Paste a page and see what a robot that does not run scripts finds in it: how many words, which headings and links, and the first 500 characters it would read.
One public page, raw HTML only, no scripts run. Nothing is stored.
A browser downloads your page, runs its scripts and paints the result. Many AI robots stop after the first step: they read the HTML the server sent and move on. If your site builds its text with JavaScript in the browser, as many React and Vue sites do without server rendering, those robots receive a nearly blank page with a title and an empty container.
Nothing looks wrong to you, because you always see the finished page. That is why this checker shows the raw text itself rather than a score: if the first 500 characters in the result are your menu and cookie notice, that is what the assistant gets too. Our study of 731 websites lists text that only appears after scripts run among the five reasons assistants find a site hard to read.
This checks one page. The free OperStack check reads your home page and up to nine more the way an assistant does, scores them out of 100, and asks ChatGPT the questions your buyers ask.
By Vercel's analysis of crawler traffic (December 2024), OpenAI's robots (OAI-SearchBot, ChatGPT-User, GPTBot) and Anthropic's ClaudeBot fetch JavaScript files but do not execute them, while Gemini, through Googlebot, and Applebot do render pages (Vercel: The rise of the AI crawler). Robots change, so treat the raw HTML as the safe lowest common denominator.
Google does render pages, in a second step that can come later than the first crawl (Google: JavaScript SEO basics). ChatGPT and Claude are not Google. If your text exists only after scripts run, the assistants that read raw HTML get the empty shell.
Under 100 words a page is about one short paragraph, often just the menu and footer: too little for anyone to describe what you do. Between 100 and 250 the page can be read but is thin. These lines are a rule of thumb we chose, not a standard: a contact page can be short on purpose.
Make the server send the text. Frameworks call it server-side rendering or static generation: Next.js, Nuxt, Astro, SvelteKit and Remix all do it. If rebuilding is not an option, a prerendering service can serve robots a rendered copy. On WordPress, Wix, Squarespace and Shopify the text is usually in the HTML already; if it is not, a page builder or an app widget is drawing it with scripts.
It is what the raw HTML holds, with scripts, styles and hidden templates removed. A real robot may also skip menus and footers, may be served a different page by your firewall, or may be blocked in robots.txt. The AI crawler checker covers the last two.
No. We read the page once while this page loads and keep nothing: no account, no cookie, no database.
Built by Maksim Shchegolev, founder of OperStack. Free, no account, no cookie. The address you check is not stored.
OperStack