// Start here
Find out what the machines say about you
Three weeks. Fixed price. Your prompt set, your citation gaps, every wrong answer traced to its source, and a 90-day roadmap you keep.
Most brands have no idea what AI says about them
You can read your own site. You cannot read the answer a model gives a buyer who asks about you at 11pm, and neither can your analytics. That answer is being assembled right now, from your pages, your competitors' pages, Reddit threads, aggregators and explainers you have never seen. It is either accurate or it is not, and either way it is being served to everyone who asks.
The audit is the first honest look. We build the questions your buyers actually ask, run them through every major engine several times each, and show you exactly what comes back.
What three weeks buys
Your prompt set. The real questions, catalogued from sales calls, support tickets, community threads and keyword data, then ranked by the demand behind each one. In crypto and fintech they cluster hard: is [your brand] a scam, is [exchange] licensed in Singapore, best crypto exchange for beginners in Canada, safest way to stake ETH, best cold wallet for institutions, [exchange] vs [exchange] fees, and whatever happened in 2022 that a model still brings up unprompted. This is the durable asset: every measurement afterwards runs against the same fixed set, which is what makes a number a benchmark instead of an anecdote.
A citation gap map. Where you appear, where your competitors appear instead, and which sources the engines reach for when they answer your category. That last list is measured from the citations the engines attach to their own answers, not assembled from reputation.
Every wrong answer, traced. When a model states something false about you, we find the page it learned it from. In practice a large share of those pages turn out to be ones the client owns and can edit the same afternoon.
A readiness score with the blockers named. Nine dimensions, three pass/fail gates, from crawler access to whether your pages render anything at all without JavaScript.
And then we check it by hand
Everything above runs through the provider APIs, logged out, which is the only way to get a number that means the same thing in October as it did in July. It is also not quite what your customer sees: a signed-in user carries memory, custom instructions and whatever tests the provider is running on their account that week.
So on your highest-value prompts we also open a browser and look. Not as the measurement, because a personalized answer is not reproducible and two people would get two different results. As a check on the measurement, to catch the case where the API and the product people actually use have drifted apart. Anyone selling you "logged-in tracking" as a continuous metric is selling you a number that cannot be compared with itself.
What we will not tell you
We will not give you a trust score, a safety score or an accuracy score for your brand. No such number is defensible and we will not invent one. Every finding is framed as what a named engine actually said, on a named date, with the sources it cited. Where we cannot trace something, the report says so in those words rather than guessing.
We also will not promise that fixing everything on the roadmap raises your citation rate. Our own research found that technical readiness does not predict how often engines cite a brand, and that brand size dominates. Readiness removes the reasons you cannot be cited. That is worth real money and it is not the same claim.
Who this is for
Crypto and fintech brands where a wrong answer is a compliance problem rather than a marketing one. If a model is telling people the wrong fee, the wrong custody model or the wrong incident history, that is the thing to find first.
Straight answers.
What does the AI Visibility Audit cost?
$3,500, fixed, three weeks from kickoff. We publish the price because a category where every competitor hides behind "book a call" is a category where buyers assume the number is bad. The scope does not move and neither does the price.
What do I actually walk away with?
Your prompt set, a citation gap map across every major engine, every wrong claim AI makes about you traced back to the page that taught it, a prioritized 90-day roadmap, and a verified fact sheet for your brand. All of it is yours to keep, whoever executes it afterwards. An audit you cannot act on without the agency that wrote it is a sales document, not an audit.
Do I have to sign a retainer afterwards?
No, and the audit is deliberately built so you do not have to. It is a complete piece of work and it is priced as one. Most clients do go on to a retainer, but that is because the roadmap turns out to contain more work than they have people for, not because we withheld anything.
How is this different from the free scan?
The free scan reads one page the way a crawler does, in about two seconds, and tells you whether anything structural is in the way. The audit asks your real buyer questions to every engine, several times each, and checks what comes back against verified facts. One tells you the door is unlocked. The other tells you what people find once they walk in.
Why does a human review every finding?
Because the automated pass produces false positives, and a confidently wrong finding in a document with our name on it does more damage than a missing one. In our own testing at least one finding in eleven did not survive review. Software gets you a triage queue; a person gets you a finding list. That review is most of what you are paying for.
See what the engines see.
Run the free scan and see your site the way the machines see it. No login. When you want the full picture, the Audit takes three weeks.