How to track your brand's AI visibility
A practical, repeatable framework for measuring whether ChatGPT, Claude, Gemini and Perplexity mention your business, and what to record when they do.
Tracking AI visibility means answering a deceptively simple question: when the people you want to reach ask an AI system for help, does your brand show up? This guide gives you a practical, repeatable framework that works with a spreadsheet or a dedicated tool. It covers ChatGPT, Claude, Gemini, Perplexity and Google's AI features.
Before you start: what you're actually measuring
AI answers are generated, not retrieved from a fixed index. The same question can produce different answers depending on wording, the platform, the person's location and when it's asked. That means AI visibility isn't a single number you look up once. It's a pattern you observe across many prompts, platforms and dates.
Keep that in mind throughout. The goal is not to catch one perfect answer. It's to build a consistent sample that shows where you stand and which direction you're moving.
1. Identify the questions customers actually ask
Your prompts are the foundation of everything else, so start with real language rather than keywords you wish people used.
- Mine your own conversations. Sales calls, support tickets, chat transcripts and reviews show how customers describe their problems.
- Cover the whole journey. Include discovery questions (“how do I…”), comparison questions (“what's the best… for…”) and decision questions (“is [your brand] good for…”).
- Add specifics. Real prompts often include a location, a budget, an industry or a constraint. “Best CRM” and “best CRM for a three-person real estate team” can produce very different answers.
- Start small. Twenty to fifty well-chosen prompts are more useful than hundreds of generic ones.
2. Track brand appearances
For each prompt and platform, record whether your brand is mentioned. Then go one level deeper. Note your position (first recommendation, somewhere in a list, or a passing reference) and whether the mention is a recommendation or just an acknowledgment. Over many prompts, the share where you appear, and where you appear near the top, becomes your core AI visibility measure.
3. Record competitors
Every answer that doesn't mention you probably mentions someone else. Record those names. After a few rounds, patterns emerge: a small set of competitors that appear repeatedly, sometimes ones you wouldn't expect. These are the brands AI systems currently treat as representative of your category, and they're worth studying closely.
4. Examine citations
Where platforms show sources, log the cited domains. Group them into types: your own site, competitors' sites, review platforms, directories, news and industry publications, forums. Citations reveal what AI systems rely on when describing your market. If the same third-party sources keep appearing, consider whether your business is accurately represented there.
5. Monitor sentiment and context
Read the sentences around each mention. Are you described accurately? Positively? As a premium choice, a budget alternative, or with a caveat? Outdated pricing, discontinued products or old complaints can linger in AI answers. Catching them is often the most immediately actionable result of tracking.
6. Compare changes over time
Run the same prompts on a regular schedule. Weekly works for manual tracking, and daily is possible with a tool. Keep your prompt set stable so changes reflect the answers, not your method. Annotate your log with things you did, such as a new page, a press mention or a site fix, so you can see whether visibility moved afterwards. Look at trends across many prompts rather than reacting to any single answer.
7. Connect AI visibility with traditional search performance
Finally, put AI visibility next to your traditional search data. Where you rank well and appear in AI answers, you have a strong position to protect. Where you rank but aren't mentioned, look at how clearly your pages state who you are and what you do. Where competitors appear in both, they're likely earning authority you can learn from. Seeing both views together turns two partial pictures into one.
A simple tracking log you can copy
If you're starting manually, a spreadsheet with these columns is enough. The row below is an illustrative example of how to fill it in, not real data.
| Date | Platform | Prompt | Mentioned? | Position | Competitors named | Cited domains | Tone / notes |
|---|---|---|---|---|---|---|---|
| Week 1 | Perplexity | Best bakery for custom cakes in Brooklyn | Yes | 3 of 5 | Bakery A, Bakery B | review site, local news | Positive; mentions old hours |
Manual tracking vs. a tracking tool
Manual tracking is a good way to learn. Doing it by hand teaches you how AI systems describe your market. But it scales poorly. Ten prompts across five platforms every week is fifty answers to read and record, and the results are hard to compare consistently.
A dedicated tracker runs your prompts on a schedule, records mentions, positions, sentiment and citations in a consistent format, and keeps history automatically. That frees you to spend your time on the part that matters: deciding what to do about the results.
Common mistakes to avoid
- Judging from one answer. A single response is a sample of one. Look for patterns.
- Using only branded prompts. Asking “what is [your brand]?” tells you little about whether new customers will discover you.
- Changing prompts constantly. A stable prompt set is what makes trends meaningful.
- Ignoring tone. A frequent but negative mention is a problem, not a win.
- Confusing measurement with improvement. Tracking shows where you stand; the AI visibility checklist covers what you can actually improve.
New to the concept? Start with our overview: what is AI visibility?