It is Monday morning, and your boss has just asked ChatGPT who the best provider is in your industry. Your name is missing from the list. Nobody quite says it is your fault, but at the next management meeting you will have to explain what you plan to do about it. Where do you start?
One measurement is no measurement
It is tempting to log in straight away and start asking a few test questions. Do not use your own logged-in account: it is shaped by previous activity and is not a reliable measuring instrument. Still want to run a quick check? Use a logged-out account or a temporary chat. Treat the answer only as an initial signal, not a firm conclusion.
Rules for a reliable baseline measurement
To understand where you really stand, you will often need a GEO specialist. Do not settle for quick assumptions about the cause. Ask for a methodical approach. A sound measurement always follows these rules:
- Neutral and unbranded: Test with questions that do not include your brand name, distributed across the different buying stages. That shows whether potential customers who do not yet know you encounter your brand at all.
- Repeat the test: AI answers fluctuate constantly, so a measurement should repeat each question at least five times. Only by looking at the variation can you see which results are genuinely stable.
- Keep platforms separate: Do not lump ChatGPT, Gemini and Perplexity results together or add them up. Always report the systems separately.
- Record everything: Log the AI system, market, language, date and any failed attempts for every measurement. Always require the raw answers, including citations and the definition used for “being mentioned”.
From data to an action list
Do not be dazzled by a polished dashboard. It is useful only if you can see exactly which competitor is taking the spotlight, on which AI search engine and with what content—and, just as importantly, what you need to do to catch up.
Do not jump to conclusions about why you are invisible in AI. Use transparent raw data to build an evidence-based action list for your content and technology teams, and clearly agree who owns each action. Later, run a follow-up measurement with exactly the same questions and under exactly the same conditions. It does not guarantee you will immediately overtake the competition, but you will have a strong, verifiable story for the next management meeting.
This translation was made by AI.