Plenty of businesses have now asked ChatGPT about their industry, seen a competitor's name come up, and felt a jolt. Fewer have turned that jolt into a measurement process. That is the difference between worrying about AI visibility and managing it.
Here is how to track it in a way that produces numbers you can trust.
Why one-off checks are not enough
Asking an AI tool a question once tells you almost nothing. The same question can produce different answers on different days, in different sessions, and with slightly different wording. Models get updated. Retrieval results shift.
A single check is a snapshot with a wide margin of error. Useful measurement means asking a consistent set of questions, on a schedule, and looking at the rates across many runs. If you have not done any checking yet, start with our guide to a one-time AI visibility audit, then build the ongoing process below on top of it.
Step 1: Build your prompt set
List the questions where being mentioned would actually drive business. These usually fall into three groups:
Category questions. "Best CRM for small manufacturers." "Top digital marketing agencies for e-commerce."
Problem questions. "How do I fix declining organic traffic?" Questions where your expertise is the answer.
Brand questions. "What is [your brand]?" "Is [your brand] legit?" What the models say when asked about you directly.
Write each question the way a real customer would phrase it, and include a few phrasings per question, because wording changes answers. A workable starter set is 20 to 50 prompts.
Step 2: Run them consistently
Run your prompt set across the tools that matter to your audience: ChatGPT, Google's AI features, Perplexity, and any others relevant to your market. Use fresh sessions so earlier questions do not color later answers. Repeat each prompt more than once per cycle, because a brand that appears in three runs out of five is in a very different position than one that appears in one out of five.
Monthly is a reasonable cadence for most businesses. Doing this by hand gets tedious fast, which is why tooling exists for it, but the process matters more than the tool.
Step 3: Record more than yes or no
For each response, capture:
Mention: did your brand appear at all?
Position: first recommendation, or an afterthought in a list?
Sentiment and framing: how was your brand described? Accurately?
Citations: which pages were cited, yours or someone else's?
Competitors: who else appeared, and how often?
Citations are the most actionable field in the whole dataset. They tell you exactly which pages are feeding the answers, which is a to-do list for your content and PR work.
Step 4: Turn the data into metrics
A few simple metrics cover most needs:
Mention rate: the share of runs where your brand appears, per prompt and overall.
Share of voice: your mentions compared to each competitor across the same prompt set.
Accuracy rate: the share of mentions that describe your brand correctly.
Citation share: how often your own pages appear as sources.
Track these over time and the trend line becomes the story: is your visibility growing, flat, or losing ground to a specific competitor?
Be honest about correlation and causation
One warning that will save you from bad decisions: AI answers change for many reasons, including model updates that have nothing to do with anything you did. If your mention rate rises a month after you published new content, that is a correlation. It is a hopeful sign, not proof.
Label it that way in your reporting. Overclaiming causation is how measurement programs lose credibility. The honest version, "our mention rate rose from X to Y over this period, alongside these actions," is still a compelling story, and it stays true.
Connect it to business results
Mentions are a leading indicator, not the finish line. Watch for the downstream signals: referral traffic from AI tools in your analytics, leads who say "ChatGPT recommended you," and branded search growth. Tying the pieces together takes some setup, and our analytics and custom reporting work covers exactly that.
And when the data shows a gap, the fix loops back to fundamentals: the sources and signals covered in how AI search engines decide which brands to mention. For local businesses, the playbook in our guide to AI visibility for local businesses is the natural next step.
If you would rather have the tracking, reporting, and fixes handled as one program, that is the core of our AI visibility services.