I. The funnel has a new first chapter
There’s a moment in the buying process that marketers couldn’t reach for decades. The moment before the moment. The question someone asks themselves before they start searching. The inner consideration that precedes every click, every search query, every comparison.
That moment was always invisible. You couldn’t advertise on it. You couldn’t build a campaign around it. You could only try to influence it through brand advertising — broad, expensive, hard to measure.
AI models have made that moment visible. And simultaneously taken it over.
When someone asks ChatGPT or Gemini today “which bank is best for a mortgage in Belgium” or “which car insurance would you recommend for a young family,” something unusual happens. The model gives an answer. Not ten links to weigh yourself. Not sponsored results at the top. An answer. With a recommendation. With a reason. Sometimes with a name.
That user, if the model does its job well, is convinced before they reach your website. They already have a shortlist in their head. They already have a preference. You were in it or you weren’t.
That’s the new first chapter of the funnel. And most of the marketing industry isn’t writing it yet.
II. How AI is redrawing the awareness phase
The classic funnel starts with awareness. Someone becomes conscious of a brand, a product, a category. Traditionally you reach that phase via paid media: display, video, social, out-of-home. You buy visibility with people who aren’t yet searching but are receptive.
The model worked well as long as the orientation phase was fragmented. People stopped browsing, started googling, clicked on results, compared websites, arrived at a decision. Throughout that journey there were dozens of touchpoints where you could intervene.
That journey has become shorter. Much shorter.
A growing share of orientation now takes place in a single conversation, with a single interlocutor: an AI model. That model summarises, compares, weighs and advises — in a few sentences. The user doesn’t need to visit five websites. They ask one question and get a condensed answer that compresses weeks of comparison behaviour. In SEO circles, the term “zero-click search” has existed for a while — queries where the answer already appears in the search results and the user never clicks through. AI radicalises that phenomenon: the AI answer is the final destination, not the starting point of a search.
For paid media, this means something fundamental: you can keep putting money into the classic awareness funnel, but if your brand isn’t present in the AI answer that precedes the search behaviour, you’re reaching a user who has already decided. You’re advertising to someone who already has their shortlist. You’re paying for contact at the moment it’s already too late.
That’s not a hypothetical risk. It’s a shift that’s happening now.
III. The most efficient top-of-funnel there is — without a media budget
This is where it gets interesting for marketers used to thinking in cost per acquisition.
A user who reaches you via an AI answer is differently qualified than a user who clicks on a paid ad. They actively asked a question. They formulated an intent. The model answered their question and judged your brand as relevant. By the time they reach your website, the biggest hurdle has already been cleared: they know who you are, they have a reason to look further, and that reason didn’t come from an ad but from a recommendation.
And the cost? If your brand is correctly represented in AI answers — if you’re findable, if your entity is strong enough, if your content sends the right signals — there’s no cost per click, no bid strategy, no media budget. Organic AI presence isn’t bought — it’s earned: it’s the result of who you are in the information flow that models process. Free in media budget, not in effort. The investment lies in content, structure and consistency — not in buying reach.
That also makes it differently manageable. Paid media can be switched on and off. AI presence is built over time, by consistently making the right information available. It’s slower to build and harder to steer — but once built, it’s an asset that complements paid reach in a way paid reach can never do on its own.
A user who reaches you via a paid ad thinks: I see an ad. A user who reaches you via an AI answer thinks: this is a recommendation. That difference in perception has a name in the marketing literature. We call it credibility.
IV. Paid and AI visibility aren't competitors — they're communicating vessels
It would be a mistake to see AI visibility as an alternative to paid media. They serve different functions in the buying process, and the brands furthest ahead are the ones that treat them as complementary.
Paid media is still the fastest way to generate reach, activate new audiences, push seasonal offers or put price promotions into the market. For those functions, there is no better solution. That doesn’t change.
What changes is the question of what precedes paid media and what runs alongside it. If a user encounters your brand via an AI answer before they see your ad, that ad works differently: it confirms rather than introduces. It activates rather than convinces. The threshold is lower, the conversion higher.
The reverse also holds. Paid visibility increases brand awareness, which increases the chance that an AI model considers your brand relevant when making a recommendation. Brands that are more present online — in news, in reviews, in conversations — are better represented in the training data and in the live sources that models consult.
The two reinforce each other. But only if you manage them consciously together.
The marketer who optimises their paid strategy without knowing how their brand scores in AI answers is missing half the picture. They know how many impressions they’re buying, how many clicks they’re generating, how many conversions they’re achieving. But they don’t know how many decisions were already made before someone clicked on their ad — and whether those decisions went in their favour or against them.
V. The measurement problem — and how to approach it anyway
The honest caveat is that AI presence is harder to measure than paid media. There’s no impression count. There’s no click-through rate. An AI model doesn’t log who read its answer or what recommendation it made.
But that doesn’t mean there’s nothing to measure. You can’t see it in your dashboard. But you can see it happening.
What you can measure is presence: are you mentioned in answers to relevant prompts, and if so, how? That’s the foundation of LLM visibility monitoring — a methodology I described in earlier posts. You build a set of prompts that represent how users query your product category, you run them through multiple models, and you analyse whether your brand appears, how it’s positioned and what sentiment the answer carries.
That gives you three usable metrics: visibility (in the answer or not), citation rate (being referenced as a source) and sentiment (framed positively, neutrally or negatively). Not exact funnel attribution, but a direction. Is your brand present on the questions that matter? Is it described correctly? Is it associated with the right products and strengths?
Connecting to business outcomes is the next step, and one the industry as a whole is still developing. What already works: use AI visibility data as a leading indicator. A drop in presence or sentiment can precede a shift in brand preference that you’ll only see later in brand tracking or conversion figures. It’s an early signal in a world where most signals come too late.
VI. What you can do today
AI visibility is not a futuristic concept. It’s a shift that’s measurable now, strategically relevant now and requires action now. And it doesn’t have to start with complexity.
The first step is simply looking. What does an AI model say about your brand when someone asks about your product category? Test it yourself, in ChatGPT, in Gemini, in Perplexity — not once, but systematically, with prompts your target audience would realistically ask. What you find determines your starting point.
The second step is a mental shift in how you think about content. AI models don’t learn from slogans. They learn from informative, consistent, well-structured content that answers questions users actually ask. Every FAQ page, every product description, every article that answers a relevant question is a potential source a model can pick up and reproduce. Content as a knowledge source, not as campaign material — that’s the difference in mindset.
The third step is organisational: bring your paid and your AI strategy into the same conversation. Talk to your media agency or internal paid team about what AI models say about you. Those two conversations rarely happen in the same room right now. They should.
And finally: define what you want AI to say about you. Just as you have a brand message for campaigns, you need a desired AI profile. Which products should be associated with your brand, which strengths, which context? That desired profile is measured against reality — and the gap between the two is your work agenda.
The best ad you can make today is an AI answer that recommends your brand to someone who wasn’t yet looking for you but was looking for a solution to a problem you solve. That answer isn’t made in a media agency. It’s built by who you are in the information flow that AI models process.
