GEO gives thought leadership a job description

Thought leadership has always suffered from a vague definition. AI visibility gives it a more concrete role: create the distinctive evidence and narrative that people and machines have a reason to retrieve.

I have always found thought leadership slightly frustrating as a term because it can mean almost anything: an executive opinion piece, a research report, a LinkedIn post, a branded podcast, a white paper, a keynote or a campaign with a survey attached to it. Sometimes it means genuinely original thinking; sometimes it just means content written in a more serious tone. That looseness makes it difficult to argue for precisely.

GEO is almost the opposite. Whatever terminology eventually wins — GEO, AEO, AI search optimisation or AI visibility — businesses are taking the underlying problem seriously because it feels concrete. They can ask whether they appear in AI-generated answers, how they are being described, which competitors appear instead, what sources are being used and which important buyer questions they answer badly. These are operational questions that can be audited, compared and re-tested.

That creates a useful connection: GEO can give thought leadership a much more specific job. Rather than producing impressive content so the brand looks authoritative, extracting a point of view from an executive or publishing original research simply because it feels valuable, thought leadership can be tied to a clearer purpose.

A more useful definition might be:

Thought leadership creates the evidence, interpretation and narrative an organisation wants to become known for. GEO tests whether that knowledge is actually visible, retrievable and represented in the information environment.

That is a much tighter operating model.

The mistake would be to turn thought leadership into another optimisation tactic

There is an obvious bad version of this idea. A company runs an AI visibility audit, finds 40 prompts where a competitor appears more often and tells the content team to produce 40 pieces of “thought leadership” against them. It has simply rebuilt the keyword-content machine with a different interface.

I have already argued that AI search risks optimising the wrong layer. If the underlying answer is generic, restructuring it for retrieval does not make the organisation more interesting. The stronger use of GEO is upstream, where a visibility review can expose weaknesses in an organisation’s public account of a topic.

Editorial judgement then has to distinguish between several different problems. The answer may already exist but be badly structured; useful evidence may be scattered across research, customer stories, videos and expert conversations; or the organisation may hold strong knowledge internally without ever expressing it clearly. In other cases, it may simply not know enough yet.

That last category is where thought leadership becomes especially useful, because the answer to a visibility gap is sometimes not another page but an instruction to go and learn something worth publishing.

Thought leadership becomes an evidence-creation function

This is the connection I have been moving towards in Editorial Intelligence. The useful sequence is not simply topic → content → distribution. It is closer to:

buyer question → current representation → evidence check → knowledge gap → research/reporting → evidence → argument → narrative → activation → AI representation → new signals

That changes what thought leadership is for. Instead of merely expressing expertise, it can help create new organisational knowledge.

An important buyer question may reveal that everybody in the category is repeating the same public information. That creates an opportunity to interview customers, analyse first-party data, investigate practitioner experience, commission research, test an executive hypothesis or use an event to gather evidence.

The eventual output may be a report, article, video series, executive narrative or campaign, but the most valuable thing was created before the format was chosen: the organisation now knows something more specific than it knew before. That knowledge is harder for competitors to reproduce than another well-written explainer, and it gives AI systems more useful source material to encounter.

I would not turn this into the simplistic claim that “original research ranks in AI”. There is no mechanical guarantee that a piece of proprietary evidence will cause a particular model to cite you. The stronger argument is that original evidence can make an organisation’s public information environment more specific, attributable, differentiated and credible; GEO can then help test whether that improvement changes how the organisation is represented.

Original narrative matters as much as original evidence

Evidence on its own is not thought leadership. A survey can generate 50 statistics and still produce no meaningful position, which is where I think Narrative Architecture matters. The point is not to manufacture a succession of unrelated “insights”, but to connect evidence into a storyline that becomes stronger over time.

A research finding may support an argument, a customer story demonstrate it in practice and an event test it with practitioners. An executive can then develop its implications, while a product story shows where the organisation can act on it and a journalist or analyst begins using the same language. Eventually, a customer may repeat the framing back or another piece of research extend it.

At that point, the organisation is no longer simply publishing assets about a topic; it is building an intellectual territory. Memorable positions are easier for people to understand and repeat, while an AI system encountering a coherent body of evidence has a clearer account of what the organisation is associated with than one encountering a collection of disconnected pages.

This is the part of thought leadership that has often been underspecified. The unit is not the individual thought-leadership asset but the narrative being built across assets.

Hidden Hours is the kind of shape I mean

The closest example from my own work is Hidden Hours. It did not begin as a GEO project, and that matters because I do not want to reverse-engineer a case study around a newer label. It began with research into the operational burden on accounting professionals.

The value was not simply that the research produced statistics. The evidence supported a wider storyline: a significant amount of valuable professional work was happening around the edges of what firms formally scoped, charged for and recognised.

That storyline could travel through practitioner conversations, editorial articles, events, customer evidence, regulatory developments and product relevance. Each new piece did not need to invent a completely different campaign idea; it could extend the same territory.

This is the shape I now find interesting through a GEO lens. Imagine that an AI visibility review around the territory identifies a question the existing evidence cannot answer strongly enough: why do professionals repeatedly do work outside scope without charging for it?

The weak response is to optimise an article containing plausible explanations. The stronger response is to treat it as a reporting question: talk to practitioners about the last time it happened, identify the mechanism, capture the tension and test whether the initial narrative holds.

That process produces new evidence, which strengthens both the narrative and the surrounding public information environment. AI visibility can then be tested again, creating a genuine loop between thought leadership and GEO.

GEO can make thought leadership less floaty

One reason thought leadership becomes vague inside organisations is that its objective is described in equally vague language: build authority, own the conversation, create cut-through or be distinctive. These are reasonable ambitions, but they are difficult to operationalise.

AI visibility adds another observable surface. You can ask:

Which buyer questions do we want to be associated with?
How are we currently represented around those questions?
Which evidence appears when an AI system explains the issue?
Does our original research surface at all?
Are our concepts or frameworks being understood accurately?
Are competitors supplying the evidence for a conversation we claim to lead?
Is our narrative consistent across research, product content, customer proof and third-party sources?
Which gaps are technical, which are editorial and which require new knowledge?

This does not solve the measurement problem completely. AI answers vary, attribution is incomplete and citation behaviour is not stable enough to pretend that we have recreated keyword rankings. It does, however, give thought leadership something useful that it often lacks: a feedback surface.

Instead of publishing a flagship report and declaring victory because it generated downloads or impressions, the organisation can examine whether the ideas it invested in are becoming part of the way the category is represented. That is a more interesting question.

Thought leadership can also make GEO less generic

The relationship works in the other direction too. Without thought leadership, GEO can easily become a programme for making existing information more legible. That work matters: important pages should be accessible, claims explicit, entities understandable and useful evidence should not remain trapped in a PDF or a video without a transcript. Organisations should also know when AI systems are representing them badly.

But if every competitor has access to similar optimisation techniques, optimisation alone is unlikely to create a durable intellectual advantage. Thought leadership supplies what the technical layer cannot manufacture:

proprietary evidence;
first-hand customer and practitioner experience;
original interpretation;
distinctive concepts and language;
coherent strategic narratives;
expert judgement;
useful disagreements;
a developed point of view that becomes stronger as evidence accumulates.

GEO can make these things easier to find, but it cannot invent them on the organisation’s behalf. The two disciplines therefore solve different parts of the problem: thought leadership builds something worth retrieving, while GEO improves the chances that it can be retrieved, understood and measured.

A practical operating model

I would now connect the work in six stages:

Start with consequential buyer questions. Not every prompt deserves a content response, so identify the questions and decision criteria that actually matter to the market, category or buying process.
Test current AI representation. Use an AI Visibility Audit to understand how the organisation and its competitors are represented, what evidence is surfacing and where the account is weak.
Diagnose the real gap. Separate technical retrieval problems from fragmented evidence, weak positioning and genuine knowledge gaps.
Create the missing knowledge. Where the organisation does not know enough, use research, reporting, customer access, expert interviews, events, first-party data or other signal gathering to build a stronger evidence base. This is where research-led thought leadership becomes a practical intervention rather than a vague content label.
Build the narrative. Use the evidence to develop a clear argument and connect it through Narrative Architecture, rather than treating each output as an isolated asset.
Activate and re-measure. Publish, distribute, earn third-party repetition where possible and then test AI representation again. Examine what changed, what did not and which new questions appeared, then feed those signals into the next cycle.

That is much closer to an operating system than a campaign.

The important distinction: visibility is not the idea

There is one boundary I would keep very firm: AI visibility should not dictate what an organisation believes. If the only reason to create a point of view is that a prompt-tracking tool found a gap, the organisation has not developed thought leadership; it has developed demand-responsive content production.

The editorial task is still to decide what is worth knowing, what is worth investigating and what the organisation has earned the right to say. Sometimes the correct response to a visibility gap will be to ignore it; sometimes the evidence will contradict the position the business hoped to take. The strongest piece of thought leadership may even introduce a question that no visibility tool was tracking because the market had not yet framed the problem that way.

That is why the work still requires judgement. GEO can provide demand intelligence and a feedback loop, but it should not become the commissioning editor for the entire organisation.

A less vague definition of thought leadership

This connection has helped me make the term more useful for myself. Thought leadership is not simply senior people having thoughts in public, “premium content” or a particular format. I increasingly think of it as a deliberate process for creating and developing an organisation’s useful point of view through evidence.

Evidence gives that position credibility, narrative gives it coherence, activation gives it reach and third-party repetition gives it external weight. AI visibility adds another way to observe whether the position is becoming part of the information environment buyers now use.

That gives the supposedly floaty discipline a harder edge and GEO a more interesting ambition than getting another generic page cited. The goal is not simply to become visible to AI, but to become visible for something worth knowing.