Free tool · rendering
Your page looks fine. Does it arrive that way?
This fetches your page and does not run a single line of your JavaScript, which is what an answer-time crawler does. What comes back is what a model has to work with.
01Why it matters
The crawler that decides whether you get cited is not running a browser.
An answer-time agent fetches a page while an assistant is composing a reply, under a latency budget measured in seconds. Executing a JavaScript bundle and waiting for the data it requests costs time that no engine has committed to spending.
So the practical question is not whether a page can be rendered. It is what the page contains before anything is rendered, which is the only version some crawlers will ever hold. A site that assembles its content in the browser has a body of work that, from the outside, does not exist.
Search made this feel solved. Google renders in a second pass, so client-rendered sites still index, and a decade of teams learned that the problem had been handled. The answer-time agents did not inherit that behaviour, and the sites that trusted it are the ones now invisible in assistant answers while their rank reports look untouched.
02Questions
- Do AI crawlers run JavaScript?
- Assume not. Answer-time agents fetch pages under a latency budget while a reply is being composed, and executing a bundle costs time none of them are obliged to spend. Some engines render some of the time; none of them commit to it, so a site that needs rendering to be readable is gambling on behaviour nobody has published.
- My page looks fine in a browser. Why does this show almost nothing?
- Your browser ran the JavaScript that assembled the page. This check does not, so what it displays is the HTML the server actually sent. A large gap between the two is the entire finding.
- Is Next.js or React a problem?
- No. Both server-render by default in their current configurations, and a React page can arrive complete. The problem is a client-only setup, where the server sends a mount point and a bundle and the content is built afterwards in the browser.
- What word count is enough?
- There is no threshold that makes a page citable, and this check does not pretend otherwise. What matters is whether the content that answers the question is in the HTML. A page that arrives with a hundred words of navigation and no substance has failed, whatever the number says.
- Does Google render JavaScript?
- Google does render, in a second pass after the initial crawl. That has led plenty of teams to conclude rendering is a solved problem, which was true for search and is not true for the answer-time agents that decide whether an assistant can cite you.
- How do I fix a page that fails this?
- Server-render the content, or pre-render it at build time. Both put the substance in the initial HTML response, which is the only thing this check, or a crawler, ever reads.
Next
Arriving is not the same as being quotable.
A page can server-render perfectly and still have no passage a model can lift. The answer readiness check reads your sections one by one and says which of them survive being quoted on their own.