How to Audit Your AI Visibility
An AI visibility audit is the foundation of every credible AI search optimization program. Here is the framework Aethon uses: what to test, what to measure, and how to turn findings into action across ChatGPT, Claude, Gemini, and Perplexity.
What an AI visibility audit measures.
A good audit answers five questions. Most teams skip straight to the second. The first matters more, because many brands turn out to be effectively invisible.
Does AI know you exist?
When AI talks about your category, does it know your brand exists at all? This is the question most teams skip, and the one where many brands discover they are invisible.
How often does it name you?
When AI recommends in your category, how often does it actually name you against how often it names everyone else?
What does it say about you?
When AI names you, what does it actually say? Is the description accurate, current, and framed the way you want, or did AI get something wrong?
Who does it name instead?
When AI doesn’t name you, who does it name? Which sources is AI citing, and which of those do you have presence on?
Do it yourself or use a tool.
You can run a basic audit by hand. Write 20-30 prompts real buyers would say in your category, run each across ChatGPT, Claude, Gemini, and Perplexity, and note when your brand is mentioned, when it is not, and who is mentioned instead. Check the citations, especially in Perplexity. This works for a one-time read, but breaks down fast.
By hand: good for a first read
A manual pass tells you whether AI knows you exist and roughly where you stand. It is the right place to start, and it costs nothing but time.
By hand: where it breaks down
AI recommendation patterns shift constantly, and the prompt volume needed for statistical significance is high. A snapshot goes stale quickly.
With a tool: continuous coverage
Tools like Aethon handle this continuously across the four engines, tracking mention rates, sources, and competitors over time instead of in a single snapshot.
A six-step audit framework.
Run this in order. Each step builds on the previous one. The first three steps are measurement, the last three are diagnosis and prioritization.
Build your prompt set
30-50 prompts mixing life-moment language, business-trigger language, direct category queries, and verification questions about your brand. Include geographic variations if relevant.
Run prompts across all four engines
ChatGPT, Claude, Gemini, Perplexity. Each will surface different results, and that is the point. Record verbatim responses.
Code mention rates and positioning
Per moment: was your brand named? First-named or later? How was it described? Did AI get anything wrong?
Identify the cited sources
Which publications, communities, and review platforms is AI citing? For Perplexity this is explicit; for the others you can probe with follow-up prompts.
Benchmark against competitors
Run the same prompt set treating competitors as the target. Which engines name them more often? In which moments? Citing which sources?
Prioritize the gap closures
Your action queue: which moments to fix first (high-volume and low mention), which sources to chase first (high-frequency-cited and low presence).
Common questions
How often should I run an AI visibility audit?
Quarterly at minimum for a snapshot, weekly with a tool for ongoing optimization. AI recommendation patterns shift continuously, so monthly is often too slow.
How many prompts do I need to get reliable results?
For a one-time snapshot, 30-50 prompts per moment category gives directionally useful results. For statistical confidence (90 percent plus), you need hundreds of prompts and continuous monitoring.
Should I audit AI for direct queries or life-moment queries?
Both. Direct queries (“best CRM”) tell you where you stand in the explicit-shortlisting moment. Life-moment queries (“we just hired a VP of Sales”) tell you where you stand in the upstream conversation that creates the shortlist.
Can I audit AI visibility without a tool?
Yes, for a first read. But continuous monitoring across all four engines, source-authority tracking, and competitive benchmarking quickly outgrow manual processes.
What if AI doesn’t know my brand exists?
That is an early-stage finding. The next step is source-authority work to get your brand into the conversations and citations AI sees. Aethon’s Action Engine prioritizes this for brands with low baseline visibility.
How is AI visibility different from SEO rankings?
SEO ranks pages for keyword queries. AI visibility measures how often AI engines name your brand in conversational answers. Two different layers of the funnel.
Related reading.
The vocabulary beneath the audit, the full playbook, and what working with us looks like.
See where your brand stands in the conversation.
We’ll show you where AI is naming you, where competitors are landing instead, and the highest-leverage actions to take next, across ChatGPT, Claude, Gemini, and Perplexity.