I blocked GPTBot. Does that mean ChatGPT can't cite me?

No, and this is the most common confusion. GPTBot collects text for training future models. The crawler that fetches a page so ChatGPT can answer with it is OAI-SearchBot, and ChatGPT-User when someone's question triggers a live fetch. Blocking GPTBot while leaving those open is a coherent position: don't train on me, do cite me.

Is llms.txt worth adding?

It costs an hour and no assistant currently requires it. Treat it as cheap insurance rather than a fix — if someone tells you it will get you cited, they are overselling a convention that is still settling.

My page passed every check but I still don't appear in AI answers.

That is expected and the checks are not lying to you. Readability is the precondition; being chosen is a separate contest decided by authority and relevance, the same things that decide search rankings. Passing here means you are eligible, not selected.

Do you store the URLs people check?

No. The check runs when you submit and the result is rendered straight back to you. Nothing is written down, and result pages are excluded from search indexes.

Why does it only check one page?

Because a full site crawl is a different, slower job, and most of what matters here — the robots.txt rules above all — is decided site-wide anyway. Paste your homepage for the access questions, or a product page if you want the structured-data check to apply.

AI visibility checker: can ChatGPT and Perplexity actually read your site?

Markazi Panel17 September 2026Free, no signup, nothing sent anywhere

People increasingly ask an assistant rather than a search engine, and the assistant answers from pages it was able to fetch. This checks whether yours is one of them.

The short answer

Three things decide whether an AI assistant can use your site to answer a question. Whether your robots.txt lets its crawler in. Whether there is readable text in the HTML, as opposed to content a browser assembles with JavaScript. And whether your pages carry structured data saying what is being sold and for how much.

The first is the one that is usually wrong by accident. A robots.txt rule copied from a template, or added by a plugin, can block the exact crawlers that fetch pages to cite them — and nothing about your site looks broken afterwards. You simply stop being an answer.

It is worth separating two kinds of crawler, because most advice does not. Some collect text to train future models; blocking those is a normal choice and changes nothing about whether you are cited today. Others fetch a page so an assistant can answer a question with it, now, with a link back. Only the second kind matters here, and this check only fails you for the second kind.

What each check means

AI crawler access. Read from your /robots.txt. A Disallow: / against OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-SearchBot or Claude-User means an assistant cannot fetch your pages to answer with them.
llms.txt. A short plain-text file describing what the site is and which pages matter. An emerging convention rather than a ranking factor — cheap to add, and no guarantee of anything.
Product structured data. JSON-LD with name, price and availability. It is how a machine quotes your price instead of guessing at it. Only reported on pages that look like product pages.
The page as a crawler receives it. We fetch without running any JavaScript, because that is what almost every AI crawler does, and show you the text that came back. If that is nearly empty while your site looks full in a browser, the difference is the problem.

Why the JavaScript part catches people out

A modern storefront often renders its product grid, prices and descriptions in the browser. Open it yourself and the page is full. Fetch it the way a crawler does and you can get a near-empty shell with a loading state.

Google renders JavaScript, eventually and imperfectly. Most AI crawlers do not render it at all. So a site can rank acceptably on Google and be almost invisible to an assistant, and nothing in your analytics will tell you, because a crawler that finds nothing worth citing does not send a visitor to be counted.

What this does not tell you

This checks whether you are readable. It does not check whether you are actually being cited, which depends on authority, relevance and competition, and varies between assistants and between one question and the next. Being readable is a precondition, not a promise. A site can pass every check here and still not be mentioned.

It also checks one page — the one you paste. Robots rules are site-wide, so that part holds everywhere; the structured data and the text check are about that URL alone.

Questions people ask

I blocked GPTBot. Does that mean ChatGPT can't cite me?: No, and this is the most common confusion. GPTBot collects text for training future models. The crawler that fetches a page so ChatGPT can answer with it is OAI-SearchBot, and ChatGPT-User when someone's question triggers a live fetch. Blocking GPTBot while leaving those open is a coherent position: don't train on me, do cite me.
Is llms.txt worth adding?: It costs an hour and no assistant currently requires it. Treat it as cheap insurance rather than a fix — if someone tells you it will get you cited, they are overselling a convention that is still settling.
My page passed every check but I still don't appear in AI answers.: That is expected and the checks are not lying to you. Readability is the precondition; being chosen is a separate contest decided by authority and relevance, the same things that decide search rankings. Passing here means you are eligible, not selected.
Do you store the URLs people check?: No. The check runs when you submit and the result is rendered straight back to you. Nothing is written down, and result pages are excluded from search indexes.
Why does it only check one page?: Because a full site crawl is a different, slower job, and most of what matters here — the robots.txt rules above all — is decided site-wide anyway. Paste your homepage for the access questions, or a product page if you want the structured-data check to apply.

This asks whether machines can read your store. The COD profit calculator asks what an order leaves you once it does sell, and what Google Analytics does not tell a store owner covers the traffic you are already getting and cannot see.

Markazi Panel watches this continuously instead of once — whole-site crawls, and which AI assistants and crawlers actually reached you — see a dashboard with a month of data in it.

The other tools

COD profit calculator: what one cash-on-delivery order actually leaves youWhat one COD order leaves you after goods, freight, deductions and returns — in rupees, with your numbers.
Break-even ad spend for a COD store: the most you can pay per orderThe most you can pay to acquire one order, once you account for the parcels that come back.
Courier remittance checker: did the settlement pay what it owed?Whether the money the courier sent matches the parcels they delivered — for one settlement cycle.