Everyone wants to know if ChatGPT, Perplexity, Gemini and the rest are recommending them, and a wave of tools has appeared to sell you the answer. Most hand you a number. Here is the question almost nobody asks: is that number measured, or modelled? Because that difference decides whether you are making real decisions or decorating a dashboard.
Strip away the branding and there are only two things you can actually want to know:
That is it. Everything else is presentation. The honest test is what a tool does with each one.
Let us be straight about how this works, because some marketing makes it sound like magic. To find out whether an assistant cites you, you ask it. You send prompts to the models and read the answers. We do exactly that. So does everyone else. Firing prompts is table stakes; it is not the moat, and anyone telling you their prompt list is secret sauce is selling you the wrong thing.
The difference is what you are handed back. A score collapses dozens of answers into one figure, and you cannot audit a figure. You cannot see which question triggered it, which model answered, what the answer said, or who got recommended instead of you. Evidence keeps the receipt: the exact question asked, the exact model that answered, cited or not, the verbatim sentence if you were, and which competitor showed up when you were not. One of these you can act on Monday morning. The other you can put on a slide.
Here is where most scoring quietly cheats: it treats a vague brand mention and a hard, sourced citation as the same point. They are not. There is a ladder of proof, and an honest tool grades where you actually landed:
A single blended number throws all of that away. We keep the rungs visible, and we tell you honestly when a citation is only a mention. Our published methodology sets out exactly how we grade it. A tool that will not grade its own confidence is asking you not to look too closely.
Now the inbound side. AI crawlers do not execute JavaScript, so a tool watching your traffic with a browser tag, the way classic analytics works, cannot see the AI bots at all. They came, they fetched, they left, and nothing client-side noticed. Catching them means watching server-side, where the request actually lands. And catching them honestly means one more step: verifying identity, because anyone can set their user-agent to GPTBot. The check is forward-confirmed reverse DNS against published IP ranges. If a tool just reads the user-agent and believes it, its crawler numbers are inflated by every impostor on the internet.
This is the one that matters most, and the rarest. You read a report, you make a change, and then what? Most tools hand you the next report and let you assume. The honest version is a loop: you mark a gap as actioned, and about a week later the same question gets asked again so you can see, with evidence, whether the citation moved. Before and after. No "we got you ranked" promises, because AI answers shift for a hundred reasons and anyone claiming clean causation is guessing. Just: here is what happened, measured.
You do not need to take anyone's word, mine included. Ask the tool to show you the receipt:
If the answers are yes, it is measuring. If all it can offer is a number that went up or down, it is modelling, and a model of your visibility is a guess wearing a lab coat. We built Unsourced because we were tired of scores we could not audit. The whole product is one stubborn idea: evidence, not scores. If you want to see what that looks like with real receipts, the demo is open, no login.
A score collapses dozens of answers into one figure, and you cannot audit a figure. You cannot see which question triggered it, which model answered, what the answer actually said, or who got recommended instead of you. You are asked to trust the arithmetic. Evidence keeps the receipt: the exact question, the exact model, cited or not, the verbatim sentence, and the competitor who won when you did not.
To check citations, yes, everyone sends prompts to the models and reads the answers. We do too. Firing prompts is table stakes, not a moat, and anyone claiming their prompt list is secret sauce is selling the wrong thing. The difference is entirely in what you are handed back: a blended number, or an auditable record.
AI crawlers do not execute JavaScript. A tool that watches your traffic with a browser tag, the way classic analytics works, cannot see them at all. Catching them means watching server-side, and catching them honestly means verifying identity with forward-confirmed reverse DNS, because anyone can set their user-agent to GPTBot.
Ask it to show you the receipt. Can it show the exact answer text where you were cited? The competitor that won the answer you did not? Whether a real, verified crawler fetched your page? Whether your last fix moved anything? If yes, it is measuring. If all it offers is a number that moved, it is modelling.
The same argument, applied to a specific tool comparison.
Read it →Real captured answers, no login.
Read it →Exactly how we grade a citation from mention to proof.
Read it →Unsourced captures which AI assistants cite you, proves which crawlers really fetched your pages, and re-checks after you act — evidence, not a score.
© Unsourced — the evidence layer for AI search.