[ai visibility audit]
Ask ChatGPT which prop firm to use, or which broker is safe, and it will name a few. Most firms have never checked whether they are one of them. This tells you, with a number you can measure again later.
A buyer asking an AI tool for a recommendation does not get ten links. They get an answer with a handful of names in it. You are either in that answer or you are not, and there is no second page to be on.
A fixed set of buyer questions, run across the main AI tools, with your firm either mentioned or not. That gives you a visibility score you can track.
The competitors that come up in your place. Often not the sites ranking first on Google, which is the part most firms find surprising.
The pages the models lean on when they answer. Sometimes your site. Often a review site or a forum thread you have never looked at.
Old rules, old payouts, old fees. If a widely cited page has stale numbers about you, the AI answer repeats them.
Not every question is worth chasing. The ones that count are asked right before someone decides.
The same questions, re-run later, so you can see whether anything moved. Without a baseline there is nothing to compare.
In a small market one cited recommendation reaches a real share of your buyers. If your whole market is a few thousand traders or a few hundred operators, being named in the answer is not a small edge.
The other reason is that wrong information spreads quietly. A firm changes its drawdown rule, the web keeps showing the old one, and the AI answer keeps repeating it. Nobody notices until a trader complains about a rule you stopped using a year ago.
Usually ten to fifteen. They are the things a buyer types right before choosing, not vanity questions about your brand name. You see the list before anything runs.
The same questions, same wording, recorded against each model. It is done carefully rather than automatically, because the automated tools in this area are not reliable yet.
What came back, what it means, and the shortlist of fixes in the order I would do them. Some of those will be on your site. Often the most useful ones are not.
It seemed only fair to measure myself before selling it. My first reading came back at a visibility score of 1.82 percent, with an average position of 8 across the question set. That is low, and it is a normal place to start.
I mention it because a starting number that looks bad is still more useful than no number. You cannot improve something you have never measured, and most firms in this industry have not measured this at all.
If the report shows gaps worth closing, that work is AI Overview optimisation. If the underlying problem is that search engines do not clearly understand what your brand is, the background reading is entity SEO.
I wrote up how this plays out in this industry specifically in how prop firms get cited in AI Overviews.
I will run the questions and show you what comes back. If you already look strong, I will say that rather than invent a problem.
A written report showing what the main AI tools say about your firm right now. Which questions name you, which name a competitor instead, what sources they cite, and where the description of you is simply wrong. Plus a walkthrough call.
ChatGPT, Google AI Overviews and Google AI Mode, which are the three that show their sources. I also check Claude and Gemini for mentions. Grok can be added if you want it.
An SEO audit looks at whether you can rank. This looks at whether you get named in an answer, which is a different question with different fixes. Plenty of sites rank well and never get cited.
No. These systems change without notice, so a guarantee would not mean much. What the audit gives you is a starting number and the list of things that are actually in your control.
Quarterly works for most firms. The point is the comparison, so the second reading matters more than the first. I set the questions up once so they stay the same each time.
The audit is the report and the call. Fixing it is separate, and it is covered by AI Overview optimisation. You are free to hand the report to your own team instead.