Coming soon — not available yet
Were you named in the answer?
Everything on this site today is free, and it stays free. It measures whether an assistant could read and cite a site, and whether a search engine returns it for the questions its customers ask. That is the part we can do on our own hardware at effectively no cost per report.
The step after that is reading the AI answer itself. Google’s AI Overview names the sources it used; so do Perplexity and ChatGPT when they answer with links. Recording whether a site was among them — and which sites were instead, and how that changes week to week — is a different kind of measurement, and it is the one people actually want.
It is not free for us. Reading an AI answer costs a fraction of a penny per question through a data provider, and asking the named assistants directly costs several times that. A handful of pennies per report is nothing to bill a customer and a great deal to absorb across everyone who visits, so this will be a paid option rather than something quietly added to the free tier.
What it would show
- Whether Google’s AI answer cited you, for each of your customers’ questions, with the sources it used instead.
- Movement over time. One measurement is a snapshot; the useful version is “you were cited for two of eight questions this week, down from three”.
- Which competitor is being read. Not a ranking position — the actual pages an assistant is summarising when it answers your customers.
- The named assistants, later and at a higher price, because asking ChatGPT, Claude and Perplexity directly costs several times more than reading Google’s answer.
What we will not claim
An API answer from a model is not the same thing as the consumer product. The retrieval stack and the ranking differ, and anyone selling you “your ChatGPT citation rate” from an API call is selling you a proxy. It is a close proxy and a useful one, and it will be labelled as one.
Every measurement is a sample: one question, one locale, one moment. AI answers vary between runs for the same query, so a single check is weak evidence and a trend is strong evidence. The product will report the sample size rather than a single confident verdict.
In the meantime, free
The strongest evidence of a real citation is not something we buy — it is in your own server logs. A request from ChatGPT-User is a real person’s assistant fetching your page because they asked it something. Upload an access log on any report and we reconcile it against what we measured.
The rest of the free report already covers whether assistants can reach you, whether search returns you for your customers’ questions, which platforms own those questions, and what to publish so that you are quotable when they arrive.
Would you pay for this?
Genuinely useful to know before it is built, including if the answer is no.