Prompt set

Every AI visibility number you will ever see is downstream of the prompt set, so the quality of the panel decides the quality of everything downstream. A panel of vanity prompts, questions phrased so you are likely to win them, produces a flattering score and no information. A panel built from real buyer language produces uncomfortable, useful truth. Good sets mix intents deliberately: navigational questions about you, transactional questions with your category and location, and the conversational long-tail phrasings people actually use with assistants. They evolve, too: seasonal questions, new services and newly spotted rival comparisons earn their place over time, while dead prompts get retired. The size needs discipline as well: enough prompts for stable statistics, few enough that each one still gets sampled properly. When comparing tools, ask who writes the prompts and how; the answer tells you how seriously to take the score.