To track your brand's mentions in AI, build a fixed set of buyer prompts, run them across ChatGPT, Perplexity, Gemini, and Google AI Overviews on a schedule, and log whether you're cited, mentioned, or absent — plus which sources are cited and how rivals appear. Roll the totals into an AI share-of-voice score and watch it over time. Do it by hand, or with a dedicated tool.
You can't manage what you can't see. Most brands have no idea what AI says about them because they've never measured it — they find out when a buyer mentions a competitor "came up when I asked ChatGPT." Here's how to put a number on it.
What To Actually Measure
"Are we mentioned?" is too crude. Track five dimensions on every prompt:
- Presence — cited (named positively), mentioned (named neutrally), or absent.
- Framing & sentiment — how you're described, and whether it helps or hurts the decision.
- Citations — which sources and URLs the engine references about you, and how many you control.
- Competitors — who shows up when you don't, and who's framed as the default choice.
- Accuracy — is what AI says correct and current, or is it quoting stale prices, old products, or a rival's claim?
The Manual Method
For a lighter starting version on just nine prompts, use our 10-minute AI citation self-test.
Manual Or A Tool?
Manual tracking is free, precise, and teaches you exactly how each engine treats you. It's also time-consuming and hard to scale past a few dozen prompts. Tools automate the runs, the alerts, and the dashboards.
| Manual | A tool | |
|---|---|---|
| Cost | Free, but your time | Subscription |
| Best at | Depth, nuance, learning the engines | Scale, scheduling, competitor alerts |
| Limit | Hard to scale & sustain | Measures — doesn't diagnose or fix |
If you go the software route, see the best AI visibility tools. Either way, remember the limit: tracking tells you where you stand, not why. Turning the gap into action is the work in our keystone guide, how to get cited by ChatGPT, Perplexity & Gemini.
Rankings can hold steady while your AI presence collapses. AI share-of-voice — how often you're cited, accurately, versus rivals — is the metric that now tracks revenue. It has to be measured directly: traditional rankings predict only ~45% of AI visibility, and the engines publish no official mention analytics (the 2026 AI search data). Track that, not vanity mentions.
Frequently Asked Questions
How do I track my brand's mentions in AI?
Build a fixed set of buyer prompts, run them across ChatGPT, Perplexity, Gemini, and Google AI Overviews on a regular schedule, and log whether you're cited, mentioned, or absent — along with which sources are cited and how rivals appear. Track the totals as an AI share-of-voice score over time, manually or with a tool.
What should I measure when tracking AI mentions?
Five things: presence (cited, mentioned, or absent), framing and sentiment, which sources the engine cites, how competitors appear on the same prompts, and accuracy — whether what AI says about you is correct and current.
How often should I track AI brand mentions?
At least monthly. AI answers shift as engines update and as sources change, so a fixed prompt set re-run on a schedule is the only way to see whether your visibility is improving or eroding.
The AI Visibility Audit runs 25+ buyer prompts across every engine, scores your share-of-voice and accuracy versus rivals, and tells you exactly why the gaps exist — and what to fix first.