LLMs will not recommend the most visible brands but the most clear and credible. The real question is not how much to invest to appear, but whether your brand is worth choosing
Marketing departments are probably making the same mistake with LLMs as with Google. For years, many wanted to believe that being well referenced was enough to be chosen, or even to become a brand powerful. It was a mistake. Word of a child’s performance.
This widely disseminated belief, supported by measurable results – but what were we actually measuring? — had a very clear influence on the shift in communication budgets.
Yes, it was a mistake.
Google has never magically made a brand powerful. Above all, it amplifies the resonance of those who already had everything to succeed: a clear proposition, a useful offer, an established reputation, available evidence, reassuring experience, a latent or already existing demand.
Google does not create demand. He captures it at the moment when it becomes explicit.
Search players know it well: there is no lasting performance without a powerful brand. There are, at best, glass ceilings, a few moves to be made, rarely a lasting dynamic.
Today, with the generative artificial intelligencethe same reflexes return. We’re talking about GEO, optimization for ChatGPT, presence in Gemini or Claude. We are already trying to understand the new algorithm, to produce the right content, to structure the right data, to occupy the new space.
All this will be useful. But that won’t be enough.
Because the real question is not: how to be visible in LLMs? The real question is much more disturbing: why should an AI recommend your brand over another?
This question changes everything.
To be cited is not to be chosen. Being visible is not being credible. For years, digital gave marketing an illusion of control because it made it possible to measure everything: clicks, impressions, conversions, cost per lead,ROAS,attribution.
This culture of measurement has had its virtues. She professionalized the execution. It forced the teams to pilot, test, referee.
But it has also produced a major blind spot: believing that what is measured best necessarily creates the most value.
However, the power of a brand does not only come into play in what is clearly shown on a dashboard. It is built in memory, repetition, trust, experience, conversations, comparisons, proofs that others agree to relay.
It is built in this rare ability to become an obvious option before the customer even begins to compare.
This logic is not entirely new. Google had already started to value experience, expertise, authority and trust with EEAT. LLMs extend this movement, but with a major difference: they no longer just classify the answers, they formulate them.
This is where the break is deep.
Google offered lists. LLMs produce recommendations. In a search enginea brand could still exist by simply being well placed. In a generated response, the space shrinks. Three brands will be cited. Sometimes two. Sometimes just one.
The others will not exist.
This is exactly what weak brands have experienced in the world of mass distribution. Due to lack of power and therefore space, they disappeared from the shelves with the arrival of distributor brands.
This reality should worry many marketing departments. Not because they are behind AI, but because AI risks making suddenly visible what their marketing has sometimes been hiding for years: vague positioning, interchangeable discourse, weak evidence, an insufficiently constructed reputation, a difference that no one really knows how to formulate.
LLMs will not correct this. They will expose it.
This is why it would be dangerous to reduce this subject to a new technical discipline. GEO must not become the new surface SEO, with its recipes, its hacks, its tracking tables and its promises of visibility.
Of course, it will be necessary to understand how the models read, select and render information. Of course, you will have to structure your content, strengthen your signals, work on your presence in the sources that matter.
But this work will only be effective if the brand is already worth retaining.
This is bad news for weak brands. This is great news for real marketers.
Because LLMs can become a formidable tool of truth. They allow you to test what the market understands about you, to identify the blind spots in your positioning and to understand what you are missing to become, or remain, a strong brand.
A strong brand reduces doubt. It makes the choice easier. It improves conversion. It increases the effectiveness of media investments. It better protects the margins. It allows it to be researched, cited, compared, recommended.
She gives to algorithmscustomers, journalists, experts and communities with good reasons to make it exist.
Conversely, a weak brand pays more for everything. She must buy the attention she failed to earn. She must repeat what no one remembers. She must push what no one is asking for. It must continually optimize because it has failed to become obvious.
Also, let’s not get the focus wrong. Our role is not just to appear in AI responses. It is about building brands that are strong, clear and credible enough to deserve to be recommended.
The decisive question is therefore not how much to invest to appear in the answers generated by LLMs. It is whether your brand gives enough reasons to be recommended.
It’s less comfortable than a GEO plan. But it is infinitely more strategic.
