How to Measure AI Visibility: Why One Mention Does Not Mean Your Brand Is Part of the Decision

In the previous article, we looked at a shift already under way: as users increasingly ask AI directly, ‘Which option is right for me?’ or ‘How should I compare these choices?’, AI is beginning to shape the consideration set before users visit an official website or submit an enquiry.

That raises a more practical question. If we see our brand in an answer from ChatGPT, Doubao or another model, does that mean our AEO work has succeeded? For Jeffery Asia, the answer is not that simple. A screenshot can prove that the brand appeared once, but it cannot show whether the brand appears consistently in AI-generated recommendations or what changed after optimisation.

AEO measurement should answer more than whether AI mentions a brand. It should show whether the brand appears consistently, ranks prominently, receives sufficient coverage and shows measurable change after optimisation.

Why a Screenshot Is Not Enough to Demonstrate AI Visibility

AI-generated answers are not fixed search results. The same question can produce different recommendation orders, cited sources and wording at different times, across different models, and even across repeated responses from the same model.

If we ask once and capture one screenshot, we learn only that the brand appeared at that particular moment. We do not know whether the appearance is consistent, where the brand sits within the answer, how much attention it receives or how it compares with competitors. Meaningful AEO measurement must turn a one-off ‘yes or no’ into data that can be tested repeatedly and compared over time.

From a Single Mention to Four Measurable Dimensions

In our current work, we assess a brand’s performance in AI-generated answers across four dimensions.

The first is Mention Rate (MR). We test the same type of question repeatedly and measure the proportion of answers that actually mention the brand. This tells us whether the brand appears consistently across different responses.

The second is Occurrence Frequency (OF), which measures how often the brand appears within a single answer. Because answers vary in length, occurrence frequency should not be assessed solely as an absolute count. It also needs to be considered in relation to response length; otherwise, longer answers will naturally tend to contain more brand mentions.

The third is Average Rank (AR). When AI presents several universities, products or services, a brand usually occupies a relative position within the recommendation list. First and fifth place both count as a mention, but they do not have the same influence on a user’s consideration set. Recommendation position is therefore an important signal in its own right.

The fourth is Brand Content Ratio (BCR). This goes beyond counting how many times the brand name appears and looks at how much of the answer is genuinely devoted to the brand. Does AI simply include the brand in a list, or does it spend meaningful space explaining its strengths, relevant audiences, points of difference and reasons to choose it?

Once normalised and combined according to a consistent set of rules, these dimensions can produce an AI visibility score that can be tracked over time. The purpose is not to create an impressive-looking number. It is to make the same brand’s performance comparable across different periods, audiences and models.

AI Visibility Must Be Assessed by Persona and Scenario

There is no universal AI visibility score for a brand. An applicant considering a career change and one seeking a degree to support promotion may be similar in age, location and educational background, yet ask AI entirely different questions. The answers AI finds and organises will change accordingly.

Visibility measurement should therefore answer more than ‘What is this brand’s AI score?’. It should show which audiences and decision scenarios make the brand more likely to be seen, and where competitors are more likely to be recommended instead.

AI visibility becomes strategically useful when a brand and its competitors are compared using the same personas, questions, models and testing period. This allows us to identify the audiences where the brand needs to gain ground and the scenarios where an existing advantage should be protected, rather than pursuing broad exposure across every possible question.

Change Before and After Optimisation Matters More Than the Absolute Score

AEO must ultimately show what changed and whether performance improved. After a website structure has been corrected, an article reorganised or new content added to target sources, we can repeat the test using the same personas and questions from the previous round. We can then assess whether mention rate, ranking, content share and cited sources have changed.

This before-and-after comparison is particularly important because AI models still operate as black boxes. We cannot promise that an article will be cited after publication, nor can we assume that one effective optimisation will remain effective indefinitely. What we can do is keep a record of how models change, which content they are more likely to retrieve and which actions are followed by observable improvement.

In other words, AEO is not a one-off exercise designed to prove success. It is a repeatable system for diagnosis and feedback.

The Score Is Only Part of the Picture

Once visibility has been quantified, the next set of questions becomes more important. Why did AI recommend this brand? Which sources did it cite? What content did it rely on when recommending a competitor? Does the same approach to writing work equally well across different models?

This is one of the clearest differences between AEO and conventional exposure monitoring. A visibility score tells us where the brand currently stands. Analysing cited sources and answer structure helps us decide what to change next.

For Jeffery Asia, AI visibility is therefore not a standalone KPI. It provides a diagnostic view of how user intent, the competitive context, source ecosystems and content optimisation interact. Only when visibility can be observed, compared and reviewed over time does being ‘seen by AI’ become a growth capability that a brand can manage.

In the next article, we will explore why website optimisation is only the first step in AEO, and why different AI systems draw answers from very different source ecosystems.

If you would like to understand your brand’s visibility across leading models such as ChatGPT, Claude, Gemini, Doubao, DeepSeek and Qwen, and see how it performs across different audiences, scenarios and competitive contexts, contact success@jeffery.asia or visit jeffery.asia.