The AI Presence Score is the headline number of every benchmark: a 0–100 composite of how favorably AI represents you when buyers ask evaluation questions. It’s written at scoring time from the same underlying data every other number in your report reads, so the score and its drill-downs never disagree.

What goes into it

The score combines four dimensions, each measured across every scored response in the run:

• Presence — do you show up at all, and how prominently? Every response is classified by how much of it is about you, from being the primary subject down to a passing mention or complete absence.
• Competitive outcomes — in head-to-head comparisons, does the AI pick you, a competitor, or call it a tie?
• Accuracy — is what AI says about you true? Factual errors and outdated claims count against you; each flagged claim is verified against live web evidence first.
• Framing — when you do appear, is the sentiment and positioning favorable?

Visibility

Visibility is the share of evaluations where you were mentioned at all. One important detail: prompts that name your company directly (head-to-head prompts like “Compare X vs Y”) are excluded from this number — an AI mentioning you because the question forced it to isn’t real visibility. What’s left measures organic visibility: how often you surface when a buyer asks an open question about the category.

Win rate

Win rate covers the comparisons where the AI actually made a call. Of those decided head-to-heads, it’s the share you won — with ties counted against the total, since a hedge is not a win. Every win and loss links to the specific responses behind it.

If the AI recommends one of your own products over another, that’s a win for your brand, not a loss — scoring understands your product family. See Subjects: brands and products.

Reading the score

• The trend is the signal. AI answers have natural variance run to run; a two-point wiggle means nothing, a ten-point move across three runs means something. Every run writes a snapshot, and your home page charts the trend.
• Comparable by construction. Because your prompt library persists across runs, every benchmark asks the same questions — score movement reflects changes in AI’s answers, not changes in what you asked.
• Per-product scores are separate. Each product you track gets its own score over its own prompts, alongside the brand-level score.