Home / Features / Cockpyt Score
A score out of 100 that answers three questions at once: do AI platforms talk about you, where in their answer, and does a link point back to your site. One number to follow, three levers to pull.
Get my score for free 14 days, no card requiredReading citations one by one leads nowhere. You open twenty AI answers, you note that your brand shows up in eight of them, and you still have no idea whether that is good, whether it is improving, or what you should fix first.
The problem is not the volume of data, it is the absence of a unit of measurement. Being cited once at the top of an answer with a link to your site is not worth the same as being cited five times at the end of a paragraph with no link. Simply counting mentions flattens that difference.
The Cockpyt Score brings these three dimensions together into a single number. You follow one figure over time, you compare it with your competitors, and you know which of the three components is costing you points.
Each answers a different question, and each is fixed through different actions.
Do AI platforms talk about you?
The baseline component. It measures how often your brand appears in answers, across all your tracked questions and all three platforms.
Weak presence rarely signals a content quality problem. It usually signals that AI platforms find no source connecting you to the subject.
Where in the answer?
An AI rarely cites a single brand. It lines several up, and the order matters: the first one named is the one the reader remembers, and often the only one they click.
This component measures your average rank inside the answers where you appear. It is the closest equivalent to the notion of ranking in search, applied to written text rather than to a list.
Can people reach you?
Being named without being linked is awareness without traffic. This component measures the share of your citations that come with a link to your site.
It is the dimension most often ignored, and the one that best explains the gap between a brand that is highly visible in AI answers and a brand that actually gets visits from them.
If no AI platform cites your brand on your tracked questions, your score is zero. Not 12, not 30. The other components compensate for nothing, because they have nothing to measure.
That choice is deliberate. An indicator showing a reassuring number while nobody talks about you is not an indicator, it is decoration. A zero is unpleasant to look at, but it tells you exactly where you stand, and that is the starting point you need.
Two brands can share the same total score for opposite reasons. It is the component breakdown that tells you what to do.
Your presence is good, your position decent, but AI platforms name you without pointing back to your site. Your brand is known, it generates no visits.
Work on: pages AI platforms can cite as a source, and their technical reachability.You appear often, but always after three or four competitors. The reader has already made a choice before reaching you.
Work on: third-party content connecting you to the subject, not just your own site.A single question is pulling your score up. Your visibility rests on one subject, and you do not cover the rest of your market.
Work on: widening the subjects on which you are a legitimate source.Platforms do not rely on the same sources. A brand that performs on Perplexity can be invisible on ChatGPT, and the reverse is equally true.
Work on: the source types favoured by the platform where you fall behind.It is not a Google ranking. Search measures your rank in a list of links. Here, the question is whether your brand is named inside a written text, and that text is not built like a results page. The two are tracked separately, for that reason.
It is not an estimate. No extrapolation, no predictive model, no stored database of answers. Every score comes from real queries sent to the AI platforms at the moment of the scan. An answer recorded last month says nothing about your visibility today, because AI platforms go looking for sources every time a question is asked.
It is not an absolute figure. A score of 45 does not mean the same thing in a market dominated by three players and in a fragmented one. What matters is your movement over time and your gap with the competitors you track.
Every scan asks your questions under identical conditions, with no user account, no conversation history and no personalisation. A change in your score therefore reflects a real change, not a change of protocol.
If our measurement method evolves, the change is annotated in your history. You always know what you are comparing, and you never mistake a technical change for a market movement.
The score tells you where you stand. These modules tell you why.
There is no universal threshold, and be wary of any tool that gives you one. In a market dominated by two or three established players, a modest score can represent a solid position. In a fragmented market, the same figure is disappointing. The two useful reference points are your own movement over time, and your gap with the competitors you track inside your project.
Because AI platforms are not deterministic: the same question asked twice does not produce exactly the same answer. Small variations are therefore normal and call for no reaction. What matters is the trend across several weeks. It is also why we measure weekly rather than daily: daily tracking would mostly display that variability, not your real visibility.
It depends on which component is costing you points, and that is exactly what the breakdown shows. Weak presence is addressed through third-party sources connecting you to your subject. A low position is addressed through the authority of those sources. Weak contact is addressed through the technical reachability of your site and the presence of citable pages. The monthly GEO audit gives you these actions in priority order.
You get both an overall score and a breakdown per platform. That matters, because platforms do not rely on the same sources: a brand that performs on one can be invisible on another. The overall score gives you the wide view, the breakdown tells you where to concentrate your effort.
Yes. The competitors you add to your project are measured on the same questions, under the same conditions and during the same scans. The comparison is therefore direct, and it is often the most actionable information: it tells you not only that you are not cited, but who is cited instead.
Allow 3 to 8 weeks. AI platforms take time to pick up new or edited content, and the delay varies between them. If you want to check the effect of a change without waiting for the weekly scan, an on-demand scan gives you an immediate reading.
Create your project, run a first scan, and get your Cockpyt Score with the breakdown of its three components.
Get my score for free 14 days, no card required, no commitmentPRODUCTS
RESOURCES
COMPANY
