Can an AI assistant actually read your store?
A store can rank perfectly well in Google and still be invisible to ChatGPT, Perplexity and Gemini. This checks the four things that decide it: whether your robots.txt lets their crawlers in, whether there is a sitemap, whether the page survives without JavaScript, and whether your structured data says what you sell.
Eight requests, no signup, no email. Results cached for a day.
Why a store that ranks in Google can still be invisible to ChatGPT
They are different indexes, not one internet
Google’s AI surfaces read Google’s index. ChatGPT’s browsing reads Bing’s. Perplexity runs its own crawler. A store indexed in Google and absent from Bing is invisible to ChatGPT with browsing switched on, and nothing in Google Search Console will ever tell you. Verifying the site in Bing Webmaster Tools and submitting the sitemap there takes about ten minutes and is the highest-value thing most stores have not done.
AI crawlers use their own tokens, and blocking them is easy to do by accident
GPTBot, OAI-SearchBot, ClaudeBot, Claude-SearchBot, PerplexityBot and Google-Extended each decide access from your robots.txt independently of Googlebot. Plenty of stores block them — sometimes on purpose, more often because a plugin or a well-meaning agency added a block during a wave of “stop AI scraping” advice. For a public storefront that trade rarely makes sense: the pages are already public, and the cost is that an assistant answering a buying question in your category recommends someone else.
If the page needs JavaScript, they mostly do not see it
AI crawlers frequently fetch the raw HTML and read that. A storefront that renders client-side looks complete to you and nearly empty to them. The check here is deliberately crude: strip every script and tag, count the text left. On a homepage, a few hundred characters means the machines are reading a shell — and no amount of content strategy fixes a page they cannot see.
Structured data helps the index-based systems, not the browsing ones
Controlled testing in 2026 found that live-browsing assistants read the visible HTML and treat JSON-LD as plain text — models happily extracted values from deliberately invalid schema. Index-based systems such as Google’s AI surfaces and Bing parse it properly. The rule that follows is simple: publish the structured data, and make sure every fact in it also exists in the visible text, because the systems that matter most read the text.
Questions about AI crawlers and storefronts
Which crawlers do AI assistants use, and should I allow them?
OpenAI uses GPTBot for training and OAI-SearchBot for ChatGPT's search results, plus ChatGPT-User for pages fetched during a conversation. Anthropic uses ClaudeBot, Claude-SearchBot and Claude-User. Perplexity uses PerplexityBot and Perplexity-User. Google uses Google-Extended for Gemini, which is a separate token from ordinary Search indexing. For a public storefront the answer is almost always to allow all of them: your product pages are already public, and blocking them means an assistant answering a question in your category describes a competitor instead of you.
Do AI crawlers run JavaScript?
Often not, and you cannot rely on it. A storefront that renders its content client-side may look complete in a browser and read as an almost empty page to the crawler. The test in this scan is deliberately crude for that reason: strip every script and tag from the raw HTML and count what text is left. Under a few hundred characters on a homepage means the machines are seeing a shell.
Why does structured data matter for AI search when the model can read the page?
Less than the SEO advice suggests, and more than nothing. Controlled testing in 2026 showed live-browsing assistants read the visible HTML and treat JSON-LD as plain text — one experiment had models extracting data from deliberately invalid schema. Index-based systems like Google's AI surfaces and Bing do parse it properly. So structured data is a second copy for the systems that read it, never the only copy: any fact you want quoted has to be in the visible text too.
My store ranks well in Google. Does that mean AI assistants can find it?
Not necessarily. Google's AI surfaces draw on Google's index, but ChatGPT's browsing reads Bing's, and Perplexity runs its own crawler. A store indexed in Google and absent from Bing is invisible to ChatGPT with browsing on, which is a common and entirely silent failure. Verifying the site in Bing Webmaster Tools and submitting the sitemap there takes minutes and is the single highest-value thing most stores are not doing.
What does this scan not check?
Almost everything a proper audit would. It does not grade your policy pages, accessibility, page speed, checkout, or the accuracy of your product data — it makes eight requests and answers one question: can a machine read this store. For product data specifically, the Merchant Center check on this site goes deeper.
What this scan is, and what it is not
Eight requests: the homepage, robots.txt, up to two sitemap candidates and one product page. It identifies itself as MagistryToolsBot with a contact address, follows redirects with a cap, and refuses anything that is not a public host.
It is not an audit. It does not grade your policy pages, accessibility, page speed, checkout or product-data accuracy. Magistry’s own storefront crawler does that over fifty pages for connected stores; this answers one narrower question that no other free tool answers at all. For product data specifically, the Merchant Center check goes deeper.
The robots.txt parse is conservative on purpose. It only reports a crawler as blocked when a Disallow: / clearly applies to it. Telling a merchant they are blocking ChatGPT when they are not would be worse than saying nothing.
Last reviewed 3 August 2026. Results cached 24 hours per domain; no IP addresses stored.