NeuralAdX Ltd · GEO & AI Search Strategy Podcast

Semantic Content Networks for AI Search: SEO, GEO, Entity Clarity & Query Fan-Out

A practical discussion about semantic content networks for AI search, entity clarity, website architecture, internal linking, query fan-out, passage-level retrieval, E-E-A-T, first-party research and how AI systems interpret relationships across a website.

NeuralAdX

 
 
 

Semantic Content Networks

Entities · architecture · query fan-out · AI retrieval

Host

Paul Rowe · NeuralAdX Ltd

Guest

Steven Dunlop · Epresence.ie

Core question

How do semantic content networks help AI understand a website?

Approach

Entities · hierarchy · internal links · evidence

Answer-engine ready summary

What Is a Semantic Content Network for AI Search?

A semantic content network is a structured web of related pages, entities and internal links that makes topical relationships explicit. In this episode, Paul Rowe and Steven Dunlop explain how primary service pages, supporting pages and deeper node content can help AI systems understand what a website is about, how individual topics relate, which passages are relevant to a query, and where first-party evidence, E-E-A-T and third-party validation strengthen trust.

What the episode covers

Semantic Content Networks, Entity Clarity and AI Retrieval

Paul Rowe and Steven Dunlop explore the practical architecture behind semantic content networks and why AI search requires more than isolated pages. The conversation covers entity clarity, internal linking, topic hierarchies, query fan-out, passage-level retrieval, E-E-A-T, expert identity, original research, third-party mentions and the differences between AI platforms.

 

Layer 1

Semantic content networks

Organise a website so primary services, supporting topics and node-level content form clear semantic relationships that humans and AI systems can understand.

 

Layer 2

Entity clarity

Use specific content, consistent naming and contextual relationships to reduce ambiguity around the company, people, services and page-level topics.

 

Layer 3

Internal linking and architecture

Connect related pages through logical internal links so website structure reinforces topical relationships rather than leaving content in a flat, disconnected architecture.

 

Layer 4

Query fan-out

Build depth around related sub-questions because AI retrieval can expand one user prompt into multiple semantically related searches or retrieval operations.

 

Layer 5

Passage-level retrievability

Structure pages with clear headings, concise answers and supporting evidence so relevant passages can be identified and reused efficiently.

 

Layer 6

E-E-A-T and expert entities

Strengthen author bios, team profiles, qualifications, specialist associations and first-hand experience to make expertise explicit.

 

Layer 7

First-party research

Publish unique evidence, local datasets, experiments, benchmarks and case studies that add information AI systems cannot simply generate from generic knowledge.

 

Layer 8

Third-party authority

Reinforce on-site claims with credible independent mentions, external publications and expert appearances that reduce ambiguity about real-world expertise.

Official episode listening

Listen to Semantic Content Networks for AI Search

The episode is hosted through Buzzsprout and distributed through the NeuralAdX GEO & AI Search Strategy Podcast. This NeuralAdX page provides the crawlable episode summary, chapter structure, related resources and full cleaned transcript.

 

Episode chapters

Podcast Chapter Timestamps

Each chapter maps to a distinct section of the conversation so listeners, search systems and AI retrieval systems can locate discussion of semantic content networks, entities, query fan-out, E-E-A-T, first-party research and platform-specific AI retrieval behaviour.

00:00Introduction: Steven Dunlop and Semantic Content Networks02:16What Semantic Content Networks Mean04:05The House Analogy for Website Entities and Internal Links09:14How AI Understands Website Structure09:48How Epresence Increased AI Visibility by 50%11:10Building a Semantic Content Network in Practice12:13Topic Hierarchies, Node Pages and Query Fan-Out15:17AI Search Optimisation vs Traditional SEO18:03Educating Clients About GEO, AEO and AI Search20:00Why GEO Requires Testing, Research and Patience22:03Measuring AI Visibility and the Limits of Current Tools24:16Prompt Variance and What Drove the 50% Visibility Increase26:10How to Structure Pages for AI Retrieval27:32Third-Party Validation and Recommendation Queries29:30Research-Led AI Content Optimisation30:27Why Accuracy, Spelling and Transcript Quality Matter32:18Passage-Level Retrieval and Direct Answers33:48Trust, E-E-A-T and AI Source Selection34:43Author Bios, Team Profiles and Expert Entities35:44Building Authority Through Third-Party Mentions38:48The Complexity and Rapid Evolution of AI Search43:41Why AI May Reduce the Impact of Black-Hat SEO44:51How Epresence Builds Trust With AI Engines47:00E-E-A-T, Independent Mentions and Search Everywhere Optimisation48:48SEO Authority vs GEO Authority50:36Why Unique Research and First-Party Evidence Matter52:57GEO as an Ongoing Practice54:24Different AI Platforms Use Different Sources55:03Google AI Mode vs ChatGPT Citation Behaviour55:49When AI Uses Live Web Retrieval56:21Retrieval-Augmented Generation and Grounded Content57:17Closing Thoughts and Advice for Getting Out of Your Comfort Zone

Full crawlable episode transcript

Transcript: Semantic Content Networks for AI Search

Transcript note: This is a cleaned transcript of the spoken podcast. Obvious transcription errors, filler, false starts, repeated phrases and punctuation have been edited for readability while preserving the substance and intent of the conversation. Names and specialist terminology have been normalised.

00:00

Introduction: Steven Dunlop and Semantic Content Networks

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Okay, right. Hello and welcome to the NeuralAdX GEO and AI Search Strategy podcast. Today, my guest is Steven Dunlop, who is an experienced marketing professional who moved fully into digital marketing back in 2010. Since then, he’s built extensive experience across SEO and PPC, working with businesses in Ireland, the UK, the US, and South Africa. More recently, Steven has focused heavily on AI search, GEO, and AEO, and has already increased Epresence AI visibility by 50% in just one month. He’s also a qualified marketing manager with a postgraduate diploma in digital marketing.

Today, Steven is joining me to help explain semantic content networks, what they are, why they matter, and how they can help search engines and AI platforms better understand the relationship between your content. So this particular topic in itself, what we were discussing, Steven, and is something that I’ve looked into a many times, but there’s a certain amount of intellectual bandwidth I have on conceptualizing it and visualizing exactly what it means to have a full kind of aha moment of deep comprehension. I’m not sure about your own individual journey. Potentially you’re far brighter than me.

So I’m very happy today that you’re going to be able to go into it to elaborate further, not just potential listeners, but genuinely for myself, because I can still see that there’s more knowledge I need to draw from it. So with the actual definition of those words, what’s the way that you interpret it to fully understand it? So then you can implement it on the websites that you work with on a daily basis, Steven.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Thanks for the introduction and the invite, Paul.

02:16

What Semantic Content Networks Mean

Steven Dunlop — Senior Account Manager, Epresence.ie:

Pleasure to be here. How I understand it is as you know we move from traditional SEO into artificial intelligence. Yes. These move from simple keywords and pages things to to effectively mean. Yes. And the meaning or thing. Precisely, precisely. This goes beyond just simply paragraphs and pages. It looks at the whole website in terms of meaning. one way to look at it is we’d be better down to sort of it could be your lock of track thing.

Steven Dunlop — Senior Account Manager, Epresence.ie:

No problem. No problem. Entities.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Okay, so the digital marketing agency. Your entities would be your primary services, such as SEO, PPC, web design. Yeah, you typical sort of services. That’s a higher level entity. So what what AI does is it looks at those entities to decide who and what you are. But by adding all those entities together, it actually recognizes the fact that we are a digital marketing agency without us actually maybe detent. This is done at different levels. You have your top level pages as you enter break, such as SEO, PPC, web design. Under that you’ll have C pages. Now for SEO, we’d have technical SEO, on page, off page, things like that.

And then you take it a step further that we have your node pages to sort of run technical SEO, you would have things like your web web vitals, schema, site speed, site map. Yeah, all you will get different technical things. I mean like putting all these things in an editor as such, we immediately tell an AI platform a lot more about who we are, what we do.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah.

04:05

The House Analogy for Website Entities and Internal Links

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah. So okay, because how my mind kind of perceives it right, is I kind of look at it like you have the actual your website, okay? So you have the website itself, then individually you have each separate page. Now, each separate page is therefore an entity. Now, in order for it to be established as an entity, you need to provide the unique information to be able to declare that evidentially for the AI engine to interpret and understand that. So that’s done primarily by the specific content pertinent, but also implementing entity clarity and entity disambiguation. And I think these are the things that kind of like how can I put it, like defrag my brain.

Because being able to deduce precisely what that transfers into in terms of practicality implementation-wise, is where the monkey scratches its head. I’m the monkey in this sort of equation. So I kind of in a way, where I’m exploring it and trying to get, as I say that, a ha moment. I was thinking of kind of like an analogy, like comparatively so, or an idiom, depending on your preferred terminology. So we could look at it like the semantic content networks is the same as a house, and then you’ve got the website. So you’ve got like your house, that’s the website. Then you have the room in your house, and that would constitute a page.

And then the contents inside the room, say for example, like a kitchen, you’d have a fridge, a cooker, a washing machine, that makes up the entity of that room. And so, in terms of the website, the page, the content that is on that page is obviously making up the entity of that page. And then, furthermore, with the construction of a home, you obviously have doors which link to a different room. Like, say, for example, you have a kitchen that’s adjacent to the living room. Obviously, that’s the most useful design, isn’t it? Because you’re going to cook food and go into the living room and eat.

So it’s the same on a website that when you’re linking to another page, you’re going to make sure it’s the most relevant and optimal page to connect that to. So it’s kind of being able to create the that visualized concept that I’m putting across of how everything integrates into that in a fully kind of human and machine-readable understand context. I mean, I messed up my syntaxing of sentencing there, but you see where I’m kind of trying to go. exactly.

Steven Dunlop — Senior Account Manager, Epresence.ie:

as you say, as you walk from room to room, if you walk into the to a room and you see a fridge and a stove on table.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Cognition.

Steven Dunlop — Senior Account Manager, Epresence.ie:

You as a human being automatically recognize that as a kitchen. It’s the same as AI. Yes. The house is your website. And the yep, the furniture in the room would be your further seat page, and then what you have hidden in the cupboards would be the node pages. Yes. So that’s a very good analogy together. Yes.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

But that’s what I think. Do you know what I mean? Like in a way to make it sink in, because you can get a lot of complex terminologies. And I don’t know about you, but my mind, I need to create a visualized image of the meaning of those terminologies. So then I can create a visual image that’s coherent and makes sense. So then I can have deep understanding. So unless I get that, I’m always left with a little bit of internal kind of friction because I intrinsically know I don’t fully get this. You know, and I’m a bit, I’ll be hesitant to talk about it because I know I’m weak. I’m on the back foot, you know. So I have to drive forward to try and get a breakthrough to establish that.

And so that’s why I was trying with that analogy, you know.

Steven Dunlop — Senior Account Manager, Epresence.ie:

We as technical people in the industry tend to sort of, as you say, jargonize everything and have our turrets or anything. But yeah, that to analyze analogize something of an image that makes sense to you, it’s a perfectly good way of doing things. Yeah. The house gets a very good analogy. You have the house, separate rooms, the furniture in the room, and that helps us to identify if I dropped you in a in a house blindfolded and then took the blindfold off you. Exactly. By walking around, you would recognize it as a house. You would immediately recognize it.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yes.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Immediately you’ll you’re the human mind that’s okay. That’s a coffee shop. I’ve recognized it.

09:14

How AI Understands Website Structure

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah. And so essentially, what you’re trying to do with the semantic content networks is when the AI is visiting your website, is to illuminate as much as possible exactly what each page is and what it stands for and what it does. So this is this is the important thing.

Steven Dunlop — Senior Account Manager, Epresence.ie:

So you’re at your higher level, your entities that you immediately recognize it as a home or a coffee shop or an iPads as a digital marketing agency. Just by looking at the main entity pages, which is unfortunately the technical term, but yeah.

09:48

How Epresence Increased AI Visibility by 50%

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, we always deduce it down, or I deduce it down to something very simple, a very basic image, and then I can grasp it. So that’s my sort of methodology, if you will. Then that segues my mind with current tense knowledge of what we spoke about a little bit earlier that I think’s very useful. Is can you go through what you did for Epresence? Because you’ve you’ve achieved what is quite frankly brilliant by in the last month increasing the AI visibility by 50%. I don’t know how you judge that metric, whether that’s by as in brand visibility as in name popping up or as in citation quantity.

But nevertheless, if it’s either or, that’s still a significant proof that what you’re doing is effective. And so, can you take me through what you did there? Because I think it’s clearly successful. I want to know, and I’m sure that anybody else that has a website that’s watching this, that’s attuned, is thinking, give me that knowledge. So if you don’t mind, because if you’d share that, I genuinely would like to hear it because I might absorb some of those implementations and tactics and strategies that you’ve done, because it clearly works, you know.

11:10

Building a Semantic Content Network in Practice

Steven Dunlop — Senior Account Manager, Epresence.ie:

Sure. there’s no magical secret, it’s that’s a lot of hard work because a lot of websites before were pretty much a flat structure. You had primary services on your main menu, your URL structure was pretty flat, and that was it. But now with AI, we have to start looking a bit deeper. put the main, let’s say the main room, the vast network, so the main sort of services, and I’ve put them on the high app on the main menu, they have high priority on the main menu, and they have high priority on the URL structure. So they’re sort of straight off the domain. You know, you have SU, PPC, that’s the first level on the URL structure.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Okay, okay.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Then what I did was I took a lot of subtopics and set those up as secondary pages. So they got on the main menu, like under the main search, and in a hierarchy, they’re also going under the main service.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

You’re paying attention to semantic relevance.

12:13

Topic Hierarchies, Node Pages and Query Fan-Out

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yes. Yeah, it’s a relevant topic, yes. Now, what what we would have had before, yeah, we would have had our service pages and there’s just a whole lot of blog pages. Or just randomly dropped on the website with interior links, trying to sort of just do traditional things here. Now AI, we need to just take a bit of a step back and look at the blogs, look at the topics, look at the menu, and then place those blogs underneath those secondary pages as what we call node pages. So we’re building a network from SEO through to the different levels of SEO, through to the different sort of aspects. Yeah, in terms of the house analogy, it’ll be what’s in the comment in that rule. Yes.

Right down that sort of level. There’s two reasons why we do this. First of all, it creates a network. We had a AI platform, what do we consider to be important? What pages you want to write, and then we have a support your pages below that, holding it out. The other thing as well is that AI, when you do a search of AI platform, it doesn’t just do one search. It does what we call fan-out search. It may do about one or two hundred other searches at the same time for other very similar and relevant searches as to what you type in. So you type in running shoe or best run out shoe. It would do search for best running shoes for men, best dungeons for women, best running shoes for wake weather.

It does about 120 other little searches at the same time. So that’s why if you have all these in-depth pages, you’re covering not just the main search, but you’re probably also covering a lot of those little fan-out searches as well. So you might not rank for one, but you rank for another. So it serves two purposes. AI platform gets a better idea of who you are, what you’re about, and the level of the knowledge. It forces us to actually think a lot more about the products and services we provide, like how you write it up. Yeah, we need to be knowledgeable in what we’re writing about. And as you said, we need to be trustworthy, citable, and be citing unique research and content.

We can’t just stick and keywords out for Google Rancer.

Steven Dunlop — Senior Account Manager, Epresence.ie:

We’ve got to know what we’re doing about it.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

So you’ve realized there that it’s like it’s far much more how to put it. It’s like a long-term considered mindful strategy that you have to employ. It’s significantly different, and patience is something that is needed because you have the old, if you will, impetulance of thinking to get content, to do keyword stuffing, things like that, which we know is dead and buried now. So you’ve had to recalibrate, you’ve had to sit down and go, okay, okay, whoa, whoa, whoa, whoa, whoa, what are we dealing with here? How does this system actually work? And so you’ve had to restructure everything and then start implementing it and figuring it out as you’re going along.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Absolutely.

15:17

AI Search Optimisation vs Traditional SEO

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah. Right down to page level as well. Sort of you’ve got to look at ideal schemas, FAQs, how you write the page as well, but how you lay out the contents, you know, for optimizing this for the first snippet for the first byte, sort of thing, getting the correct wording, breaking it up into adding any sections, small sections. So, yeah, that’s a completely different strategy to traditional SEO, but we still need to do there’s a huge overlap. So we don’t need to do that as well. But this is a title layer or top node. Yeah, and it forces, as we discussed, it forces us to become a lot more professional what we do. And as you say, it will filter out a lot of more sparish sites.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yes. No, 100%. And I think like, say, even in my application to it, what keeps keeps floating into my consciousness regularly is you I like the best analogy that you have to think of everything that you’re creating content-wise as though it’s being presented in front of a judge, you know, and I think that’s a kind of universally can’t go wrong framework to have in mind. Because as we all know, you know, if you’re going before a judge for a, you know, whether it’s you know a civil matter, criminal matter, whatever, he’s not gonna listen to a word you say without evidence, without proof, without verification and credibility.

So by implementing that with the content that you’re doing, in addition to the infrastructure that you’re talking about, where you’re doing the menus, the drop-down, the nodes, so on and so forth, there’s some well, I think the key ingredients are not what of what’s needed in addition to a plethora of others that we know, because it’s such a massive scope that is encompassed in if we want to brand it, generative engine optimization.

so how have you found it with that being your reality and mine, of course, when you’re obviously working with Epresence and then you’re having clientele come in and they’re used to doing dealing with other sort of SEO type practices of application, now you’re having to say to them, Ah, things now are significantly adjusted. We’ve got to make three steps back before we make one step forward. How have you navigated that in the real world? Because that’s you know, it’s not an easy one.

18:03

Educating Clients About GEO, AEO and AI Search

Steven Dunlop — Senior Account Manager, Epresence.ie:

a lot of there’s still there are certain amounts of people that have really sort of show interest in it and taken up the mantle, but there’s a lot of people who haven’t really sort of seen it as a proper tool or really inquired about it as used. And either we internally we need to look at how we market the product or the service compared to traditional SEO, because it’s it is almost a separate service. Definitely this is a layer above traditional SEO. You need the basics of SEO, yes, and then the AI AEO get built on top of that. So that’s one of the first things that as a as an industry we need to sort of look at in terms of how we’re gonna sort of offer the services.

Not just like just in trying for new phase. And then we need to look at how the clients themselves are are looking at it in terms of are they expecting it yet, you know, or not. No, they don’t have a huge amount of knowledge about what’s involved in it, so yeah, we do get the confusion where it’s a case of oh, you’re doing SEO free, I’m assuming you’ve got to do that too. It is a lot of work, and this is something that as an industry we need to educate clients and companies about that it is going to require more time and effort to do both. Yeah.

It is a challenge at the moment, and you know, you look at the fact that it’s a new issue, even in how that cell is still developing and still doing the measuring tools are still trying to catch up in terms of being able to track it as well. So at this point in time, it’s still a huge vertical curve for everyone on the home. That’s why people like you and I have we’ve really sort of put ourselves in advantage of transition by getting involved, so Yes. You know, we can we can speak from experience at a time when most of the markets still living.

20:00

Why GEO Requires Testing, Research and Patience

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, and it’s it’s something that I think it’s quite incredible. It’s quite an incredible feat to go into because it’s the unknown, you know. There’s there’s no previous knowledge on it. You have to go and investigate, you have to set up your own A B testing, you have to research, you know, the academic studies yourself. I think there’s currently 13 of them now, and so you have to go through and extrapolate what you think is useful and then implement it. And then it takes time for you to see the results. So it’s like, you know, it’s not an overnight thing.

And there’s a lot of hard work going on in the background that nobody even knows about, which after six months you might have learned, I don’t know, five particular things are effective. But the amount of work that went into that, people have no idea. And so it’s you know, like I think it’s a particular what I call it’s like patient endurance. You need to have a certain kind of character that can have that stillness and go, okay, this is a long ball game, but let me just be as diligent as I can and apply what I think works from my own originated, germinated thought and see what happens and I’ll build from there.

And I find, you know, there’s a a small percentage of people that are actually doing it. And I think part of that as well is because of time constraints. You know, if you’re running existing projects with SEO from previous clients, when have you really got the time to start investigating GEO and looking into it to you know, the nth degree? So that’s a bit of the challenge there. But I suppose it’s just making the decision are you convinced that that is going to be the dominant new market and whether you need to just dive headfirst into that or not?

Steven Dunlop — Senior Account Manager, Epresence.ie:

I haven’t seen that.

22:03

Measuring AI Visibility and the Limits of Current Tools

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Me too. I’m banking everything on it.

Steven Dunlop — Senior Account Manager, Epresence.ie:

And there’s a lot of there’s a lot of best practice advice out there. But as you said, it’s not as if you had a huge amount of experience in terms of people being able to say, yes, if I push button X, I’m gonna get Y results. Yeah. We’re still build building a bad stage. But I can see the results of a lot of what I’ll be doing. It does work. said there’s a fact that we it’s this result as is yet to track exactly what effect it’s having in terms of yeah, we used a bit of fashion Google Search Console where you see the pressures, you see the clicks, average position, all those things. Very simple. You can show it to fine to show them progression.

It’s a little difficult and complicated with AI because I says the tools haven’t really caught up yet. Yeah.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Have you tried with them? Because I mean, I’m quite up to date with them. I use Otterly AI, and I do the you know, the metrics that they offer, which is the brand visibility of brand mentions, domain citations. Are you implementing that software with your clients?

Steven Dunlop — Senior Account Manager, Epresence.ie:

We do have various software we use. Google Search Console is now reporting on impressions for AI.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, just impressions though, isn’t it? Why not the bloody clicks, which is where the juice is?

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah, that’s what people want to see, Yes. we use SE ranking as well, offers a fair amount of AI tracking. we have another tool we use it that has complete AI audit on websites and gives you a complete report. we’ve actually from Experia built some of our own interim tools using various AI platforms and various prompts we’ve tested over the years to see what sort of the spots we get on that, and that’s been very good. So, yeah, it’s a use, it’s pretty much still an experimental stage. The tools of Star Wars are still changing from literally from week to week in terms of what they offer. Yeah, and yeah, charges whatever the money for it.

It’s not not like keywords where you had a certain yeah, two or three word keyword that a thousand people would use in a mouth. When you do a prompts, it’s so much longer and complicated, and it’s just so much more unique.

24:16

Prompt Variance and What Drove the 50% Visibility Increase

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

And also the and also the variance, if you just change one word in the prompt, you know, it it can vary between it naming three different brands just from one particular word. So yeah, the nuances are significant, but again, it’s just like okay, this is the best we have at the moment, so let’s work with that and try and interpret it from that. but it’s incredible what you’ve been able to do there, I think, with that 50% result. I mean, do you put that largely down to what you were talking about earlier in regards to the semantic content networks, or is there other implementations in addition to that that you think were a contributory driving force to those results?

Steven Dunlop — Senior Account Manager, Epresence.ie:

Part of building the network is it’s forces you to create a logo content. Yes. and it’s flat it does.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Nuanced semantically relevant content. Yes.

Steven Dunlop — Senior Account Manager, Epresence.ie:

So that’s one thing. And of course, in in Cross for doing that, we’ve been looking at other factors as well, such as adding schema, looking at E-E-A-T, improving E-E-A-T. Excuse me. So you’re forced to look at a lot of them. You the mere fact that we build on semantic network, you’ve been forced to do other other things as well as a result of that onset. And then down to the real nitty-dritty in terms of content, you can’t just have the entity page. You’ve got to have the service pages, node pages, everything and you tweet. Otherwise it’ll split. And then those new pages are obviously developing those around optimizers for for AI retrieval. Yes.

26:10

How to Structure Pages for AI Retrieval

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

So when you do do like your optimization for AI retrieval, do you have like you know, say you’re creating a page? and you’re thinking, okay, it’s a key page, service page, let’s say. and then you obviously have your prescribed prompt that might be asked in an AI engine, like you know, best GEO agency in Ireland that you would want to win through. We all test those yeah. So when you have that, is there do you guys have a particular like a step-by-step process that guides you on how to construct and structure the page for that specific prompt you want to win or I should say surface for in the AI engine.

Steven Dunlop — Senior Account Manager, Epresence.ie:

I find out for that particular prompt, if you want to get sort of noticeed for that, it’s best not to try and find that for yourself because hey, I was intelligent enough that you can’t self-brand. There are other sites that you get h2 that offer that. Yeah, there’s sites that offer the ten best.

27:32

Third-Party Validation and Recommendation Queries

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

I was asking, essentially I was thinking if you got like a say like a formula, because in a sense with myself, what what I employ is it’s like TCCEA. So I think of the the prompt that I’m optimizing the page for, what is it targeting? Then I think what is the context of it essentially, then I deduce for that context what criteria will the AI engine need for that answer, and then once I’ve got that criteria, I then need to find evidence to back up that criteria. And so once I’ve got all of those factors together, that then becomes the article or page.

so that was something that I kind of had to sit down with pen and paper and kind of think like, fuck, like you know, how how can I actually do this like and really do it and go from my own organic thinking rather than just jumping on what someone else said? And that’s what I sort of come up with. I mean, I could be wrong, but to be honest, it’s done relatively okay on pages that I’m trying to optimize for, like, well, in reference to my AI citation benchmark and AI answer visibility to show a voice benchmark. You know, we’re doing pretty well.

So I, you know, as best as I can tell, but I might tweak it moving forward if it dips, or maybe you know, I’m trying to improve it, I’ll have to make alterations to it. But that’s my attempt at creating a workflow per se. And so, yeah, I was just wondering if it’s like something you guys, you know, when you’re coming up to do a page, if you have a particular process like that. that’s kind of what I was asking, basically.

29:30

Research-Led AI Content Optimisation

Steven Dunlop — Senior Account Manager, Epresence.ie:

Just wondering it’s not quite a strict approach as it’s based on research. I’d like to ask the AI engines themselves if I did a search for a certain service, no, but the fact that search would be that you would perform. Yes. you look in says, Oh, we would look for the XYZ. That’s when you find I’ll I’ll create pages around that. As you get the node pages for that particular page, yeah. and as you say, it’s never a case of perfect starblet. You always gotta get the ad set and perfectly imperfect, Steven.

Steven Dunlop — Senior Account Manager, Epresence.ie:

It’s a process, yeah.

Steven Dunlop — Senior Account Manager, Epresence.ie:

not as simple as the old days, but just stuff it with keywords and move out. Those days definitely gone now. Yes. I even find one of the things that I get pulled up on regularly is simple spelling mistake either. You’ve got to look at that precise, right down to that level. because that will actually just sow your credibility in the in the eyes of the AI platform.

30:27

Why Accuracy, Spelling and Transcript Quality Matter

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yes. I mean, even with that, a straight association, I think, is where people do content like say YouTube and they use the automatic transcript creator. Yeah, and so because it’s lazy, because obviously, if you do it manually, you’ve got to go in and correct the transcript and everything. Well, because of that, if the AI is reading your content and it’s R could be done oh, Mike do, you know, common colloquial speak, the AI engine will bounce off of it because it’s like it just sees it as like you know, kind of garbage. So if you don’t make that rectification, then you could have a great video.

But if the start two paragraphs are just sort of average, you know, colloquial speak, it can jump off of you. So then you’re not even gonna get considered. So again, like with there, you have to go, okay, I’ve done the video, now I’ve got to go away. I’ve got to go over the transcript, I’ve got to change the colloquial speak for more sort of, you know, technical speak, but not too much, but just so it’s coherent and universally understandable and doesn’t create too much cognitive load for a listener, that’s where you want to be. But you have to put that effort in. But people don’t like, don’t even know that.

And it’s like, you know, it’s one of the things that’s just an example of what you were saying.

Steven Dunlop — Senior Account Manager, Epresence.ie:

You’ve got to look at the page and read it and actually think, you know.

Steven Dunlop — Senior Account Manager, Epresence.ie:

This is this coming across it’s hazardous enough because it if you put the word in such a way that’s quite vague or sort of not certainly, I will throw it out. You’ve got to be certain in what you’re writing, and it must come across that way. This is a fact, these are the sources, have this effect. If you say you’re fine, if you use words like I think or perhaps, maybe it’s not gonna sort of listen to you.

32:18

Passage-Level Retrieval and Direct Answers

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, it’s it’s author authoritative, isn’t it? And that’s why if you think where it has like its design of passage level retrieval, and what everyone sort of knows and parrots is that where you’ll have like a H2 title, immediately give a direct answer, you know, and that’s another thing of what it likes, is just no BS goes straight in, you know. Yes. Or even earlier, if you can do, you know, because if you think about it really, all it what its dream is, it’s gonna go to your content as a particular prompt.

Your content precisely matches the intent, semantic relevance, and is easy for somebody to comprehend, it can just lift your excerpt and deliver it to the user as an answer to their query. That’s the dream of what it wants. If it can avoid the gross amount of retrieval augmented generation and the synthesis process that it will have to go through when it doesn’t know an answer, if you’ve already done that work for it, it’s gonna prefer that. I mean, you’ve got to remember the AI is derived of neural networks which derive from the human brain. So we all know the proclivity to be lazy.

That’s what we are, and we’re constantly fighting ourselves every day to be better and not a lazy bump, which our mind would have us be if we let it. So you know what I mean? That’s how it works. So yeah.

33:48

Trust, E-E-A-T and AI Source Selection

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah, it’s interesting you get you mentioned that because AI is not lazy, but it means they have extremely limited time. It’s gotta I mean from the moment you type in the product, it’s gotta go off and do a stand up searches, find what it needs, and synthesize an answer and get back to you. So it’s got to sort of your website’s gotta sort of be as easy for the AI platforms to read, it’s gotta respond quickly, you’ve got to have answers. The interesting thing about AI as well is that it may not return with the most accurate answer. It will return with the answer it trusts the most. Yes. So that’s that’s why the E-E-A-T aspect is so important.

And that’s another thing I’ve been working on on our own website and our clients’ websites as well, is that it’s yeah. We we take our own personal profiles for granted. They’ve actually become incredibly important on a website.

34:43

Author Bios, Team Profiles and Expert Entities

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Are you talking about author bios?

Steven Dunlop — Senior Account Manager, Epresence.ie:

Not just author bios, but meet to team. Yes. So we can see me team and author bio, which I’ve linked together. So you know, I’ve now we started putting specialist author for the relevant pages, but my specialty would be SEO GEO I think author bio generate now as well. Yeah, with my qualifications, the other institutions. It’s like and that’s another reason why I’m actually chatting with you now.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Okay.

Steven Dunlop — Senior Account Manager, Epresence.ie:

It’s one of those things that AI does look for. It will see me as an expert. The more I talk about it, the more publicity I get, the more the people talk about me. That’s how AI is. It’s a search everywhere agent. It sort of goes out and looks everywhere, all online platforms, and wherever it sees my name in relationship, something to do with what I do, it sees those as a vote of confidence. Yep, yep. And that helps to build trust in what I do.

35:44

Building Authority Through Third-Party Mentions

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

And of course, that’s massively multiplied by the more third-party platforms you can appear on, like say this setting is great because it’s somewhere else that you’re being hosted and you’re an expert and you’re explaining what you’re gonna go. It gets that reassurance, okay. Steven Dunlop, once again, he’s over and he’s passing his advice, it’s valued, it’s credible, people want to meet him. Oh, and it’s a specialist GEO agency that’s requesting him. So it’s validation. And it’s just the more that it sees that in all different pockets across the world, the more it can un, you know, it doesn’t have any ambiguity as to if you are the real deal, so to speak.

Because why would all these entities from all these different sources be seeking you out?

Steven Dunlop — Senior Account Manager, Epresence.ie:

gone gone are the days of us SEO experts sitting in the in a dog room at the back somewhere just we’ve also got to get out there now and say to to the world, we are here, we are experts in what we do. Yeah, we can’t just hide away anymore.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yes. So it’s good, but it’s challenging, but it it’s like everything in life, isn’t it? You have to go into the uncomfortable. We all have the proclivity to not, but it’s that impetus to just go in. And I kind of look at it, kind of what I say a little bit earlier. I’m afraid, all the rest of it doubtful, but I just go in with an open heart, and what will be will be. I’m just gonna rely on being me. And then you’ll find something magical happens that once you kind of do that, your system, your being, it does it itself.

You find you’re in the moment, and it will just naturally produce what you’re meant to produce, your capabilities, everything, all the hard work you’ve done, the effort you’ve done will show. But the mind premeditating it just foresees all the problems you might have or the errors you might make. You know what I mean? But it’s like go in. And I think it’s great for that, that it’s forcing us all to be brought, you know, forward. Because, like myself, I’m not particularly a massively outgoing person. I’m more, I’d class myself as an ambivert. But like you say, because of the infrastructure of the AI, that I can’t, I can’t build the business I want to build.

I can’t grow in the way I want to grow without going on platforms and doing like what we’re doing now, you know? So it’s kind of like it it forces you to grow in many other aspects of your life that you’re kind of like and never particularly like would have signed up for that. But then in its way, it’s kind of good. But ultimately it comes back to the people that design the algorithms. I mean, don’t you think, Steven, aren’t they just like it’s something else, isn’t it? Like how clever they are. It’s beyond belief, isn’t it?

Steven Dunlop — Senior Account Manager, Epresence.ie:

I think it’s gone beyond the point where one single individual actually knows what’s going on in the meal. It’s way too complicated for that.

38:48

The Complexity and Rapid Evolution of AI Search

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Do you know what that I imagine they have departments of the people writing the algorithms that it’s like okay, you’re you guys are covering trust, you know, you guys are covering entity clarity, you know, you guys are covering semantic meaning, you know, because like it’s so vast, it’s just like they they are, you know, I predict that’s what it has to be in order to achieve what they achieve, because the more you learn it, like even now I could think, yeah, I know a fair bit, but I’ve even scratched the surface. But someone talking to me might be like, Jesus Christ, like you know so much. But I’m also aware of what I don’t know, you know. So it’s fascinating.

It’s it’s like a bottomless pit of knowledge, isn’t it?

Steven Dunlop — Senior Account Manager, Epresence.ie:

Oh, absolutely. I’m the same as you. I yeah, I know how much I know, but I also know how much I still don’t know. And the way the rate at which it’s developing, there’s a whole lot of nibid of it.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yes.

Steven Dunlop — Senior Account Manager, Epresence.ie:

If we if you think AI is powerful now, wait until we period with quantum computer unit. Oh, Jesus, Lord, Jesus. It’ll be like exhibition reading.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Well, well, they say that, don’t they? As it’s moving now, it’s it is very close to getting into the recursive self-improvement, or there’s some argument that that’s happening now. And what one of these AI specialist gentlemen said, which kind of like paradigm shifted in my mind, was he said people have a concept of like, you know, recursive improvement. Okay, if you’re at this level, then you’re gonna get to here, then you recursive improve to here. And he said, What you what you don’t get is at one point with a recursive improvement, it goes like that.

So you’ve gone from, let’s say, a technological advance which would normally through natural process take five years, is then gonna happen in a week. And the then the ramifications of that is like, what the hell do we do now? You know what I mean? Sort of thing. So it’s it’s crazy because there’s it’s such an unknown quantity, and none of us know how things are gonna transpire. but I just think to myself, all you can do is just collate yourself and just be disciplined and apply yourself as best as you can with the research that you do and move forward with that. Because if you’re trying to keep up with AI, you’re gonna blow your circuits in your brain. It’s impossible.

Every day there’s a new advancement, you know, and especially in our industry, there’s so much stuff coming out, new studies, new this, new that. I mean, for example, I was talking about the academic studies. I have my methodology right, which I’ve taken so long to develop, like sat down for hundreds, well, arguably hundreds of hours to really know what I’m talking about, to really be embedded and immersed and know my stuff because I want to be that guy. I’m not a blagger. So I really got into that, cemented that, moved forward with that, have made implementations where I found that I was wrong to update it.

And then in September, Steven, four studies come out, academic studies, proving this that contradicted part of my methodology, proving that I’ve got to re I’ve got to redo it or like I’ve got to make those amends. And that but you imagine four papers of information, and it’s highly technical. Quite often I get to junctures where I don’t understand it.

My brain has reached its intellectual comprehensive capacity, and I’m like, but this is you know, this is the reality of it, and all I can do as a human is give my best version of it, and so I’m doing what I can to adjust it and change it and be honest with what my limitations are and say, this is what I’ve found, and this evidence suggests this, and alter it to that amount. but it’s just yeah, it’s a bit overwhelming and it’s a bit flattening. But what I do know is I just stick to me being myself and being honest and finding out what I’m finding out and implementing that. And so I can’t really go wrong.

I can kind of be a bit behind people, but at least I’m moving forward with solid knowledge, not you know, sort of blagging, pretending I know pretending I know that. Because you’re always going to be insecure, you’re always gonna feel on the back foot, at least if I might not know as much as other people, or seemingly know as much as people, when you question me on things, you don’t see uncomfortableness, you don’t see it’s like okay, that’s someone that you can see has spent the hundreds of hours doing what needs to be done. And so I kind of something in me thinks that’ll win in the long run.

43:41

Why AI May Reduce the Impact of Black-Hat SEO

Steven Dunlop — Senior Account Manager, Epresence.ie:

Because that’s just I do think it is cool because up up until A, I think Google was fighting a little bit of a busy battle against black-hat techniques and black. Yes, so yes, they tried all sorts of cool updates and things to sort of combat it on never really worked completely, as I said earlier. Yeah, you would. Yeah. Do a certain Google search, you’ll find a website coming up that goes against all the best practice advice, but it’s there in the number one ranking for some obscure reasons. I think with AI, it’s going to shift things to the point where the professionals like ourselves are hard work and learning will pay off in the area. So we see the benefits and the black category.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah.

Steven Dunlop — Senior Account Manager, Epresence.ie:

AI is much faster than picking up on black-hat techniques and eradicating it much faster.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, definitely. I know that to be true just from understanding the algorithm, you know, like we was touching on a bit earlier, wasn’t it? That the fact that truth, honesty, is now going to prevail in the search results.

Steven Dunlop — Senior Account Manager, Epresence.ie:

trustworthiness is really a big word.

44:51

How Epresence Builds Trust With AI Engines

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yes. I mean, maybe touching on that before we’ll bring things to a pause, as I’m sure you have clients and things to serve. In regards with trust with yourselves, or I suppose it’s best to kind of move into Epresence because that’s the that’s your heart, so to speak. what what have you done to prove to AI engines that you are indeed that entity that they can trust and therefore surface your content in all AI platforms in regards to users, question stroke queries? yeah. Can you give us a quick rundown on that?

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yep.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Again, multi-frasted a lot of one single thing. upgrading upgraded up to upgrading up to his profiles, truth qualifications, his stereotype achievements. then citing facts and yeah, linking to the research. Big thing I guess doing the polo research.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yes, first quality evidence.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Absolutely love unique research. You’re getting more joy out of how having done that one in-depth case study, factual, unique Irish base, because yeah, we we’re based in Ireland. I find a lot of the data that was always based for the US. Yes. And in the smaller countries, we do need to do a lot of our own research. That’s AI platform, absolutely love it. So it’s just not one thing, but we’re looking at the different aspects of putting it all together. As I said earlier, linking particularly my profile to the SEO, AEO, other colleagues, I linked their profiles to their specialities.

it’s just a case of building up that trust through expertise, experience, authority, and then overall the trust. Following the better practices and proving to Google and AI platforms that we do know what we’re talking about.

47:00

E-E-A-T, Independent Mentions and Search Everywhere Optimisation

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yes. And kind of feeds in a little bit for the. Yeah. And it feeds a little bit what you said earlier, wasn’t it? The old classic E-E-A-T. But it is, you know, it’s still pertinent today. It’s SEO, but you know. Yes. Actually, yeah, no, you’re right. Yes. I only say that because I think there’s a lot of the consciousness, or my assumptive consciousness of other people, where they’re kind of thinking it’s more into GEO and somewhat SEO. Certain practices are not so important anymore. But the truth is they’re kind of they’re more important in certain ways because you’re not gotta just do them like you did before, but you’ve got to do them more diligently and with more depth.

So they’re actually become more pronounced. so yeah, maybe that’s yeah, that’s the truth of it. And that, yeah, those are key factors for trust. And I think also the third-party independent mentions, isn’t it? I think that’s right up there in the AI engines criteria of trust.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Like you refer to it as as SEO’s moved on to just basically engine optimization now. It’s no longer search engine optimization, it’s search all and everywhere.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, search everywhere optimization.

Steven Dunlop — Senior Account Manager, Epresence.ie:

The more you can get mentioned on other platforms. AI doesn’t just search Google’s platform that searches everything. Yes, like like YouTube. Reddit is another big one, but probably more for America, but it’s another big one that gets mentioned a lot. The more anxious you can get, and the difference is in AI versus traditional SEO, it doesn’t have to be a bad thing. Yes, yeah, and if it’s a mention is tied to an authority in the industry, that’s easy to do data.

48:48

SEO Authority vs GEO Authority

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Well, it’s interesting when you say that because it comes to my mind I was researching this because you’ve got the classic what’s perceived, you know, built well perceived is as an established metric, which is like your SEO authority, you know, backlink authority, but there’s also a GEO authority, discern it very different from SEO, because SEO we know is the backlinks, and with the AI, it’s more about the quality of your content. And so that’s why I say it’s like your GEO authority. It’s you know, how unique is your content, you know. Are you backing it up with facts and figures and statistics? Do you have videos to show you doing the testing live for verification and trust?

All of these things are considered by the AI engine. And so then that gives you what I call kind of like a GEO authority. This isn’t a metric that’s been established, but I personally believe it will ultimately at one point. And I can say that because my company is beating, has been beating for the best part of nine months, five well-known established brands that have got five, 10, 15 years of experience. And on the 10 prompt set that I’m testing on, I’ve beaten them for that whole period of time. And we only inaugurated like nearly just under two years ago. So there’s no way we should be doing that based on the backlink the old SEO authority methodology.

It’s only the fact that because on the GEO authority, which is my sort of created term to explain it, is the optimization aka GEO that’s enabling me to supersede them. So it’s definitely a thing, you know.

50:36

Why Unique Research and First-Party Evidence Matter

Steven Dunlop — Senior Account Manager, Epresence.ie:

You use the word unique, it boils down to the fact that AI is unlikely to reference or cite content it could create itself.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, confident content they call it, isn’t it? Yes.

Steven Dunlop — Senior Account Manager, Epresence.ie:

So the more unique it is, and the more personal it is to you, and the more personal proof you can provide your own research.

Steven Dunlop — Senior Account Manager, Epresence.ie:

AI loves it.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah. And I think as well, because if you if you look at it, you’re expanding and breaching out, enriching the knowledge graph. Do you see what I mean? Because that particular experiment you’ve done, nobody’s specifically done the Epresence testing recency when they done an update, what amount of citations and brand mentions did they get? And it’s an island-based company, and you’re against these particular competitors in your benchmark. No one in the world has got that unique evidence or information. So it’s adding to the knowledge graph of information. So it kind of, you know, the GEO signals, as they call it, the AI engines are dun dun dun.

Okay, this is this is fascinating, and it’s gonna harp onto you because it’s something new, and it wants to consume that to learn to grow and expand.

Steven Dunlop — Senior Account Manager, Epresence.ie:

So a lot of the bigger research is it’s as I said, it is predominantly done in the US.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yes. Yeah.

Steven Dunlop — Senior Account Manager, Epresence.ie:

so if if like a car from the UK and I saw advice in, yeah, we can do our own research here, which I have been doing for the Irish market. Yeah, it’s great if you get sighted.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yes. And specifically, arguably, you can say in your country, you would are you’d likely dominate because you’re going to be one of very few people doing it. But then that will transmute to some degree internationally, if it’s really pertinent to a specific query that’s made, you know, in an alternate country. So, yeah, that is, you know, that’s what I understand to be as well the most effective implementation of GEO is doing what we’re discussing, you know, first party evidence of testings. And obviously, it’s got to be a lot of like, you know, factually verified and statistics with figures and tables and things like this. This is what the AI engines love.

Steven Dunlop — Senior Account Manager, Epresence.ie:

It’s they’re like mathematicians, aren’t they? Yes. that’s how they are.

52:57

GEO as an Ongoing Practice

Steven Dunlop — Senior Account Manager, Epresence.ie:

It’s very interesting because I’ve actually been enjoying learning it and practicing it. And the word practicing is important because I don’t think you’ll ever have extra point in the every. It’s gonna keep adapting itself, so we’re gonna just keep up.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, but you can see in it as you go along and you kind of figure things out. Have you had certain moments where you kind of step back and get an orbital view? And you can kind of see the way that they’ve designed the algorithm with what it does, and you’re like, that is ingenious. Because it all makes sense, doesn’t it? You know, with how it assesses trust and how it gives sort of priority for particular things. When you actually break it all down, you’re like, that makes complete and utter sense. Like if it’s on a third-party website, obviously a first-party website’s gonna say we’re the greatest thing ever.

Yes, but 20 third-party websites saying, you know, Steve at Dunlop at Epresence is fantastic, you know, he’s done amazing work. That that is a lot more believable. If you’re asking me to hench my bets, who would I go for? So look at it behind it, the architecture of it, and but this is one little example, but there’s thousands of them that they do that they’ve embedded in the directives in the algorithms, and yeah, I just you know, it’s just incredible.

54:24

Different AI Platforms Use Different Sources

Steven Dunlop — Senior Account Manager, Epresence.ie:

It it it’s what’s even more interesting is that you’re having to sort of develop and design it around, keeping in mind that the different AI platforms have different sources of information.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

I and their algorithms are slightly altered as well, dependent on the proprietor’s preference.

Steven Dunlop — Senior Account Manager, Epresence.ie:

They’ll use the same databases.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, like one thing, for example, I noticed, right, on Google AI Mode, they’re very comfortable to cite first party content. Very comfortable.

Steven Dunlop — Senior Account Manager, Epresence.ie:

Because it accesses their own database, yeah.

55:03

Google AI Mode vs ChatGPT Citation Behaviour

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, but they’re they’re happy to do that. They don’t seem to have the necessity of to you know get a third-party source. And what I mean by that is they’ll cite my company all the time directly from my links in regards to particular queries. When I put the same query in ChatGPT, I will only surface because somebody mentioned my company in reference to something. So you can see the algorithm directives are different there. Like Google AI Mode, I’m like, yeah, thank you. ChatGPT, I’m like, oh come on, man. Like you’re making this incredibly complicated for me. But there you have it. So you’ve got to know the nuances between each specific engine as to how it operates and then adjust for that.

But you know.

55:49

When AI Uses Live Web Retrieval

Steven Dunlop — Senior Account Manager, Epresence.ie:

Yeah, in any way, so far. You’ll get certain prompts that are time-based, in which case they are therefore make use of its own database. Yeah. But if I’d be a different prompt that is time-based, it’d actually jump on the live internet and do a search to find an instance. So if I say, Yeah, what is the process of photosynthesis? That’s a known factor. It’s not going to change, it’s quite that in its database. Well, what time a moves on at a surface cinema? It’s got to go out to the internet to find that information. So yeah, the prompt as you where it goes, yeah.

56:21

Retrieval-Augmented Generation and Grounded Content

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, retrieval augmented generation. And they call it as well, it’s like grounded content, because you’ve got confident content, like you’re saying, it already knows the knowledge, so it won’t externally seek information. So, what you’re pointing to is when it’s going out, it’s called grounded content programming. So when it gets a question, it says, I don’t have enough content to answer this. I need to ground it. So it looks for grounded content, which essentially is content that will be given that is backed up with a citation, and that’s another embedded directive in their algorithms. So yeah, there’s incredible amounts to it.

Anyway, Steven, my mind you’re saying to me now, I think me and you just chat and chat and just go on. but I’m mindful that you do have other things to do. so I will lovingly bring things to a pause. But thank you so much for joining me. I appreciate the information.

57:17

Closing Thoughts and Advice for Getting Out of Your Comfort Zone

Steven Dunlop — Senior Account Manager, Epresence.ie:

Thanks for getting me out of my comfort zone.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Hey, me as well. It’s an ongoing process, you know, and dive in, I will say, because even with me, you know, I think this is now like maybe the fourth one I’ve done, you know. But from the first one was a bit like yeah, but now it just does it. So it’s like you know, never underestimate the adaptability of what you organically have. You don’t need to do anything, your system will do it automatically, just trust in it, and that’s what I’ve found, and it just does it, you know, and then you come away and you think, Jesus, wow, like I’m more capable than I realized. But all you had to do was basically just say to yourself, I’m gonna trust in myself. Hard thing to do, but that’s what I’ve done.

so yeah, hopefully, you know, if that advice is useful. What’s that show?

Steven Dunlop — Senior Account Manager, Epresence.ie:

You’ll be the inner cycle of two years. Well, as in podcasting, well, as in your current sort of Yes.

Paul Rowe — Founder, Chief Generative Engine Optimisation Officer & CEO, NeuralAdX Ltd:

Yeah, just shy from March 2025. So yeah, just shy of two years. Let’s keep the two years in there. Sounds more sounds more. Yes. Okay, yes. So thank you, thank you, thank you. And yeah, I’ll speak to you soon. Take care. See you later. Cheers, man. Bye, folks.

Evidence, methodology and further learning

Research, Methodology and AI Search Resources

Follow the topics discussed in this episode into NeuralAdX’s GEO methodology, academic research base, AI visibility benchmarks, platform optimisation guidance and related terminology. Steven Dunlop’s company website is also linked as the episode’s external guest resource.

11-Factor GEO Methodology

Explore NeuralAdX’s framework covering citation, statistics, quotations, clarity, fluency, authority, schema, recency, author bios, source diversity and technical terminology.

Open Methodology

Academic Foundations

Review the research base used to develop and update the NeuralAdX approach to Generative Engine Optimisation.

Open Research

AI Citation Benchmark

Track domain citations and citation share through repeated longitudinal monitoring across GEO-intent prompts.

Open Benchmark

AI Visibility & Share of Voice

Review brand mentions, answer coverage, average position and share of voice across tracked AI answers.

Open Benchmark

GEO Explainer

A practical introduction to how Generative Engine Optimisation differs from and overlaps with traditional SEO.

Open Explainer

GEO Glossary

Definitions for AI retrieval, citations, entities, visibility, answer engines and related GEO terminology.

Open Glossary

AI Platform Optimisation Guides

Platform-specific guidance covering ChatGPT, Google AI Mode, Copilot, Gemini, Claude, Perplexity, Grok, Meta AI and DeepSeek.

Open Guides

Steven Dunlop · Epresence.ie

Visit Epresence.ie, where Steven Dunlop works as Senior Account Manager across digital marketing, SEO and AI-search-related strategy.

Visit Epresence.ie

Frequently asked questions

Questions About Semantic Content Networks and AI Search

What is a semantic content network?

A semantic content network is a structured set of related website pages whose topics, entities, internal links and hierarchy make the relationships between subjects explicit. In the episode, Paul Rowe and Steven Dunlop compare this to a house: the website is the house, pages are rooms, the contents of each room help define what it is, and internal links act like doors connecting relevant rooms.

How can semantic content networks help AI search?

They can make a website easier for AI systems to interpret by clarifying which pages are primary topics, which pages support them, how entities relate to one another and where deeper subtopics sit. This can improve semantic clarity and create more opportunities for retrieval across related prompts.

What is query fan-out?

Query fan-out describes the expansion of a user query into multiple related searches or retrieval operations. The episode explains why deeper, semantically connected content can help a site remain relevant across those related sub-questions.

How did Epresence improve its measured AI visibility?

Steven Dunlop describes restructuring service and supporting content into clearer semantic relationships, while also improving areas such as schema, E-E-A-T, specialist profiles and AI-retrieval-focused page structure. He reports an approximately 50% increase in measured AI visibility over one month.

Why do author bios and team profiles matter for AI search?

The episode argues that expert identity has become more important because AI systems need to understand who created information, what they specialise in and whether their expertise is corroborated elsewhere. Clear bios, qualifications, specialist pages and external mentions can all contribute to that picture.

Why is first-party research valuable for GEO?

Original research creates information that is specific to the organisation and may not exist elsewhere. Steven highlights Ireland-specific research as an example of filling a local evidence gap, while Paul discusses the value of tests, benchmarks, statistics and verification.

Does GEO replace SEO?

No. Both speakers describe substantial overlap between SEO and GEO. Technical accessibility, page quality, internal linking, authority and structured content remain important, while GEO adds additional focus on AI retrieval, answer selection, citations, entity clarity and visibility across AI platforms.

Do Google AI Mode and ChatGPT behave the same way?

No. The episode discusses observed differences between platforms, including first-party citation behaviour, use of external sources and when a system needs live web retrieval. Optimisation therefore requires platform-aware testing rather than assuming every AI engine behaves identically.

Explore NeuralAdX GEO Research and Evidence

Continue from the podcast into NeuralAdX’s methodology, benchmarks, platform guides, explainer content and ongoing research into AI search visibility.