How to measure AI visibility
Published by Krister Ross · Updated July 2026
You cannot improve what you do not measure — and AI visibility is fully measurable. Here are the metrics that matter, the method that makes them comparable over time, and the tools that automate the job.
In short
AI visibility is measured by running a fixed set of questions — the ones your customers actually ask — through ChatGPT, Gemini, Perplexity and Google AI Overviews on a schedule, and logging four things: how often you are mentioned, how often your pages are cited as sources, what sentiment the descriptions carry, and your share of voice against competitors. Repeated monthly, this gives a trend line you can manage against, exactly like rank tracking did for SEO.
The four core metrics
Together they answer: does AI know you, trust you, like you — and prefer you?
Mentions
How often the models name your brand when answering the questions that matter in your category. The baseline metric for whether AI considers you at all.
Citations
How often your URLs appear as sources in generated answers. Citations show the models retrieve and trust your content — and they drive qualified traffic.
Sentiment
How the models describe you: positive, neutral or negative, and which attributes they attach. Two brands can both be mentioned while one is quietly talked down.
Share of voice
Your slice of all brand mentions across the question set, against named competitors. The single best summary number for AI visibility over time.
The method: fixed questions, fixed rhythm
Build a question set from real customer questions — not brand searches
Run it through the major models: ChatGPT, Gemini, Perplexity, AI Overviews
Log mentions, citations, sentiment and share of voice per model
Repeat monthly with the same set, so results are comparable
Connect changes to the work you shipped, and prioritize next month from data
Tools: manual or automated
You can start manually: pick twenty questions, run them through the models, log the results in a spreadsheet. It works, but it is slow, and models vary their answers — single runs mislead.
CitationLab AI Monitor automates the whole loop: scheduled runs across models, mention and citation tracking, sentiment scoring and competitor share of voice — reported monthly with the trend lines that show whether the work pays off.
Frequently asked questions
The questions we hear most often about measuring AI visibility.
Dedicated monitors like CitationLab AI Monitor run your question set through ChatGPT, Gemini, Perplexity and AI Overviews on a schedule and track mentions, citations, sentiment and share of voice. Manual spot checks work for a first baseline.
Track which of your target queries trigger an AI Overview and whether your pages are cited in it. Combine Search Console impression data with scheduled Overview checks — impressions without clicks often mean you appear in an Overview.
Monthly is the practical rhythm. Model answers vary run to run, so scheduled repeated runs with the same question set are what make trends real rather than noise.
It depends on category size, but the goal is direction: growing month over month, and ahead of your named competitors on the questions with commercial intent.