GEO Audit in 1 Hour: Check Your AI Visibility Yourself

In about an hour, you can check for yourself whether AI assistants know your brand, represent it correctly and recommend it. The short version: draft five core questions your buyers actually ask, feed them into ChatGPT, Gemini and Perplexity, and note verbatim whether and how you are named. Then check who does get cited and whether your own pages are technically readable for AI bots. You finish with three quick wins and a priority. No tools, no subscription, just you and a clock.

This self-audit is a baseline, not a final report. It does not tell you everything, but it tells you the most important thing: are you on the shortlist that AI assembles, or not? That is exactly the question classic SEO reports do not answer. If you want to understand the full framework first, read the GEO guide. If you just want to know where you stand today, set a timer and follow the six steps below.

Below you see how those sixty minutes break down: from drafting your core questions to prioritizing your findings, each block with a fixed duration so you finish within the hour.

What you need

Little. A laptop, access to ChatGPT, Gemini and Perplexity (the free versions are enough), a blank document or spreadsheet to note everything in, and 60 minutes without interruptions. Have your screenshot tool ready. Work as much as possible in a private window or logged out, so your own search history does not color the answers. That way you test what a real prospect sees, not what the AI thinks you want to hear.

Step 1 (10 min): draft your core questions

Do not start with “do you know company X?”. That is too easy and says nothing about commercial visibility. Instead, write down five prompts your ideal customer would actually type at the moment they are looking for a supplier. Think at the level of the category, not your brand.

Concretely, for a B2B service provider:

• “What is the best [type of service] agency in [region]?”
• “Which suppliers of [product] do you recommend for a mid-sized company?”
• “I’m looking for a partner for [problem]. What are the options?”
• “Compare the best-known providers of [category].”
• “By what criteria do I recognize a good [type of supplier]?”

Common mistake: entering only your own brand name. Then you measure recognition, not recommendation. The real gain sits in the category questions, because that is where the shortlist is decided. Feel free to add one brand question as a control, but put the weight on the four commercial prompts.

Step 2 (15 min): test your brand in ChatGPT, Gemini and Perplexity

Feed each prompt into the three assistants and note three things per answer: are you named (yes/no), in what context (recommended, in passing, or as a footnote), and which competitors are at the top. Take a screenshot of every answer. This is your reference point for later.

Why three models? Because they weigh different sources. ChatGPT leans heavily on its training data plus web results, Gemini draws differently from its index, and Perplexity explicitly shows the sources it cites, which is worth gold for this audit. Anyone missing from all three has a fundamental visibility problem. Anyone who appears in one but not in two knows immediately where the shoe pinches.

Common mistake: testing once and concluding. AI answers vary by phrasing. Ask the same question two or three times, worded slightly differently. If you never see your brand, it is no coincidence. If you sometimes do, your visibility is shaky and therefore steerable.

Step 3 (10 min): check your source position and citability

Now it is all about the why. Look in Perplexity at which sources it cites for your category questions. Are they comparison articles, review sites like G2 or Capterra, industry blogs, or competitors’ own pages? That list tells you exactly which places AI trusts in your market. If you are on none of them, you know where to build authority.

Click through to one or two cited pages and look at why they are citable: a clear answer at the top, figures, a clear definition, a well-defined topic. Compare that honestly with your own relevant page. Often the difference is not your product, but the form: AI cites content that is easy to extract. You can read more about writing citation-worthy content for AI.

Common mistake: looking only at your own site. GEO is largely won or lost off-site, on the sources AI already trusts. So note not only “our page is not good enough”, but also “we are on none of the sources cited in our field”.

Step 4 (10 min): check your technical crawlability for AI

An AI model can only cite you if it reaches your content and can read it. Two quick checks. One: open your most important page, right-click and choose “view page source”. Do you see your core texts in the plain HTML? Good. Do you see mostly empty containers and scripts? Then your content is only built after loading and many bots miss it. That problem is called client-side rendering and it makes you invisible to AI.

Two: type yourdomain.com/robots.txt in the address bar and check whether you block AI bots like GPTBot, ClaudeBot or PerplexityBot. Many companies do this unknowingly and lock themselves out this way. You will find more context on blocking AI bots in robots.txt.

Common mistake: assuming that “we rank in Google” equals “AI sees us”. Google sometimes indexes what AI crawlers do not fetch, and a block in robots.txt or a heavy JavaScript framework can keep you perfectly visible in the classic search engine while you disappear from AI answers.

Step 5 (10 min): note 3 quick wins

Look back at your notes and filter for what you can fix yourself this week. Deliberately limit yourself to three. Examples of real quick wins from an audit like this:

• Unblock an AI bot that was accidentally in robots.txt.
• Place a clear two-sentence answer at the top of your most important service page, instead of a woolly intro.
• Make your company details (name, service, region) consistent across your site, LinkedIn and directories, so AI recognizes you as one consistent entity.

Common mistake: making a list of twenty actions you never carry out. Three finished quick wins beat twenty open intentions. The deeper structural work (building source authority, rolling out schema) you note separately as “later”, not as this week.

Step 6 (5 min): prioritize

Rank your three quick wins by impact against effort. Start with what directly affects your visibility and takes little work. An unblocked bot or a rewritten opening paragraph can have an effect within days. Add a date and plan one repeat of this audit in three months, with exactly the same five prompts. That way you see movement instead of gut feeling.

A worked example

Suppose: an IT service provider from Ghent does the audit. On “best managed IT partner for SMEs in Flanders”, he is named in none of the three models. Perplexity cites two comparison blogs and one industry portal. In the page source, his services page is largely empty (React app), and robots.txt blocks nothing. His three quick wins: have the core texts rendered server-side on the services page, put an answer paragraph at the top, and contact the industry portal’s editors for a mention. Priority one: the rendering issue, because without readable content the rest is pointless. Within a quarter, he surfaces in Perplexity for two of the five prompts. That is the pattern we also saw when we brought a client to the first page of Google and kept them there: measure, adjust deliberately, repeat.

Common mistakes in brief

• Testing only your brand name instead of the category questions buyers ask.
• Asking once and concluding, while AI answers vary by phrasing.
• Looking only at your own site and ignoring the off-site sources AI trusts.
• Thinking that ranking in Google automatically means visibility in AI.
• Noting twenty actions and finishing none.

Checklist: GEO audit in 1 hour

• Five core prompts drafted (four category, one brand as control)
• Each prompt tested in ChatGPT, Gemini and Perplexity, with screenshots
• Named yes/no, context and competitors noted per answer
• Cited sources in Perplexity mapped
• Page source checked: are core texts in the plain HTML?
• robots.txt checked for blocked AI bots
• Three quick wins noted
• Prioritized by impact against effort, with a date
• Repeat scheduled in three months with the same prompts

From self-audit to structural visibility

This one-hour audit gives you an honest baseline, but it deliberately does not cover everything. If you want to go deeper, fully work out the four layers of crawlability, entities, citations and schema in a thorough AI visibility audit, or use a checkable 25-point GEO checklist. And if you want to know which signals to track structurally, read the 5 core indicators of AI visibility and how to measure AI visibility without expensive tools.

The reason to do this well is twofold. You not only improve your own findability, you also become the source AI cites when others ask your category questions. That is precisely the heart of GEO for B2B: not tricks, but making sure models know, understand and recommend your brand correctly. If you do not want to keep tracking this yourself, we take it over as a GEO agency: we measure, adjust and keep you on the shortlist. Ready to turn your baseline into growth? Get in touch and we will walk through your results together.