Reading Your Score
What goes into it#
The score combines how often you were mentioned across tracked prompts, where in the answer you appeared, and how consistently that held across models. A brand named first in most answers scores far above one that shows up occasionally at the bottom, even if both are technically "mentioned."
Relative movement is the signal#
Models are not deterministic. Ask the same question twice and the wording, the ordering, sometimes the recommendations shift. A few points of day-to-day noise is normal and means nothing.
What matters is the trend across enough runs to average that out, and your movement relative to competitors. If everyone in your category rose together, the model changed. If you rose while rivals stayed flat, something you did worked.
Read the score with two other views#
A number without a cause is not actionable. Every time the score moves, open:
Competitors. Who was named instead of you, and in which prompts? A drop concentrated in one comparison prompt is a very different problem from a drop spread across the category.
Sources. Which URLs did the models cite in the answers where you lost? This is where the fix usually lives. See Find Citation Gaps.
Why AI traffic converts differently#
When visibility improves, the click-through numbers may not move the way you expect. Assistants answer part of the question in the response, so a buyer often arrives already informed, later in their process, and further along than a cold search visitor.
Judge AI visibility on presence in the decision, not on raw sessions. The value is being in the answer at the moment a shortlist forms.