Direct answer
The essential definition
An AI visibility audit tests fixed brand and customer questions across answer engines, then preserves the evidence behind every conclusion.
1. Define the business context
Start with the public domain, the business description, the category, the audience, and the regions served. This context helps distinguish a genuine competitor from an unrelated name and prevents the audit from relying on assumptions taken from test fixtures.
2. Build a balanced question set
Combine branded recognition questions with natural, unbranded customer questions. Do not insert the monitored brand or a pinned competitor into discovery questions. That would convert a recommendation test into a prompted comparison.
- Recognition questions test whether the business is understood accurately.
- Customer questions test which businesses are naturally recommended for a need.
- Comparison questions should be explicit and reported separately from discovery.
3. Preserve and classify the evidence
Save the exact prompt, complete answer, provider, model, collection time, citations, mentioned brands, and recommendation framing. Report verdict, signals, evidence, and action in that order so the reader can move from decision to proof without decoding the interface.
4. State the limitations
AI answers are variable observations. An audit should state its engines, questions, date, and comparison rules. Scope treats a changed question set as a new baseline, which avoids presenting two different tests as historical movement.
Questions people ask
How many questions should an audit include?
Use enough questions to cover recognition, category discovery, comparison, and purchase intent without creating an unfocused battery. Repeatability matters more than volume. Save every prompt and reuse the same set when measuring change across reporting periods.
Should audit questions include the brand name?
Use both branded and unbranded questions. Branded questions test whether an engine understands the business accurately. Unbranded customer questions test whether the business is discovered and recommended without being named in advance.
Why should an audit save complete answers?
Complete answers preserve the evidence behind every conclusion. They let a reviewer verify mentions, recommendations, citations, inaccuracies, and competitive context. Without saved answers, a score cannot explain what happened or support a defensible optimization decision.
Sources and methodology
- Google Search Central: structured data guidelines
Primary guidance requiring structured data to match visible page content and remain accessible to crawlers.
- OpenAI: Publishers and Developers FAQ
Primary guidance on how public pages become discoverable and citeable in ChatGPT search.
- Scope product facts and limitations
First-party audit methodology, supported engines, comparison rules, and product limitations.