An AI visibility audit tests the questions buyers type into ChatGPT, Gemini and Google AI Overviews, then records what each engine says about your business on that date. Each answer is logged for three things: whether you are named, who is named instead, and whether any stated fact is wrong.
The audit is a measurement, not a forecast. Each engine decides independently what it cites, and answers can change from one week to the next. What an audit gives you is a dated, screenshot-backed record of where you stand today and which sources the engines rely on when they answer.
What does an AI visibility audit test?
An AI visibility audit tests buying questions, not keywords. A buying question is the full sentence a prospect types, such as “which SEO agency in Quezon City can help my clinic show up on ChatGPT.” Most name a service, a market and often a city, because engines answer local and national questions differently.
The question list sets the scope. AvantScale’s free audit and AI Snapshot each test 10 buying questions, and the Full Audit tests 25 to 40 prompts. The questions are drafted from your services and market, confirmed with you before testing, and kept for later retests.
Which answer engines does it cover?
A standard AvantScale audit covers three engines: ChatGPT, Gemini and Google AI Overviews. Every question runs on all three on the same day, so a 10-question audit produces 30 answers. Same-day testing matters: comparing answers from different dates mixes engine changes with real differences in visibility.
Testing conditions are recorded because they change the answers. AvantScale uses the free version of each engine, a logged-out browser and the market’s location where the engine allows it. The engine, date and location setting are logged beside every answer, so next month’s retest compares like with like.
What gets recorded for each answer?
Every answer produces one row in the audit log, with five fields:
- Named or not named: whether the answer mentions your business by name.
- Who is named instead: the competitors, directories or marketplaces the engine recommends.
- Sources cited: the pages or sites the engine links to or credits.
- Wrong facts: any stated detail about you that is incorrect or outdated, such as hours, services, prices or location.
- Screenshot: a dated capture of the answer as it appeared.
Wrong facts go into a discrepancy log with a severity level. A wrong price, location or credential is critical. Outdated hours or a missing service is high. A dated description is medium. Severity sets the order of corrections, made on sources you control, such as your website and Google Business Profile.
What do the different audit levels include?
AvantScale offers four audit levels. They use the same method and differ in depth: how many questions are tested, whether competitors or locations are compared, and whether the deliverable includes a 90-day plan.
- Free audit: 10 buying questions across ChatGPT, Gemini and Google AI Overviews, and a one-page summary of where you appear, what is missing or wrong, and the top 3 fixes. It is delivered by email or on a live call, whichever you prefer. It has no roadmap.
- AI Snapshot: 10 priority buying questions across the same three engines (30 answers), with a short written summary and top fixes. It has no 90-day roadmap.
- Full Audit: 25 to 40 buying prompts, written findings, and a 90-day roadmap that lists the month, action, owner and evidence for each step.
- Competitive or multi-location audit: named competitors or locations compared on the same questions and dates, with the scope set in the proposal.
The consulting and audits page explains how each engagement runs. Current paid audit prices are on the Pricing page.
Free audit
See where your business shows up in AI answers
We run ten of your buying questions through ChatGPT, Gemini and Google AI Overviews and show you where you stand. A person replies within 24 hours, Monday to Friday. You keep the audit either way.
What an AI visibility audit does not do
An AI visibility audit does not guarantee citations, rankings, traffic or leads. It records what each engine said on one date. No audit can force ChatGPT, Gemini or Google to change an answer. Corrections go to sources you control, and each engine decides independently whether to use them.
The free audit leaves out three things by design: a roadmap, a competitor benchmark and a technical crawl. It reports counts only (named, not named, or wrong fact) and no invented score or percentage. Those belong to the Full Audit and the Competitive audit, scoped in writing first.
How should you prepare for an audit?
Four inputs make the question list accurate and the findings easier to act on:
- Your website address and the services you most want to be found for.
- The country and cities you serve, so each question carries the right market.
- The questions customers already ask you by phone, chat or email.
- Any facts you suspect engines get wrong, such as old hours or a former address.
Every finding is reviewed by a person before delivery. Each line of the summary is traced to a logged answer and its screenshot, and the reviewer’s name and date are recorded. That trace is the core of AvantScale’s Citation Guarantee: tracked questions, source-traced corrections and human sign-off.
AI visibility audit FAQ
How long does an AI visibility audit take?
The delivery date depends on the audit level and is confirmed in writing once the question list is agreed. For paid audits, the date counts from confirmed production start after a complete brief, never from the first inquiry. All questions in one audit are tested on the same day.
Will an audit improve my visibility in AI answers?
No. An audit measures; it does not change any engine’s answers. It shows which buying questions you are missing from and which facts are wrong, so the next fixes target the right pages and sources. Whether an engine then cites you is the engine’s decision.
How often should an AI visibility audit be repeated?
Retesting the same question set each month shows the trend. AvantScale retainers retest tracked questions monthly and report AI Share of Voice per engine: the number of questions where you are cited, divided by the number tested, times 100. It is a process metric, not a promised outcome.
See where your business appears today
The free audit runs 10 of your buying questions across ChatGPT, Gemini and Google AI Overviews and shows what each engine says about you. Get your free audit.
