How to Measure AI Visibility: Complete Methods to Track Your Brand's Exposure in ChatGPT, Perplexity, and Claude

A medical device client came to us last week with a problem: 'We've spent two years building our content marketing engine, and our Google rankings are solid. So why, when customers ask ChatGPT "What blood glucose monitoring solutions are available in Taiwan?" does our company not show up at all?'

I had them open ChatGPT, Perplexity, and Claude and ask the same question in each one. Across all three responses, competitors got named four times. They got zero. SEO rankings don't matter if the AI never mentions you.

This is why you need to measure AI visibility.

What is AI visibility?

AI visibility tracking means measuring how often your brand gets mentioned, cited, or linked when AI engines (ChatGPT, Perplexity, Claude, Google AI Overviews, and others) answer user questions. The question isn't 'What's my Google rank?' It's 'When users stop clicking links and just read what the AI tells them, does my company appear in that answer?'

Search engines give you a rank. AI gives you placement in the answer itself. Rankings appear in lists. Citations appear in the text. And citations are scarce. Most AI engines pull from just two to five sources per question. If you're not in that handful, you're invisible.

Three methods you can run today

You don't need to wait for mature tools. You can measure this right now. Each approach requires different effort and offers different precision.

Manual query baseline. List 20 to 40 questions your potential customers actually ask. Use full questions, not keywords, like 'What should I watch out for when deploying a RAG system in B2B SaaS?' instead of just 'RAG system.' Run each through ChatGPT, Perplexity, Claude, and Gemini. Log three things: whether your brand appears, whether the description is correct, and which URLs the answer cites. This is the crudest method but provides the clearest results. You'll learn which pages the AI recognizes in your industry.

Citation source reverse lookup. Pull out every URL cited in the AI responses. You'll discover something uncomfortable: AI tends to cite review roundups, community discussions, industry media, and wiki-style entries far more than brand websites. This shows you where your coverage gaps are. In one fintech case, Perplexity was citing three industry forums for a topic 70% of the time, forums the client had never invested in.

Automated monitoring. Once you're running more than 50 queries and tracking week-to-week changes, manual tracking breaks. Use an API script to run the same queries on a schedule, log brand presence, position in response, sentiment, and citing domains into a database, and track the trend line. Redesign your site or publish new content, then check back two weeks later to see if the work moved the metrics.

Which metrics matter

Don't just track whether you're mentioned. These metrics tell you something:

Description accuracy is the metric most people skip. One client had the AI calling them 'a software outsourcing firm' when they actually do automation consulting. A wrong description loses deals with high-value customers, and you won't know it unless you check.

After the measurement

Measurement doesn't increase visibility. It's a diagnostic. What moves visibility: getting third-party sources cited by AI to mention you, rewriting website content into structured Q&A that AI can extract from easily, and measuring again. This playbook is called GEO (generative engine optimization) internationally and FDM (Front-Line Deployment Marketing) here. Both describe the same work: getting your brand into the AI's answer.

The first step in any FDM project is running an AI visibility baseline. It gives you a clear report on your starting point. Product building works the same way: talk doesn't matter, results do. The only metric that counts is whether the AI cites you in its response and gets your story right.