Websites don't fail loudly.
They just stop being visited.
You built one. Maybe you paid an agency, maybe you built it yourself over a weekend, maybe it's the tenth version you've shipped in five years. Then the renewal notice arrives from your registrar and you have to decide whether the site earned its $15 domain fee. For a lot of websites, the honest answer is no.
The reason is rarely the site itself
Most sites launched in the last few years look fine, load fast, and say the right words. What changed is where people find things.
For twenty years the argument was about Google. Get indexed, rank on page one, get the click. A whole industry grew up teaching people how. If your site was good enough for Google, people found you.
The click is what changed. When someone asks ChatGPT, Perplexity, Claude, or a Google AI Overview a question now, they read the answer without visiting the pages that fed it. Your site is either quoted inside that answer, cited by name, linked in a source list, or it isn't there at all. There is no page-two consolation prize.
Most sites are not there
Nearly half of the ones we have scanned actively block AI crawlers, usually without their owners knowing: a CDN default, a bot-protection tick-box, a hosting provider's opinion. Cloudflare has blocked AI crawlers on every new site since September 2025, so a site migrated in the last year probably went live already invisible.
The mechanics are different from Google. AI crawlers usually do not run JavaScript. They arrive under bot names your firewall has never heard of. They respect a different set of signals. And existing tools do not report on any of this — Google Search Console tells you about Google, and says nothing about ChatGPT, Perplexity, Claude, or the crawlers that feed them.
What we built, and what we refuse to build
Citeable is one thing: a free scan that tells you whether the assistants your customers actually use can reach, read, and quote your site. It will not tell you if your copy is any good, or if your design converts, or whether people love you. It measures the one job everyone thinks is being done for them and nobody is checking: whether the newest source of readers can see you at all.
Every check on the report is labelled with how strong the evidence for it is. The ones with no evidence, like llms.txt, are scored at zero and we say so out loud on the report. Other tools sell those as features; we refuse to. That refusal is the whole reason this exists.
If you are paying to renew a domain this year, spend thirty seconds on this first.
Scan a page
Free, no signup. Full result, every check, no cards withheld to make you upgrade. If you decide to be watched weekly after, that is a paid tier and you can read what it does before paying.