What is AI visibility, and how do you measure it?

AI visibility is how often an AI assistant names your brand when someone asks a question you should be the answer to. If a buyer asks ChatGPT for the best merino running shoe and it lists three brands, each of those brands is visible for that question and everyone else is not.

The useful metric is a rate: answers that name you divided by answers collected. Both numbers matter. A brand named in 4 of 5 answers and a brand named in 4 of 80 answers are in very different positions, and any tool that hides the denominator is hiding that difference.

Three things make a rate trustworthy. First, ask the products people use, not only the developer APIs, because the consumer apps search the web differently and answer differently. Second, ask the same prompts on a fixed schedule, so a change in the number is a change in the world and not in your method. Third, keep the answer text and its citations, so every number can be opened and read.

Measure per engine, because they disagree. Gemini and Google AI Mode lean on Google's index, Perplexity cites heavily, ChatGPT names fewer brands and leans on a handful of publishers. A brand can be at 60% on one and 10% on another for the same prompt set.

Once the number exists, it should point at work. Prompts where competitors are named and you are not are content work. Sources the engines cite that never mention you are outreach work. Facts an engine gets wrong about you are brand work. Visibility is a diagnosis; the treatment is pages, mentions and corrections.