Brand AI visibility audit checklist and step-by-step checks

A brand AI visibility audit starts by entering key questions directly into four engines (ChatGPT, Perplexity, Gemini, and Claude) and recording whether the brand is mentioned in the answers. Next, you check four areas in order: crawler accessibility, content format, competitive comparison, and metric measurement.

Consumers ask ChatGPT or Perplexity directly instead of scanning Google search results, and sometimes choose the brand the AI recommended. The way AI cites and recommends brands works separately from SEO rankings, so ranking at the top of search and appearing in AI answers do not call for the same optimization methods.

Check 1. Selecting key queries and testing each engine directly

The first step is to choose queries, enter them into each engine, and record the results. Even for a single question, the brands each engine cites do not match, so instead of measuring everything at once as you would with SEO rankings, you test each engine separately. Choosing the wrong queries distorts every measurement that follows, which is why this step takes the most time.

Query selection criteria

• Information-seeking questions that ask directly about the brand's category (e.g., "vegan sunscreen recommendations," "comparison of soothing serum ingredients")
• Comparison and recommendation questions that people would actually enter when deciding on a purchase
• Build 20 to 30 queries in each group: category queries without the brand name and queries that include it

How to test each engine

Enter the same queries into ChatGPT, Perplexity, Gemini, and Claude, and record the items below in a spreadsheet. Giving each engine its own column lets you compare the differences between engines.

ChatGPT uses its training data together with ChatGPT search results, while Perplexity answers based on real-time web indexing, so the two engines sometimes give different results for the same query. If a brand is missing from both engines, check content format and crawler accessibility first.

Check 2. Why check AI crawler access?

If AI crawlers cannot access a site, its content is not used as material for AI answers. According to OpenAI's crawler documentation, OAI-SearchBot for search and GPTBot for training are set separately. Sites that block OAI-SearchBot are not shown in ChatGPT search answers, and blocking GPTBot signals that the site's content should not be used for model training. Perplexity uses PerplexityBot to collect the pages it shows in search results. Google's Google-Extended is a token that decides whether content may be used for Gemini training and grounding, and it does not affect visibility in Google Search, so check the allow setting for each bot separately.

Crawler accessibility checklist

• robots.txt is not blocking OAI-SearchBot or PerplexityBot, and the GPTBot and Google-Extended settings match your intent
• Key landing pages, blog posts, and product detail pages can be accessed without logging in
• Pages that rely on JavaScript rendering provide server-side rendering or static HTML (when a page relies on rendering, crawlers sometimes cannot fetch the main text)
• An up-to-date sitemap has been submitted

In robots.txt, adding User-agent: GPTBot / Allow: / allows OpenAI's training crawler to collect the site; visibility in ChatGPT search is set separately with the OAI-SearchBot setting.

Check 3. What content formats does AI cite easily?

AI engines cite structured content more easily. Pages with short paragraphs that answer the question right away and question-style headings are excerpted in AI answers more often than pages where information is scattered across long paragraphs. Applying the schema.org FAQPage, Product, and Organization schemas helps AI engines recognize brand information as a structured entity. Google AI Overviews does not require any special schema and recommends that structured data match the visible text on the page.

Content format checks

• Is the answer to each key question within the first 1–2 sentences of its section?
• Are the H2 and H3 headings written as actual search questions? (e.g., "What effect does the XX ingredient have on the skin?")
• Is the FAQPage JSON-LD schema applied?
• Does the Organization schema include the brand name, category, and official URL?
• Is information broken into bulleted and numbered lists? (instead of long paragraphs)

Check 4. Comparing citation share with competing brands

You can only tell where your citation share stands in the category if you also record how often, and in what position, competing brands are mentioned for the same queries. If you record only whether your own brand was mentioned, you have nothing to compare it with.

How to run the competitive comparison

• Record every brand mentioned in the AI answers to the queries you chose in Check 1.
• Tally each brand's mention count and the number of times it was mentioned first, and calculate citation share within the category.
• Formula: brand mention count ÷ number of target queries × 100%
• Collect the content types where competing brands are cited (blogs, reviews, official sites) to find the channels you are missing.

If your citation share is at least 10 percentage points lower than a competitor's, check content format and crawler accessibility first. If neither has a problem and the gap is still large, switch to a strategy of increasing mentions in authoritative third-party media (beauty media, specialist review sites).

Check 5. Metrics for tracking AI visibility performance

An audit is not a one-off: track the same metrics at least once a month to measure the effect of your improvements. The AI search environment changes quickly with engine updates and changes in competing brands' content.

After you fix the content format, it usually takes 6–12 weeks or more for changes in citation share to become clear. So evaluate performance by the three-month cumulative trend rather than by short-term swings in the numbers.

PION's AI visibility audit combines the five checks above into one report, providing citation share for each of the four standard engines (ChatGPT, Gemini, Google AI Overviews, and AI Mode), crawler accessibility errors, and schema improvement priorities. If you want to use another provider's audit report in place of this checklist, first check whether the report includes the prompt list and the number of measurements. You can find the types of free audit reports and what to check in Free AI visibility audit reports from GEO agencies: What to check before requesting one.

To gauge how ready you are now before working through the checklist yourself, you can check eight items in the AI answer visibility basics check.