There's something most companies haven't done yet: open ChatGPT and ask it, the way a potential customer would, whether their product or service is worth considering. It sounds obvious. But the results tend to be surprising — and rarely in a good way.
The issue usually isn't that the AI says something bad about you. The more common problem is that it says nothing at all. You're simply not there.
This guide walks you through a practical audit of your brand's AI presence — something you can do in under an hour, with no special tools. The goal isn't to panic at what you find. It's to understand where you actually stand.
Why AI invisibility is the quietest problem in marketing right now
When you don't rank on Google's first page, you at least know it. The data is public. You can see your position, the search volume, the click-through rate.
In LLMs, there's no public ranking. No position one or position ten. The AI synthesizes a response and cites — or doesn't cite — whoever it considers relevant. If you're not in that response, you never hear about it. No report lands in your inbox with that signal.
It's invisibility without notification.
And the volume of queries happening inside AI systems is growing. People consult ChatGPT before deciding which software to buy, which agency to hire, which product fits their needs. If your brand doesn't appear in those responses, you're missing a growing share of the consideration funnel.
The audit: step by step
Step 1 — Build the right list of questions
Before opening any AI tool, write down the questions your customers would actually ask. Not questions about your company specifically — questions framed around their problems or needs, the way a real buyer would phrase them.
Some examples:
- "What's the best [tool category] for [specific goal]?"
- "Which [type of company] should I use for [industry/need]?"
- "How do I solve [problem your customers face]?"
- "What are the alternatives to [well-known competitor]?"
Aim for at least 8–10 different questions. Vary the level of specificity — some broad, some close to the language your actual customers use.
Step 2 — Run the questions across ChatGPT, Claude, and Gemini
Ideally, test across all three major models. Each has different training data, different retrieval behavior, and different tendencies for which brands it surfaces. Your presence — or absence — can vary significantly between them.
For each question:
- Type it without mentioning your brand. See what the AI returns unprompted.
- Note which brands appear in the response.
- Record whether your company is cited, ignored, or — worse — mentioned with inaccurate information.
Do this in a fresh session or with chat history disabled, so previous conversations don't influence the results.
Step 3 — Ask directly about your brand
Once you've run the unprompted queries, go direct. Examples:
- "What is [your company name]?"
- "What does [your main product] do?"
- "Who are [your company]'s competitors?"
- "Is [your company] a good fit for [use case]?"
This surfaces a different dimension: when you are mentioned, what does the AI actually say? Are the facts correct? Is the description current? Does the positioning reflect what you want to communicate today?
Many companies discover at this step that the AI describes a product they've since repositioned, or cites a feature they deprecated over a year ago.
Step 4 — Record everything systematically
Don't trust your memory. As you run the tests, log:
- The exact prompt
- The model tested
- Whether your brand appeared (yes/no)
- Where in the response (first mention, middle, end)
- The tone (positive, neutral, critical)
- Any factually incorrect claims
A simple spreadsheet works fine. This is your baseline — you'll want to compare it against future tests to measure whether your efforts are having any effect.
Step 5 — Benchmark against competitors
Run the same prompts and observe who shows up consistently. Which brands in your category does the AI reliably mention? What do they have that you don't?
This helps you understand the selection pattern. LLMs tend to cite companies with strong presence in the sources they draw from — industry publications, independent review platforms, technical communities, open data repositories.
What the results typically look like
Scenario 1: you don't appear at all. The AI lists four or five competitors and your company isn't among them, even though you're a real player in the market. This usually signals low presence in the sources LLMs use — not enough editorial coverage, insufficient structured data, or content that doesn't answer the questions your buyers are actually asking.
Scenario 2: you appear, but with outdated information. The AI mentions you, but describes a product you've evolved, a positioning you've shifted, or data that no longer reflects your reality. This is a signal problem — your current sources aren't reaching the AI clearly enough to update the older impression.
Scenario 3: you appear, but weakly. The AI drops your name briefly at the end of a list, while competitors get paragraph-length descriptions. Nominal mention isn't the same as perceived authority.
Scenario 4: you appear well. Good — and don't stop paying attention. LLM behavior shifts with new model versions, new training data, and new competitors who start investing in GEO.
Where to go from here
The audit is step one. From there, the most common actions are:
Building presence in sources that LLMs actually draw from — sector publications, independent comparisons, editorial coverage (not paid placement). Creating content that directly answers the questions you just tested, with clear language and structure that makes extraction easy for AI systems. Ensuring that basic facts about your company are accurate and up to date in the right places — open databases, product pages, and public descriptions.
None of this produces instant results. LLMs don't index the way Google does. But without the audit, you're working blind — patching holes you haven't located yet.
Genoma was built to make this process systematic: tracking which prompts mention your brand, how often, in which model, with what tone — and benchmarking that against your competitors over time. If you want to move beyond the manual audit, explore the platform.