An AI never answers the same way twice, and your customers never phrase their question the same way. The Deep Scan rephrases a key question, repeats it 50 times and tells you whether you are cited consistently or by chance.
Run my first Deep Scan 14 days free, no credit card, with 100 free AI credits.You ask ChatGPT your question and your brand appears. The next day, it is gone. Nothing changed on your site: that is the model's variability. AI platforms draw their answer from several possible ones, and the same question asked ten times does not return the same list of brands ten times.
Phrasing matters just as much. “Best trail watch”, “most reliable trail watch” and “watch for an ultra” are the same intent to you. Not to an AI, which can cite three different brands. Daily tracking smooths out the first kind of variability with a 7-day average. The Deep Scan measures both at once, on the question that matters most to your revenue.
One question, five phrasings, ten runs each, on the AI platform or platforms of your choice.
A tracked question, or a new question with high commercial stakes. You then select one or more AI platforms: ChatGPT, Gemini, AI Mode or Perplexity.
The question that brings you customers.
The Deep Scan writes five realistic ways of asking the same question, the way different customers would. Each phrasing is run ten times, which makes 50 answers per AI platform, all collected under the same conditions as your daily tracking.
What your customers actually see.
The robustness score sums up the share of runs where you are cited: the higher it is, the more your presence holds. The run map shows, phrasing by phrasing, where you hold and where you drop off.
Where you are fragile, and why.
The Deep Scan does not stop at the score. It shows you who takes your place and what feeds the answers.
Each cell on the map is one answer: cited, mentioned without a link, or absent. You spot at a glance the phrasing that makes you disappear.
Work first on the angle where you drop off.When you are not cited, who is cited instead? For each competitor, the report counts the runs where it appears and the phrasings where it replaces you.
Compare its page with yours on that exact angle.The sites AI platforms cite most across these 50 answers, and the searches ChatGPT runs before answering. These are your levers to get back into the answers.
Target the sources that cite your competitors without you.Every answer can be read in full, phrasing by phrasing. The report exports to PDF for your team or your client, white-label from Consultant upward.
Keep the PDF to compare before and after.Before and after optimising a page. The first Deep Scan gives you the baseline, the second measures the effect of your work on the target question. Allow a few weeks between the two: AI platforms take time to pick up edited content.
When daily tracking moves for no reason. A question goes from cited to absent? The Deep Scan tells you whether it is a real drop or a normal variation of the model.
On your high-stakes commercial questions. The question that comes before a purchase, a quote request or a meeting. The one where being cited half the time is not enough.
Cockpyt AI is the only GEO tool to offer this analysis.
A Deep Scan uses 20 AI credits per AI platform analysed, 21 on Perplexity. On ChatGPT, Gemini and AI Mode together, allow 60 credits. Your credits are included every month in your plan: 200 on Solo, 400 on Consultant, 700 on Agency, 1,200 on Scale, and 100 during the free trial. Simulate your credits
Coming next: the Deep Scan will soon simulate a conversation over several exchanges. ChatGPT increasingly asks clarifying questions before recommending a brand, and we want to measure whether you hold all the way to the final recommendation.
Daily tracking asks the same phrasing of your questions every day and gives you a 7-day trend. The Deep Scan focuses on a single question at a given moment: it rephrases it five times and repeats each phrasing ten times. One measures change over time, the other measures robustness.
They keep the intent of your question with different angles, the way several customers would: broader, more specific, price-focused, use-focused, or phrased as a request for advice. You see all of them in the report, with their answers.
It is the share of runs where your brand is cited, on a scale of 100. The higher the score, the more your presence holds up against rephrasing and model variability. A low score on a question where daily tracking shows you as cited signals a fragile presence.
ChatGPT, Gemini, AI Mode and Perplexity. You choose one or more AI platforms each time you run it, and each one is billed separately: 20 credits, or 21 for Perplexity.
No. The Cockpyt Score is still calculated on your daily tracking, so your curve stays comparable from one day to the next. The Deep Scan is a one-off analysis, shown separately.
Yes. The 100 AI credits in the trial let you run a Deep Scan on your key question, on top of a GEO audit and an AI Perception analysis. No credit card required.
Choose your key question, run the Deep Scan and see where you drop off in a few minutes.
Run my first Deep Scan 14 days free, no credit card, no commitment.PRODUCTS
RESOURCES
COMPANY
