Local & Industries · 9 min read · July 15, 2026
The GEO audit: how to show customers their AI visibility in 30 minutes
The GEO audit is your fastest way to turn a vague gut feeling into a sellable argument: you ask the same AI systems your customers already use for recommendations in their industry and show whether the brand appears, how it's described and who gets named instead. Thirty minutes, one document, one aha moment for the customer.
Why GEO is no longer a niche topic for agencies
Your customers no longer just google, they ask ChatGPT, Perplexity and Google AI Overviews. And their own customers do the same. When someone asks "Which advertising agency in Munich is good for B2B software?", it's no longer the Google position on page one alone that decides, but whether the AI includes the brand in its answer. Exactly here your new consulting field arises: Generative Engine Optimization, GEO for short.
For you as an agency this is doubly relevant. First, you need your own AI visibility under control, otherwise you're recommending something you don't live yourself. Second, GEO is a product you can sell, while classic SEO has become a commodity for many customers. The GEO audit is your door opener: a concrete, visible result that stands in under an hour and immediately provides talking points for the next pitch.
The beauty of it: you need no enterprise tool and no data-science team. You need a structured question catalog, three or four AI systems and the discipline to document the result cleanly. The rest is craft, which you'll get step by step in this article.
What the GEO audit concretely measures
A GEO audit answers three core questions. First: does your customer's brand appear at all when you ask AI systems for providers in their industry? Second: how is it described, correctly, outdated, positively, or with wrong facts? Third: who gets named instead or in addition, that is, who are the AI-visible competitors taking the spotlight from your customer?
We call these three dimensions presence, portrayal and competitive environment. Presence is binary and brutal: named or not. Portrayal is qualitative and often the emotional lever, because customers are appalled when the AI describes a service discontinued in 2019 as a current core offering. Competitive environment is strategic, because it shows who you're actually up against in the AI world, and those are rarely the same names as in the Google ranking.
What matters is the honest expectation: an audit is a snapshot. AI answers fluctuate, they're not deterministic. So you don't measure one prompt, but a series, and you document date, model and wording. Only that way does a nice screenshot become a solid consulting artifact you can compare repeatably at the next audit.
The question catalog: your most important tool
The whole audit stands or falls with the prompts. A common beginner mistake: asking "Do you know the agency XY?". That's worthless, because you've already put the name in the AI's mouth. Correct are need-based questions from the end customer's perspective. Example for a trade business as your agency's customer: "I'm looking for a roofer in the Augsburg area for a renovation, which businesses can you recommend?" That way you measure real visibility.
Build eight to twelve prompts per customer in three categories. Category one: pure need prompts without brand names ("best providers for X in region Y"). Category two: comparison prompts ("provider A or provider B, which is better for Z?"). Category three: brand prompts with names ("What is known about the brand XY?"), to check the portrayal. This mix covers presence and reputation at the same time.
For marketing and advertising agencies that run the audit on their own customers from various industries, a reusable template is worthwhile. Build yourself a prompt library in which you only swap out industry, service and region as variables. That way you turn the 30-minute audit into a scalable process that even juniors on the team can run reliably, without quality suffering.
The 30 minutes step by step
Minute 0 to 5: preparation. Open ChatGPT, Perplexity, Google with AI Overviews and optionally Gemini or Copilot in separate tabs. For ChatGPT and Perplexity use a fresh chat or incognito if possible, so your history doesn't distort the result. Have your prompt template and an empty table ready with columns for prompt, system, result and note.
Minute 5 to 25: execution. Work through your question catalog systematically, every prompt in every system. Copy the relevant passages of the answer into your table, not the complete wall of text. Immediately mark whether the brand was named (yes/no), which competitors appear and whether the portrayal contains errors. Screenshots of the most important answers are gold, because customers believe visual proof more than your summary.
Minute 25 to 30: condensing. Count the hit rate (in what percentage of the need prompts was the brand named), list the three most common AI competitors and note the two most serious portrayal errors. These three numbers plus two screenshots are already a presentable result. Polish and layout come later, but the substance stands after half an hour.
Raw data becomes a sellable story
Raw prompt results convince no one in the meeting. Your job as an agency is the translation into a story with three acts. Act one: the status quo, usually sobering ("For seven out of ten need questions you weren't named"). Act two: the competitive picture ("Instead, the AI consistently recommends these three competitors"). Act three: the opportunity ("With targeted measures this gap can be closed within months").
The strongest moment in the customer conversation is almost always the wrong or outdated portrayal. When ChatGPT claims the customer doesn't offer a service that has long been their core product, or names an old company name, that's the emotional trigger for readiness to act. Record such findings especially cleanly, because they sell the follow-up project almost by themselves.
Package the result in a concise document: one page of management summary with the three metrics, then the proof screenshots, then a short outlook on possible measures. Resist the temptation to build a 40-page tome. The audit lives on clarity and pace, not on volume. The depth comes in the paid follow-up contract.
Common mistakes that devalue the audit
The first classic is the personalized history. If you test in your private ChatGPT account, where you've mentioned the customer's brand dozens of times, then you're measuring your own history, not reality. Use fresh sessions and document that. Otherwise you present an embellished visibility that collapses immediately when the customer checks it themselves, and that costs you credibility.
The second mistake is the single measurement. A single prompt on a single day is chance, not a finding. AI systems answer variably. Measure every core prompt at least two to three times and across several systems, then talk about tendencies instead of individual cases. Honesty at this point protects you when the customer sees a slightly diverging result on the next attempt.
The third mistake is confusing visibility and ranking. GEO isn't SEO with a new name. A top Google position guarantees no AI mention, and conversely, AI answers often show providers that stand further back on Google but have well-structured, citable content and strong third-party sources. Whoever mixes this up sells past the actual problem.
From free audit to paid retainer
The audit is deliberately meant as a door opener, but it must not end in an endless loop of free analyses. A proven model: the first audit is part of your acquisition and free or cheap. Everything after that, the derivation of measures, the implementation and above all the monthly monitoring, is paid work. That way you turn an aha moment into recurring revenue.
The monitoring is the core of the business model here. GEO isn't a project with an end date, but an ongoing process, because models, training data and competitors change constantly. A quarterly or monthly report showing the hit rate over time is a strong retainer argument. You make progress visible and give the customer a reason every month to keep working with you.
For you as an agency this is strategically valuable, because GEO is a young field with little competition and high willingness to pay. Whoever builds a clean, repeatable audit process now and packs it into their offers as a standard product positions themselves as a thought leader, exactly at the moment customers start asking the questions on their own.
Your next step: the first test run
Don't start with the customer, but with yourself. Run the 30-minute audit for your own agency. Ask the AI systems which advertising agency they recommend for your specialization and your region. The result is your first real data point, your practice run and at the same time an honest look in the mirror, which often turns out surprisingly uncomfortable.
After that you choose a well-disposed existing customer for the second run. Present the result to them not as a sales pitch, but as a gift: "We took a look at how you stand in AI systems, you should see this." This attitude opens doors, because you first deliver value and only then talk about cooperation. Most customers ask about the next steps on their own.
Keep your process in a living document: prompt template, table structure, report template. After three or four audits you have a routine that everyone on the team masters, and a product you can offer with a clear price. That way 30 minutes of curiosity turn into a new business area that matures exactly when the market needs it.
Common questions
Do I need paid AI subscriptions or special tools for a GEO audit?
For getting started, the freely accessible versions of ChatGPT, Perplexity and Google AI Overviews are entirely sufficient. Paid versions sometimes deliver more current web data and are helpful for consistent results, but not a must. Specialized GEO monitoring tools only become worthwhile once you scale the audit as an ongoing retainer for several customers and need automated tracking over time.
How do I explain to a customer the difference between GEO and the SEO we already do?
SEO optimizes for a website to stand high in a results list and get clicked. GEO ensures that an AI actively names the brand in its generated answer and describes it correctly, often without any click at all. Both are connected but not identical: a good Google position guarantees no AI mention. The audit makes this difference visible and tangible for the customer.
What if the customer is completely invisible in the AI systems?
That's not a problem, but your strongest selling argument. Complete invisibility shows the customer in black and white that action is needed, and positions you as the one who saw the gap. What matters is that you present it factually as a starting point and directly deliver a perspective: structured, citable content, strong third-party sources and consistent company data are the levers to build presence.
Read on