In short: You can check what AI says about your business in five minutes, and you should. But do it right: Logged out, with the questions your customers actually ask, in several models and with several phrasings. And know what you are looking at: A spot check, not a measurement. The spot check reveals whether you have a problem. It cannot tell you why, how big it is, or whether it is improving. Here is the full recipe, the traps, and the next step.
Why should you test it yourself?
Because the moment you see it with your own eyes changes your priorities. More and more buying decisions start with a question to ChatGPT, Gemini or Copilot instead of a Google search, and in those answers some businesses get recommended while others are never mentioned. Seeing where you stand is the fastest way to understand what AI visibility means for your business specifically.
How to do the spot check properly
- Log out first. Use an incognito window or a logged out session. Logged in services remember previous conversations, and since you ask about your own business far more often than your customers do, your own answers are biased in your favor.
- Ask like a customer, not an owner. Customers do not type your brand name, they describe their need: «Which accounting firm fits a small business?», «best heat pump supplier in my region», «recommend a tool for AI visibility». Ask the way they ask.
- Vary the phrasing. Put the same need three to five different ways. AI answers are not deterministic, and small changes in the question can produce entirely different recommendations.
- Test several models. ChatGPT, Gemini, Copilot and Perplexity have different training data and sources. One model recommending you says nothing about the others.
- Note three things: Are you mentioned? What is said about you, and is it accurate? And who gets recommended instead of you?
Why is the spot check not enough?
Because a single query is just one random sample of what the model answers. If you get recommended, it may be luck. If you are missing, it may be bad luck. To know where you actually stand, you need to ask many times, with different phrasings, across several models, and repeat it over time. Three traps make the single query unreliable:
- The models do not answer the same way twice. When we asked ChatGPT and Google AI 1,000 questions about restaurant recommendations in Oslo, the models agreed on only five of ten places in their top lists. Read the test here. A single query is a dice roll, and you cannot tell whether you hit the rule or the exception.
- You do not see why. The spot check shows the result, not the sources the model built it on. Without knowing which websites, reviews and directories drive the answer, you do not know what to change.
- You have no trend. A snapshot cannot tell you whether your visibility is rising or falling, or whether your actions work. That requires repeated measurement against a baseline.
What does real measurement require?
The same as the spot check, just systematic: Hundreds of questions phrased the way your customers ask them, run across several models, repeated week after week, with tracking of which sources get cited and how your competitors score on the same questions. That gives you a visibility score you can compare over time, and a list of the sources that actually decide the answers. The full KPI setup is in our guide to measuring AI visibility.
That is the job the Synlig.ai platform does automatically: Continuous measurement across the models, source tracking and competitor comparison, in one dashboard.
What do you do with what you found?
Did the spot check and did not like the answer? Then you already know more than most of your competitors. The next step is finding out why, and that is measurable. Book a demo and we will run a full measurement of your visibility, show which sources drive today's answers, and what it takes to get recommended more often. Or start a free trial and see the numbers yourself.