There is a question that more and more managers are asking themselves in private, sometimes with a touch of unease: why, when I ask ChatGPT about my sector, does my company not come up?
This is no trivial question. And the answer has nothing to do with technology, or budget, or how much content you’ve published in recent years. It has to do with something more structural: how you’re interpreted by the systems that now decide whom to cite when someone searches for a reference.
That is the subject this post aims to explain. No jargon. No gimmicks. Just the framework you need to understand what is changing and why it matters to anyone making decisions about their organisation’s positioning.
The search engine you used to know isn’t the only game in town anymore
For years, appearing on Google meant one specific thing: being among the top results when someone entered a search query. The mechanism was well known, though not always fully understood: post regularly, get others to link to your website, and use the right keywords in the right places. A game of visibility based on volume and technical signals.
That game is still around. But a new one has come along, and it works differently.
When someone asks ChatGPT which strategic communications agency has the best reputation in Spain, or asks Perplexity who the leading figures are in corporate crisis management, or when Google displays an AI-generated summary before the usual results, the mechanism that decides what appears is not the same as the one that determined what came up first in a traditional search. If you want to understand the scope of that change, the future of artificial intelligence in marketing provides a useful context.
Artificial intelligence systems do not return a list of web pages ranked by relevance. Instead, they generate a response. And to do so, they decide which sources to trust.
That decision marks a new starting point.
How does an AI decide who to trust?
Here is the change that deserves the most attention: an AI system does not read your website in the same way as a traditional search engine did.
A traditional search engine used to crawl web pages, count signals and rank them. It was, in essence, a ranking system. A generative AI system does something different: it tries to understand what an organisation knows, what it talks about coherently, and from what perspective it does so. It doesn’t rank web pages. It builds a picture of you.
And that image depends on whether you can be interpreted as a coherent entity.
What does ‘interpretable’ mean? It means that the system can clearly answer three questions about you:
What is this organisation specialising in? Not just anything. Something specific and recognisable. An organisation that speaks coherently about strategic communication creates a clear mental framework. An organisation that talks about marketing, human resources, innovation, sustainability and digital transformation in equal measure creates none.
Where is this coming from? Not just what is said, but the stance taken. A source with its own perspective, which takes clear-cut positions and stands by them over time, is easier to interpret than a source that merely echoes market sentiment without offering any perspective. It is important here to understand how AI systems determine whether content is original or simply generated.
Is it consistent over time? A brilliant article does not establish authority with an AI system. What builds authority is when that approach is reinforced across multiple pieces, in different formats, over a sustained period. Systems interpret patterns, not flashes.
If an AI system can answer those three questions about your organisation effectively, you stand a good chance of being cited as a reference. If it can’t, it doesn’t matter how much you’ve published.
Why posting a lot doesn’t solve the problem
This is probably the most counterintuitive aspect of the new landscape, and also the most important one for content decision-makers.
The usual instinct when faced with the question ‘How can we increase our visibility?’ is usually the same: publish more. More articles, more posts, a greater social media presence, and more frequent updates. The reasoning behind this is clear: more content means greater visibility.
But when it comes to AI systems, that logic simply doesn’t hold water. In some cases, it works the other way round.
When an organisation publishes extensively on a wide range of topics, without a clear hierarchy or a recognisable stance that is reinforced piece by piece, its systems fail to establish authority. Instead, they breed confusion. People cannot grasp what that organisation knows better than anyone else, nor from what perspective it speaks. And a confusing organisation is not a source to be trusted.
The difference isn’t in the volume. It’s in the structural coherence. As we discussed in detail in ‘Posting a lot doesn’t build authority’, the most common mistake isn’t posting poorly, but posting without a system in place to reinforce your content.
An organisation comprising twenty well-connected articles that develop a coherent perspective on a specific subject is easier for an AI system to interpret than one comprising two hundred articles on disparate topics. This is not because the systems favour brevity, but because they favour readability. They need to be able to understand what you are referring to.
What is GEO and why does it matter now?
GEO is the term that has come into use to describe this new field: Generative Engine Optimisation. In other words, it refers to the set of decisions that enable an organisation to be interpreted, cited and recognised by AI systems that synthesise responses.
It isn’t just SEO by another name. They share some principles, but operate on different principles.
Traditional SEO optimises pages so they appear in search results. GEO builds entities so they are recognised as authoritative sources. SEO works with keywords and technical signals. GEO works with conceptual coherence and editorial architecture. SEO measures rankings. GEO measures whether you are cited when someone asks about your area of expertise.
The distinction is not technical. It is strategic. And it is the reason why two organisations with similar budgets, operating in the same sector and with equally well-designed websites, can achieve radically different results when assessed by an AI system: one emerges as a benchmark, the other is overlooked.
The concept that explains it all: entity
There is a term that AI systems use internally to categorise what they find on the web: entity. Not a page, not an article, not a website. An entity.
An entity is anything about which a system can construct a coherent representation: a person, an organisation, a concept, a brand. What makes something recognised as an entity is neither its size nor its age. The point is that the system can understand what it is, what it knows and what its position is.
When an organisation builds its digital presence in a way that allows for this interpretation, it begins to exist as an entity for AI systems. And it is entities that are cited. Individual pages are not.
This has a direct implication for how content is conceived. Not as a series of pieces published over time, but as a system that builds and reinforces a coherent picture of what the organisation knows and the perspective from which it speaks. Each piece does not exist in isolation: it exists because it reinforces the whole. And it is the whole that systems interpret. This is precisely what we explore in the shift from positioning content to building a digital entity.
What an organisation can do
It’s not about technical know-how. It’s about editorial and strategic decisions that any manager can understand and guide, even if they know nothing about code or algorithms.
Define a clear territory. Don’t try to cover everything. Choose the two or three areas in which the organisation has real expertise and focus on building authority in those areas. A vague territory does not constitute a state.
Take a clear stance. It is not just about describing topics, but adopting perspectives. The sources that AI systems cite most frequently are those that offer a unique perspective, not those that simply regurgitate information already available elsewhere.
Build with hierarchy, not with volume. There are sections that define the conceptual framework of a subject area, sections that explore it in depth, and sections that illustrate it with examples and analysis. It is this hierarchy that ensures the content is cumulative rather than scattered.
Connect the pieces together. An AI system does not read individual articles. It reads relationships. When an organisation’s content is linked together in a logical manner, the system can map out that organisation’s knowledge architecture. When they aren’t connected, each piece is a loose end with no map.
Be consistent over time. There are no shortcuts here. Authority in the eyes of AI systems is built in the same way as reputation and public influence: through consistent repetition over a sufficient period of time. Organisations that are seen as benchmarks do not achieve this status because they published a lot all at once. They do so because they have long been recognisable for their consistent approach.
The question worth asking
There is a simple way to assess whether your organisation is building a presence that AI systems can interpret, or simply accumulating content that doesn’t add up.
Ask yourself this: if someone asked ChatGPT to describe what your organisation is all about, what would it say? Would it be able to give a specific answer? Or would the answer be vague, generic, or simply non-existent?
If the response is non-existent or vague, the problem isn’t one of visibility. It’s one of interpretability. And the solution isn’t to publish more. It lies in ensuring that what already exists, and what is published from now on, reinforces a coherent picture of what your organisation knows and the perspective from which it speaks.
That’s what determines whether an AI cites you or ignores you. Not the number of articles. Not the content budget. It’s the structural coherence of what you build.
