Guides
Which sources AIs use to talk about your brand
When an AI talks about your brand, it does not invent it from scratch: it assembles what it has read about you. Understanding where that information comes from means understanding why the AI says what it says — and, above all, what to act on to change the answer.
The AI only knows what is published
An AI has no opinion about your company: it reflects what exists about you, filtered by what it memorised during training and, sometimes, by a live web search. If a piece of information is written clearly nowhere, the AI cannot know it — and if wrong information circulates, it may repeat it.
That is good news: you do not have to convince the AI, but to look after what is published. Answers follow the sources.
Your website: the reference source
Your website is the first source AIs consult to understand what you do. Clear pages — activity, offer, target audience, contact details — give the AI a reliable base. A vague, outdated site with no sharp description leaves the AI to guess, often badly.
Beware of stale information: an offer you no longer sell, an old positioning, a name that has changed. As long as those elements stay online, they keep feeding the answers.
Third-party sources: what others say about you
AIs give weight to what is said elsewhere than on your own site: customer reviews, press articles, professional directories, comparisons, sector pages. A brand picked up by several independent, consistent sources looks more reliable and gets cited more easily.
That is also where errors creep in: a directory with an old address, a review that confuses two companies, a listing never updated. These sources are outside your site but weigh on the image the AI reconstructs.
The references that carry authority
Some sources weigh more than others because they are structured and widely reused: encyclopaedic entries, company databases, official registries. Appearing in them, with accurate information, helps the AI identify your brand unambiguously and not confuse it with another.
Not every brand belongs there, and it is not about forcing your way in. But when these references exist and are correct, they clearly stabilise what AIs say about you.
How to find out what AIs retain
The simplest way is to ask several AIs to describe your brand, then compare with reality: what is right, wrong, outdated or missing? Each gap points to a source to fix or create. Ask too, where possible, what the AI relies on: some cite their sources.
Done by hand, across several AIs and markets, this inventory is tedious and hard to track over time. Our free pre-diagnosis does it for you: it shows what AIs say about your brand and highlights what to fix — with no account.
Frequently asked questions
Can I ask an AI to cite its sources?
Often yes, especially for AIs that do a live web search: they can list the pages used. Answers coming from the model's memory alone do not point to a specific source — one more reason to cross-check several AIs.
Is Wikipedia essential?
No, and it is not open to every brand. It is a useful source because it is widely reused, but it is not the only one: a clear website, consistent reviews and a few reliable third-party mentions are often enough to be well identified.
Can wrong information on another site affect my answers?
Yes. AIs aggregate what they find: a wrong third-party source can end up in the answer. Hence the value of spotting those sources and getting them corrected where possible.
How long before a correction shows up?
For AIs that search live, a few days to a few weeks after the source is updated and picked up elsewhere. For the models' memory, you have to wait for a new version. That is why you measure over time rather than once.