Loading…
The crawlers behind ChatGPT, Claude and Perplexity don’t run JavaScript. Vercel and MERJ confirmed it across more than 500 million GPTBot fetches. So a site built as a client-rendered app can rank in Google — whose crawler does render — and still be a blank page to every assistant.
You can’t see that in a browser. This fetches your page as each crawler, once, and shows you exactly what came back.
It tells you whether a crawler can reach your page, and what text it gets when it does. Both are mechanical, verifiable, and the same on every run.
It doesn’t tell you whether an assistant will recommend you. Nobody’s free tool can. Asking a model “do you know this product” returns a different answer on repeat runs — between 9% and 28% of decisions flip even at temperature zero, which is why serious measurement uses seven or eight repetitions per prompt. A single-shot check dressed up as a score is a coin flip with a progress bar.
What this page checks is the part that is genuinely deterministic: if a crawler gets a 403, or gets forty words of empty shell, nothing downstream can save you.
This is the mistake we see most. GPTBot collects training data. OAI-SearchBot fetches pages for the index behind ChatGPT’s search citations. Blocking the first keeps you out of training runs and changes nothing about search. Blocking the second removes you from ChatGPT’s answers entirely — and a lot of sites do it by accident, with one blanket rule.
Once a crawler can read your page, listing it on LetsLaunch gives it one more page describing what it does, in someone else’s words. Worth being straight about what that’s worth: consistent descriptions across independent pages correlate with being named by assistants, but no one — us included — has shown that any individual listing causes it.