Guides
What does ChatGPT say about your brand?
Before buying, your customers no longer just type their question into a search engine: they ask an AI. And the AI answers — naming brands, comparing offers, giving an opinion. What ChatGPT says about your company influences buying decisions, most often without you knowing it.
Your customers already ask AIs
The shift is fast: Gartner predicts a 25% drop in traditional search engine volume by the end of 2026, in favour of AI assistants. Part of your prospects already ask ChatGPT, Gemini or Perplexity — "which solution should I choose for…", "is this company any good…", "who would you recommend for…".
The difference with a search engine matters: an AI does not return ten links, it gives one answer. If your brand is not in it, the prospect never discovers you. And if the AI says something wrong about you, there is no page two to catch them.
Where ChatGPT's answers come from
An AI like ChatGPT draws on two things: its training memory, built from what was published on the web up to a certain date, and, depending on the case, a live web search. Your brand therefore has two images: one frozen in the model's memory, and one alive, depending on what AIs find today when they search.
These two images can diverge. A company that pivoted two years ago may still be described with its old activity. A young brand may be unknown to the model but well covered in web search — or the other way round. That is why a single test, on a single AI, tells you very little.
How to check for yourself
The natural reflex is to ask "tell me about [my brand]". That is useful to spot factual errors, but it does not measure your real visibility: the real question is whether the AI names you when nobody whispers your name.
So ask the questions a customer who does not know you yet would ask: "what are the best solutions for [your field]?", "who do you recommend for [your product] in [your region]?". Test several AIs, not one — their answers differ far more than people expect. And repeat the test over time: answers move with every model update.
What to watch for
Four signals deserve your attention. Absence: the AI answers your customer's question naming three brands, and yours is not one of them. Competitors: who gets named instead of you, and why them. Factual errors: activity, prices, founders, dates — AIs sometimes invent details with unsettling confidence. And staleness: a description matching your website from three years ago signals that fresh sources are missing.
This verification can be done by hand. It quickly becomes repetitive: several AIs, several questions, several markets if you sell abroad, and tracking over time. That is exactly what our pre-diagnosis automates — free, with no account required.
Frequently asked questions
Does ChatGPT give everyone the same answer?
Not exactly. Answers vary with phrasing, language, conversation history and model version. That is why a serious measurement relies on several questions asked several times, not on a single test.
Can I correct what ChatGPT says about my brand?
Not directly: there is no form to edit an answer. However, AIs rely on what is published — your website, reviews, press, directories. By clarifying and updating those sources, the answers evolve. It is groundwork, measurable over time.
How often should I check?
AI models are updated several times a year, and their web-search answers change continuously. A quarterly check is a good cruising pace; monthly if your sector is competitive or if you have just corrected information.