Tracking AI recommendations is harder than tracking search rankings because answers are personalized, worded differently every time, and spread across assistants that each read the web their own way — so there is no single rank to check. The honest baseline you can do yourself is manual and periodic: write down the handful of real questions a buyer in your category would ask, then ask each major assistant — ChatGPT, Gemini, Perplexity, Copilot — in a clean session and record whether you are named, how you are described, and who gets named instead. Perplexity is especially useful here because it cites its sources, so you can see which of your pages (or which competitor pages) it actually pulled. Repeat on a schedule, because answers drift as the web and the models change. What you are really tracking is two things: whether you appear at all, and whether the description is accurate. If you are absent or misdescribed, the fix is upstream — crawlability, consistent entity signals, reviews and corroboration — not the tracking itself. Treat the readiness of your own site as the leading indicator you can control between checks.
Most businesses score under 40/100 and have no idea. Find your score in 15 seconds, then fix it.
Run my free scan