How can I measure my brand's visibility in AI assistants?

AI answers aren't a stable ranking. The same prompt can return different brands on different days, for different users or with slightly different wording, so measurement works best as sampling.

1. Build a prompt set of a few dozen questions across the buying journey: category questions, comparisons, 'best X for Y' and problem-led queries.
2. Run each prompt on each assistant on a schedule, in a clean session where possible.
3. Record whether you're mentioned, whether you're linked, where you appear in the answer and how you're described.
4. Do the same for your main competitors, so you can calculate share of voice.
5. Treat small differences as noise until you have enough answers. The AI Share of Voice Calculator shows the margin of error for a given sample.

Complement this with data you already own. Google includes AI Overviews and AI Mode traffic in Search Console's Web search results. Visits from chatgpt.com or perplexity.ai show up in analytics as referrals, although some AI-driven visits arrive without a referrer and are counted as direct.

An automated checker saves the manual work of running and logging prompts, but the same rule applies: judge trends, not single answers.