AI Visibility Audit: How to Check Your Brand in ChatGPT, Gemini and Perplexity

An AI visibility audit, also called an AEO audit (answer engine optimization), is the starting point of any GEO work. Without it you optimize blindly; with it you know on which questions you are absent, who is cited instead and why. The method below works by hand, and every step can later be automated.

Step 1: Choose the prompts to audit

The audit is only as good as its list of questions. Build it from real demand:

• questions heard by sales and support;
• Search Console queries phrased as questions;
• People Also Ask boxes on your main topics;
• comparison and recommendation phrasings: "best X for Y", "X vs Z", "alternatives to Z".

Keep 20 to 50 prompts, tag each with an intent (discovery, comparison, recommendation, brand) and a priority. Add five brand prompts ("what is [brand]", "is [brand] reliable", "[brand] pricing") to check how the AI describe you.

Step 2: Run the prompts in each assistant

Test at least ChatGPT, Gemini, Perplexity and Claude, plus Google with AI Overviews or AI Mode where available. Use a session without personal history, note the date, and run each prompt two or three times: answers vary, and one run is not enough to conclude.

For each answer, record:

1. Mention: is your brand named, yes or no.
2. Position: first, among the first three, or further down a list.
3. Description: is what the AI say about you accurate and current.
4. Sentiment: positive, neutral or negative.
5. Sources: which URLs are cited, yours or third parties.
6. Competitors: which brands appear in your place.

Step 3: Score the results

Turn the notes into a few indicators you can follow over time:

• Mention rate: share of answers that name your brand.
• AI share of voice: your mentions compared with those of each competitor on the same prompts.
• Citation rate: share of answers that link to one of your pages.
• Accuracy: share of brand descriptions without factual errors.

Break them down by assistant and by intent. A brand often does well on brand prompts and is absent on recommendation prompts, which are the ones that bring new customers.

Step 4: Audit the technical access

If AI crawlers cannot read your pages, nothing else matters. Check:

• robots.txt: are OAI-SearchBot, GPTBot, ClaudeBot, PerplexityBot and Googlebot allowed? Google's introduction to robots.txt explains the syntax, and OpenAI lists its agents in its crawler documentation.
• Firewall and bot protection: some security settings block AI agents silently. Look for their visits in your server logs.
• Rendering: view the page source. If the main text only appears after JavaScript runs, many AI crawlers will not see it.
• Indexing: pages that are not indexed on Google and Bing are rarely found by assistants that search the web.
• Structured data: Organization, Article, FAQPage and Product markup that matches the visible content, tested with the Rich Results Test.

A free GEO audit checks most of these points on a URL in a minute.

Step 5: Audit the content

For each priority prompt, find the page of your site that should answer it, then check whether it can be cited:

• Does a direct answer appear in the first lines?
• Are headings phrased as questions, with each section understandable on its own?
• Are figures sourced and dated, and are comparisons shown in tables?
• Are the author and the update date visible?
• Is there a page at all? Missing pages are the most common finding.

The free AEO audit tool scores a page on these citability criteria.

Step 6: Audit the third-party sources

List the URLs cited in step 2 that do not belong to you. They show where the AI get their information on your market: comparison articles, review platforms, Reddit threads, directories, media, Wikipedia. For each, note whether your brand appears, how it is described and whether the information is current. This citation source audit often explains why a competitor is recommended and you are not.

What do you need to run the audit?

A spreadsheet with one row per prompt and assistant, accounts on the main assistants, access to Google Search Console and Bing Webmaster Tools, your server logs and a few hours. Record screenshots or copies of each answer with the date: they serve as the baseline for the next audit and settle any doubt when results change. Assign one person to the full audit so that scoring stays consistent.

The AI visibility audit grid

How do you turn the audit into an action plan?

Sort the gaps by impact and effort. Technical blockers come first, since they cancel everything else and are often quick to fix. Then create the missing pages for high-priority recommendation and comparison prompts. Then correct inaccurate facts on your site and profiles. Finally, plan the longer work on third-party sources. Re-run the same prompts one to three months later to measure progress, and keep the list stable so that results stay comparable. Our guide to building a GEO strategy turns these actions into a full plan.

An AI visibility audit shows on which questions your brand is missing from AI answers, who is cited instead and which technical, content or reputation gap explains it. To repeat the audit automatically, a tool like Sorank's AI visibility tracker asks your prompts in ChatGPT, Claude and Gemini (plus Perplexity and Grok from the Pro plan) and keeps the history, with a 3-day free trial.