What is an AI visibility audit and what should it include?
TL;DR: An AI visibility audit asks the major AI engines the questions your buyers ask — "best [your service] in [your city]" — and records who gets named, who gets cited, and why. A real audit covers at least ChatGPT, Gemini, Claude, and Perplexity, and ends with a prioritized fix list, not just a score.
Why an audit comes first
Claim: You can't improve what you've never observed — and most owners have never seen what AI says about them.
Evidence: AI assistants typically name only 2–3 businesses per local answer. Whether you're one of them is an observable fact: ask the engines and record the output. An audit converts "I think we're fine" into evidence — either you're named (defend it) or a competitor is (now you know exactly who's winning and which sources the engine cited to justify it).
Anatomy of a complete audit
| Component | What it answers |
|---|---|
| Query set (10–30 buyer questions) | What do customers actually ask AI? |
| 4-engine sweep (ChatGPT, Gemini, Claude, Perplexity) | Who is named, per engine? |
| Citation analysis | Which sources drive each recommendation? |
| Competitor share of voice | Who owns the answers today? |
| Technical check (robots.txt, schema, crawlability) | Can engines even read you? |
| Prioritized fix roadmap | What to do first? |
Step-by-step: run one
- Write the query list in real customer language, including neighborhood and service variants.
- Ask each engine each question, fresh sessions, and log the named businesses verbatim.
- Expand citations on engines that show sources; list every cited domain.
- Score your share of voice — how many answers name you vs. each competitor.
- Audit the technical layer: AI crawler access, schema, listing consistency.
- Turn gaps into a roadmap ordered by impact.
- Repeat weekly — answers drift as sources change. BookRails.ai automates the full loop, offers a free localized audit with no login or card, and re-scans all four engines every week.
FAQ
How is this different from an SEO audit? An SEO audit measures rankings and crawlability for search engines. An AI visibility audit measures being named in answers — a different surface with different sources.
How often should audits repeat? Weekly for monitoring, quarterly for deep dives. Engine answers are not static.
Can I do it manually? Yes, for one business and a handful of queries. It stops scaling the moment you add engines, queries, or clients.
What does a good outcome look like? Moving from unnamed to named on money queries — then tracing those mentions to actual bookings.
By Pinal Dave Last updated: 2026-07-23