Strategy · GEO
Before we talk about content or AI platforms, there’s a question that most digital marketing has never asked with any rigour: what does this company actually know how to do that no one else does quite the same way? Not what the website says. What it genuinely knows how to do.
Content marketing has a foundational problem
For fifteen years, the standard starting point for any content strategy was the same: a keyword tool. Search volume was identified, terms with manageable difficulty were selected, content was produced around those terms, and everyone waited for Google to index it.
That logic made sense in an ecosystem where ten positions existed on the first page. In genuinely competitive markets — insurance, legal, health, finance — falling to page two already meant disappearing. But at least there were ten positions, and all ten had value.
In AI there is one answer, or two. And that answer isn’t built on search volume or technical page structure. It’s built on real authority over a specific topic, backed by sources the models consider reliable.
The marketer who approaches GEO with an SEO mindset makes the mistake at the very first step: they look for the topics with the highest demand. In GEO, that’s precisely the problem. High-demand topics have been written about by more people for longer. They’re the ones AI has already resolved. That territory is occupied.
What it means for a space to be occupied in AI
The signal is direct: ask four different AI platforms the same question. If all four return the same list, in roughly the same order — that space is closed. There is no content work to be done there.
Some examples of closed spaces across different sectors:
“Best CRM for small businesses” — Salesforce, HubSpot and Zoho have dominated that answer across all platforms for years.
“How to open a franchise in the UK” — Dozens of identical guides from franchisors, consultancies and trade press. The AI has a formed view.
“How to register as a sole trader in the UK” — HMRC registration, National Insurance, Self Assessment, the £1,000 trading allowance. Written fifty times over. One more article moves nothing.
“Benefits of the Mediterranean diet” — Total saturation for over a decade.
The useful question isn’t “how many people search for this?”. It’s “what does the AI already know about this, and from which sources?”. No tool answers that. Someone who knows how to read a market does.
The groundwork nobody does
Before looking at the market, you need to look inward. Not at brand positioning. At what the company genuinely knows how to do, why it knows it, and who inside the organisation holds that knowledge.
That knowledge isn’t in any document. It’s in people that marketing rarely finds — because it doesn’t know they exist, or doesn’t take the trouble to look for them.
They know the exact points where clients go wrong, why they go wrong, and what it takes to avoid the mistake. That knowledge isn’t written down anywhere. Whoever publishes it first, with rigour, becomes the reference.
Not the buyer persona objections — the real ones. The ones that surface in the first thirty seconds of a conversation. “Does this include…?” “What I’m worried about is…” “I tried another firm and…” Those phrases are a map of what the client doesn’t understand and what holds them back. They are the questions AI still doesn’t have good answers to.
What problem do you solve that others don’t solve the same way? Not in terms of value proposition — in terms of real cases and clients who come back. Is it written anywhere that an AI can read?
If the answer is no — that’s where the work is. And if that space also isn’t occupied in AI, that’s where the opportunity is.
Why this doesn’t fit the standard process
Content marketing was built on tools that produced actionable data quickly: search volumes, rankings, engagement metrics. It was scalable and measurable from day one.
Extracting operational knowledge from an organisation doesn’t work that way. It requires interviews with people who have no habit of systematising what they know. It requires asking the right questions — which means understanding the business well enough to know what to ask. It requires persistence. Going back when the first round wasn’t enough. Crossing what the technical specialist said with what the salesperson said, and finding where the useful contradiction appears.
That doesn’t fit into a weekly content meeting. It doesn’t fit into a brief. It doesn’t fit into the process of an agency running five active clients simultaneously with an editorial calendar to fill every month.
And there’s something else: in SEO, generic knowledge worked. A well-structured article on a high-volume topic generated traffic even if it added nothing new. That’s why the process never needed to go further. In GEO, generic knowledge arrives late to a conversation that’s already closed. Only specific knowledge — the kind that comes from real experience no one else has documented — has any chance of being cited.
Fabrication doesn’t hold. Content built without real experience behind it collapses under its own weight. And in GEO, what collapses gets nowhere.
The detective doesn’t manufacture evidence. They build the case from what they find. When nothing relevant turns up in one place, they keep looking elsewhere.
That’s GEO. And for now, it’s what most of the market still isn’t doing.
Before we talk content strategy, we do the detective work. We identify what real knowledge exists inside your organisation, where there is available space in AI, and what needs to be published for the models to find you before your competitors.
Deslumbra IA · deslumbraia.com