AI Visibility Score Explained: Mention Rate, Ranking and Sentiment

An AI visibility score condenses how a brand shows up in AI-generated answers into one number. It is useful for spotting trends, but only if you know what goes into it.

The three ingredients

Ingredient	What it captures
Mention rate	How often the brand appears in answers to relevant questions
Ranking position	Where it sits when the model lists several options
Sentiment	Whether the mention is positive, neutral or negative

PeiGEO combines these into a 0–100 score that updates after every collection run. A brand that is mentioned often, near the top, and described favorably scores higher than one that is mentioned rarely or described critically.

Why "active" mention rate matters

Suppose you ask a model "What do you think of Brand X?" It will talk about Brand X, because you named it. That tells you nothing about whether the model would recommend the brand to a buyer who never heard of it.

For this reason, a useful mention rate counts only category and comparison questions ("best tools for…", "Brand A vs Brand B"), and leaves out questions that already contain the brand name. PeiGEO calls this the active mention rate. The same logic applies to ranking: the position that counts is the one the brand earns when the model chooses to list it on its own.

Reading the number honestly

• Look at trends, not single runs. Model answers vary from run to run, so a two-point move is noise. A steady climb over several weeks is signal.
• Compare against competitors. A score of 40 means little alone. If your closest competitors sit at 70, the gap is the finding. Share of voice (each brand's share of total mentions over the last 30 days) makes this explicit.
• Split by platform. A brand can be strong in Perplexity, which retrieves live web sources, and weak in a model that answers from training data. One blended score hides that.
• Check sentiment separately. A high mention rate with negative sentiment is a problem, not a win.

Awareness is a different question

The score measures answers to buying-style questions. It does not tell you whether a model knows your brand at all. PeiGEO has a separate AI Recognition Check for that: it asks each model, offline, whether it recognizes the brand, and it is not included in the overall score.

What to do next

If the score is low, the fix is rarely on your own website alone. Look at which sources the models cite for your category and improve your presence there. Our guide to tracking brand mentions in ChatGPT walks through building a question set, and the user guide explains each metric in the product.