What each metric means, in plain language. Last updated August 2026.
An AI visibility report can look like a lot of numbers at once. Here's what each one actually means, using the exact metrics ZillaPulse reports - but the concepts apply whether you're reading a ZillaPulse report or checking manually yourself.
The percentage of AI responses that mentioned your brand by name, out of every question tested. If you tracked 10 keywords across 3 AI engines (30 responses total) and your brand appeared in 12 of them, your visibility score is 40%.
This is the headline number, but it's a rate, not a guarantee - the same question can get a different answer next time you ask it, which is exactly why tracking over multiple runs matters more than any single check.
When you are mentioned, where do you land relative to competitors named in the same response? If a response lists three competitor names before yours, that's position 4. Averaged across every response where you appeared, this tells you whether you're typically the first name mentioned or an afterthought further down the list.
A 0-100 score describing the tone of responses that mentioned you - higher means more positive language, lower means more negative or hedged language. It's worth being transparent about how this works: it's a lightweight keyword-based heuristic (counting positive and negative signal words), not a trained sentiment-analysis model. Treat it as a useful directional signal, not a precise instrument.
The percentage of responses that cited at least one real source URL. Some AI assistants browse the live web and cite sources as part of an answer; others answer purely from training data with no citation at all. A high grounding rate tells you the AI is actively pulling from real, current web content for this topic - which also means your own site has a real, direct opportunity to be one of those cited sources.
Top Sources lists which domains get cited most often across all your tracked responses - useful for seeing who the AI treats as authoritative in your category, and whether your own site is (or isn't) among them. Top Prompts ranks your tracked keywords by mention rate, showing which specific questions you're winning on and which ones you're invisible for - often the most actionable single view in the whole report.
Start with Top Prompts - it points directly at which specific questions are worth your attention first. Cross-check against Top Sources to see who's winning those questions instead of you, and what they're doing that you're not. Then treat visibility score and sentiment as your ongoing scoreboard: check them on a schedule, not once, and watch the trend rather than any single number. For the practical work of actually improving these numbers, see how to get your brand mentioned by ChatGPT.
Get these exact metrics for your brand across ChatGPT, Claude, and Gemini.
See pricingThere's no universal benchmark - it depends heavily on your category and how many competitors you're up against. What matters more than any single number is the trend: is your score moving up, down, or flat over time as you make changes?
These measure different things. Visibility score is how often you're mentioned at all. Sentiment only looks at the responses where you were mentioned, and scores the tone of that mention. You can be rarely mentioned but described very positively on the occasions you are - that's a real, useful signal on its own, distinct from how often it happens.
It's a lightweight, transparent keyword-based heuristic (counting positive and negative signal words in the response), not a trained NLP model - useful as a directional signal, not a precise measurement. Treat a big shift as worth investigating, not a single point as gospel.
Being mentioned third or fourth in a list is a meaningfully weaker outcome than being mentioned first - someone reading or hearing the answer is more likely to act on an early mention. Tracking position alongside visibility rate shows whether you're just present, or actually prominent.