shown in search answers), ChatGPT-User for visits a user asks for, GPTBot for training (OpenAI docs).
Anthropic: Claude-SearchBot for search, Claude-User for on-demand visits, ClaudeBot for training (Anthropic docs).
Perplexity: PerplexityBot for the search index, Perplexity-User for on-demand visits (Perplexity docs).
Google: Googlebot for Search and AI Overviews; Google-Extended controls whether content is used to train Gemini, with no effect on Google Search ranking (Google docs). Microsoft: Bingbot feeds Bing and Copilot.

Blocking training crawlers is a legitimate choice that doesn't stop you being cited. Blocking search crawlers, on the other hand, takes your page out of the answers. Vendors also note that their on-demand fetchers (ChatGPT-User, Perplexity-User) may not apply robots.txt, since a user triggered the visit.

Structured data and reviews

Since 2019, Google no longer shows stars for “self-serving” reviews, meaning reviews a business publishes about itself through LocalBusiness or Organization markup (Google announcement). That markup isn't a serious offence, but it won't earn stars in results. What matters for AI is the review text in the page, attributed to your business.

Method and limits

The checker identifies itself as RevysChecker/1.0. It doesn't pretend to be GPTBot or Googlebot: a firewall or CDN (Cloudflare's AI crawler blocking, for example) may treat real crawlers differently.
It analyses one page at a time, up to 2 MB, within 8 seconds, following at most 3 redirects.
Only public pages can be checked: local, private or reserved addresses, non-standard ports and anything other than http(s) are refused, including after a redirect.
Nothing is stored. The limit is 10 checks per day per connection.
Being readable doesn't guarantee being cited: it's the starting point. The reviews and AI search guide covers the rest.

FAQ

Why aren't my reviews readable when I can see them on my site?

Your browser runs JavaScript and renders iframes; most AI crawlers don't. They read the HTML your server returns. If a widget adds the reviews after the page loads, or shows them in an iframe, you can see them but the crawlers can't.

Does blocking GPTBot keep me out of ChatGPT?

No. According to OpenAI, GPTBot is used for model training; OAI-SearchBot is what decides whether a site can appear in ChatGPT search answers. You can block GPTBot and allow OAI-SearchBot. The same logic applies at Anthropic (ClaudeBot vs Claude-SearchBot) and Google (Google-Extended has no effect on Google Search).

Does the check store my URL or my page content?

No. The page and robots.txt are fetched, analysed and then discarded. Nothing is stored in a database. A temporary in-memory counter limits the tool to 10 checks per day per connection.

Why paste a review excerpt?

It's the most reliable check: the tool looks for that exact sentence in the raw HTML. Without an excerpt it can only look for clues (the word “reviews”, stars, known widgets), which is less precise.

Can I add star ratings (AggregateRating) to my own business?

You can, but Google won't show them: since 2019 it ignores reviews a business publishes about itself in LocalBusiness or Organization markup. The checker flags it so you don't keep useless markup.

Is the result guaranteed to match what ChatGPT or Google sees?

No. The checker identifies itself honestly (RevysChecker) and doesn't pretend to be another crawler. A firewall or CDN may treat real AI crawlers differently, and each AI then decides on its own what to cite. The check shows whether your reviews can be read, not whether they will be cited.