Visibility Score
The visibility score is a metric that summarizes how present your brand is in answers from AI assistants and search engines. It bundles signals like mentions, citations and recommendations into a single value. The higher the score, the more often and more prominently your company appears when people ask about your topic.
Why the score matters
Visibility used to be easy to measure: you were in third place on Google or you weren't. With AI assistants like ChatGPT, Gemini or Perplexity there is no longer a classic results list. Instead, the user gets a finished answer in which you either appear or not. The visibility score makes this confusing field tangible. It condenses many individual observations into a number with which you recognize progress and justify budgets. Without such a value, AI visibility remains a gut feeling. With it you can say: 'Three months ago we were at 22, today at 41.' This creates comparability over time and against competitors.
How it works
For a visibility score, you put a fixed list of typical questions from your field to AI systems, so-called prompts. For each answer it is checked: are you named? Are you cited as a source? Are you even actively recommended? These observations get weights. A clear recommendation counts more than a casual mention. From the weighted hits across all prompts and several AI systems, an average emerges, usually on a scale from 0 to 100. It's important that the measurement is repeatable: same questions, same systems, regular rhythm. Only then are two measurements comparable at all and the score meaningful instead of random.
Common mistakes
The biggest mistake is a shaky measurement basis. Whoever asks five questions today and another twelve next month measures noise instead of progress. Set your prompts cleanly once and keep them stable. A second mistake: checking only a single AI system. Your customers use different assistants, so several belong in the measurement. It's also risky to rate every mention equally. A mere appearance in a long list is worth less than a clear recommendation. And finally: a single measurement point says little. AI answers fluctuate from query to query. Only several measurements over time yield a reliable picture.
Relation to AI recommendations
The visibility score is closely related to the question of whether and how often an AI assistant recommends you onward. That is precisely the most valuable part of the score. A mere mention brings you attention, but an active recommendation brings you customers. That's why good scores break down what share of your visibility consists of real recommendations and what share only of casual mentions. This way you recognize whether you are visible but not convincing. If you then work specifically on citability, clear facts and good findability for AI crawlers, your score shifts from passive mention toward active recommendation – and that feeds directly into revenue.
Example
Imagine a medium-sized bike shop. The owner has fifteen typical questions put to three AI assistants monthly, such as 'Where do I buy a good cargo bike in Cologne?'. In January the shop is named in only two of fifteen answers, score 14. Afterwards the team builds a clear FAQ page, clean opening hours and real customer voices. In April the shop appears in nine answers, three times even as a clear recommendation. The score rises to 52. This number shows the owner in black and white that the work has paid off.
Common questions
At what visibility score am I well positioned?
There is no fixed threshold, because scales and prompts differ by provider. What's meaningful is above all the development over time and the comparison with direct competitors on the same questions. A rising score with a stable measurement basis is more important than an absolute number.
How often should I measure the score?
A monthly rhythm is sensible for most. AI answers fluctuate, so regular measurements smooth out the chance. More important than the frequency is that you keep questions, systems and rating constant, so that two measurements really stay comparable.