There is a question we ask every CMO in our first meeting.
We open ChatGPT, type the name of their category "best CRM for medium businesses", "recommended health insurance", "e-commerce platform for retail" — and wait.
In 80% of cases, the brand does not appear.
Not in the first mention. Not in the top three. In many cases, it does not appear at all.
And that is not a technical problem. It is a business problem.
The channel that no one is measuring
ChatGPT surpasses 800 million weekly users. Gemini, Perplexity, and Claude add hundreds of millions more. Together, they are redefining how people make purchasing decisions: they no longer search for result lists to compare — they ask AI and trust its response.
The problem is that most brands are still optimizing for the old model.
Traditional SEO: ranking on the Google results page. Getting the click. Bringing traffic to the site.
That model worked for twenty years. Today it is still relevant. But it is no longer enough.
Because when someone asks ChatGPT "what is the best tool for X?", there is no results page. There is no click. There is an answer. One only. And in that answer, there are brands that exist and brands that do not exist.
The question that every marketing director should ask themselves this week is simple: which of those two categories is my brand in?
Why traditional SEO is not enough
You can have the most optimized site in the market, dominate the first position on Google for your main keywords, and still be completely invisible to AI engines.
Why? Because LLMs do not work like Google.
Google indexes pages and ranks them by relevance and link authority. Language models do something different: they learn patterns from huge volumes of text, build an internal representation of which brands exist, what they do, and in what context they are reliable — and when someone asks a question, they generate an answer based on that representation.
If your brand is not well represented in the data corpus used to train those models, it does not matter how much you have invested in SEO. For AI, you simply do not exist.
Worse yet: if you are misrepresented — described as "a budget alternative to X" when you are the segment leader, or associated with an incorrect ICP — AI will recommend you poorly. It will send the wrong leads. It will erode your positioning every time someone asks about your category.
This is not a hypothesis. It is what we measure every day in over 500 brands in Latin America.
Three concepts that change the way you think about marketing
There are three emerging disciplines that the most sophisticated marketing teams are already incorporating into their strategy. They are not buzzwords. They are concrete answers to the problem we just described.
AEO — Answer Engine Optimization
If SEO was the discipline of ranking in search results, AEO is the discipline of ranking in AI engines' answers. It is not about keywords, but about semantic clarity: structuring your content so that models can extract, understand, and cite it when someone asks a relevant question for your category.
The fundamental principle of AEO is simple: AI prefers clear sources. If your content is ambiguous, outdated, or poorly structured, the model will prefer to cite your competitor who has more accessible information.
GEO — Generative Engine Optimization
GEO works on the authority layer. It is not enough for your own content to be well-structured: AI models prioritize brands that have consistent mentions in high-credibility sources — specialized media, review platforms, communities like Reddit or Quora, content produced by opinion leaders.
An analysis by Ahrefs found a correlation of 0.7 between brand mentions in authoritative media and visibility in AI systems. Translated: media coverage is not just PR — it is infrastructure for visibility in AI.
LLMO — Large Language Model Optimization
LLMO operates at the deepest layer: how the model interprets you internally. The difference between ChatGPT describing you as "an alternative to X" or as "the leader in Y" is not a matter of luck. It is the result of how your entity is built in the model's knowledge.
Brands that do LLMO well not only appear more — they appear better. In the right context. For the right buyer. With the positioning they chose, not with the one AI assigned them by default.
The problem that no one can see (yet)
There is something that makes this problem especially difficult to manage: it is inherently invisible.
Search traffic has dashboards. Paid campaigns have real-time metrics. AI visibility, until very recently, was a black box.
How many times in the last month did someone ask ChatGPT about your category? In what percentage of those queries did you appear? In what position? With what description? Does your main competitor appear before you?
Most marketing teams do not have answers to any of these questions.
And the cost of that ignorance is real. It is estimated that between 15% and 64% of traditional organic traffic is being lost due to the migration to conversational searches with AI. Traffic coming from AI converts 9 times better than traditional search traffic — because it arrives after an explicit recommendation, not after a click.
The good news is that this can be measured. And what is measured can be optimized.
Four signs that your brand has an AI visibility problem
Before diving into solutions, it is helpful to identify if the problem exists. These are the most common signs we find in our audit:
Competitor anchoring. AI consistently describes you as "an alternative to [competitor]" instead of positioning you as a leader in your own category. This happens when the competitor has more entity clarity and the model anchors its description of you in relation to them.
Category misclassification. The model places you in a category that is too broad or directly incorrect. Result: you never appear when someone makes a specific query about your niche.
Outdated description. Models have data inertia. If you changed your product, your prices, your value proposition six months ago, it is likely that AI is still describing you with old information.
ICP inaccuracy. The model recommends your enterprise solution to small businesses, or your consumer product to corporate buyers. Each incorrect recommendation is a lead you will not close — and that will confuse whoever receives that response.
Low entity confidence. When the model does not have enough verifiable information about a brand, it prefers to omit it rather than risk giving incorrect information. If your brand appears little, it may be that the model simply does not trust what it knows about you.
Where to start: the 90-day framework
You do not need to solve everything at once. What you need is to start with clarity.
The framework we developed at Fardo has four phases, designed to go from measurement to continuous mastery in 90 days.
Phase 1 — Full X-Ray (weeks 1-2) The starting point is a deep audit in the main engines: ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews. The output is an AI Visibility Score — a number between 0 and 100 that consolidates mention frequency, context, position, authority of the sources that cite you, and breadth of thematic coverage. And a gap analysis against your direct competition.
You cannot improve what you do not measure. This phase gives you the baseline.
Phase 2 — Strategic Optimization (weeks 3-6) With the clear diagnosis, the second phase works on the root problems. Technical adjustment of schema markup. Correction of the structured information that models read about you. And the creation of a page /llm-info/ — a public and crawlable page, specifically designed to be read by language models, that accurately establishes what your brand is, for whom, and how it differs from the most common comparisons.
Phase 3 — Tactical Positioning (weeks 7-10) The third phase builds external authority. Q&A format content designed to be cited by response engines. Presence on platforms where LLMs look for consensus signals — Reddit, G2, LinkedIn, YouTube. Coverage in high-authority media that act as credibility validators for the models.
40% of citations in ChatGPT and Perplexity come from Reddit. That data alone justifies a community strategy that most marketing teams do not have.
Phase 4 — Continuous Monitoring AI visibility is not a project, it is a channel. Models are constantly updated. Narratives change. Competitors move. The last phase establishes real-time monitoring — alerts, biweekly analysis of the Score, and a continuous optimization process that ensures what you built in the first three phases does not erode over time.
What is at stake
There is a time window that is closing.
Today, most brands in Latin America are still not investing in AI visibility. This means that brands that start now have the opportunity to establish their positioning in the models before the space fills up with competitors using the same strategy.
LLMs are not like Google: they do not have positions 1 to 10 for each keyword. They are more like a trusted advisor with formed opinions. And changing a model's opinion — once it is formed — is harder and takes longer than building it from scratch.
The brand that is the answer today has a compounded advantage that accumulates over time.
The one that waits for this to become an industry priority will arrive late to a space that its competitors have already occupied.
