Summary

Ben Rees discusses how brands can build visibility in an AI-driven world by creating interconnected content clusters that help AI models form accurate representations of what a company does. He emphasises that authenticity and in-person interactions remain uniquely valuable as AI cannot replicate the genuine human connection that builds trust and drives purchasing decisions.

Find out more about Ben and his work at BJRees.com

 

Key Points

  • AI models build knowledge through networks of interconnected information, requiring businesses to create consistent content across multiple channels that reinforces their brand message [13:38].
  • Brand functions as a "prior" in Bayesian terms—pre-existing beliefs that shape how customers interpret new information, making brand awareness crucial before any marketing activation occurs [10:26] [5:33] [13:07].
  • In-person events and founder-led talks are extraordinarily valuable for small companies because they establish credibility and build prior awareness that pays off months later when purchase triggers occur [15:46] [20:33] [20:51].
  • Consistency across channels, departments, and messaging is essential because conflicting signals confuse AI models and make it harder for them to represent your brand accurately [18:12] [18:35].
  • Tracking systems and data analysis, even simple spreadsheets, enable marketers to understand what's actually working and make informed decisions rather than relying on assumptions [22:30] [23:22].

 

Transcript

Transcipts are auto-generated.

Kiran Kapur (00:01):
Hello and welcome. We are in the world of AI again, but specifically how you build your brand in a world of AI.

Ben Rees (00:10):
You need to build this kind of cluster of information, have lots of things which reinforce each other. So here's a blog post and a podcast and an email, and all of those things reinforce each other. And when you do that, the sort of AI models, they build up their knowledge.

Kiran Kapur (00:25):
And I'm delighted to welcome Ben Rees of bjrees.com. And Ben is a consultant who works in B2B marketing. Ben, welcome. You have quite an interesting background because you have psychology and neural networks studying in your background.

Ben Rees (00:42):
So yeah, I started out as a developer and then I did some neural networks, a master's course, a very, very long time ago. And one of the things I remember them saying at the time was, "We're going to study neural networks, but you'll never see it used in the real world because it's just too difficult and requires too much processing power." And that was about 20, 25 years ago. And now you just have to wait and these things become the most important thing in the world.

Kiran Kapur (01:08):
Absolutely. You have a very strong marketing background as well.

Ben Rees (01:11):
Yeah. Oh yeah. So marketing's been what I've been doing really for 10, 15 years. So prior to that developer, things like I was a product manager, product marketing, and as I say a developer. But yeah, I moved into marketing 10, 11, 12 years ago. And I've always come from a sort of technical background. So that world of quite technical software, that sort of thing is really what I know.

Kiran Kapur (01:33):
And you have a wonderful quote that I found on your website where you said, "AI can copy your words, but it can't copy the things that make people believe you." And you are all about being authentic, aren't you? And that's how you stand out in an AI world.

Ben Rees (01:48):
Yeah. And authentic's a difficult word because a lot of people can say the best way people are really great at being authentic and they can fake authenticity. But I mean directly that the example I gave is that if you are giving a talk in a room to people, then there are no barriers between you and the people and the audience. It's not as though you are reading a LinkedIn article that was written by a machine that was based on a piece of data here, which was then from a podcast two weeks ago. So the lesbians, when you're standing in front of people talking about a thing, there's an absolute direct link. And to me, that's the most powerful marketing. And just to add to that slightly, I think about the things where people have marketed to me and what's worked on me. And I can think of a few examples where I went to a conference and I spoke to the person who built a product and I understood deeply what they were doing and saying, and I ended up handing over money for that product because I spoke to the person who wrote part of the code or whatever.

(02:58):
And that's what I mean is the sort of realness of that connection between you and what the product does.

Kiran Kapur (03:05):
And I suspect reality and that realness is something that as we get more and more digitised and into this sort of AI generated marketing, it's going to become more and more valuable. Exactly. So one of the things you talk about is that your brand and the AI are linked. What do you mean by that?

Ben Rees (03:26):
Well, it's really about brand and how it affects your marketing effectiveness. So I think maybe this is jumping forward slightly, but a lot of work that goes on in marketing is trying to optimise various things. So you do SEO, search engine optimization, a sort of a discipline that's been going for 20, 25 years. And that's this whole world, and there are lots of contracts and people doing that work. But the thing that I've always found missing in a lot of those discussions is brand. And when people talk about the brand and so on, there are all these jokes about, well, nobody really understands it. We just know it's important and so on. And so I thought that's kind of a bit frustrating for me because I need to understand what that means when somebody talks about the brands being important. So I started to do investigations into how do things actually get surfaced?

(04:24):
How do people get seen rather than just trying to do SEO? And what you realise is that there actually, it's incredibly complicated world for how people get thrown up into these things. But if you start thinking in this Bayesian way, which is just a different way of thinking about things, and you think about, okay, why is it that AI models, whether it's Claude or ChatGPT or whatever, why does it quote you or not you? What you realise quite quickly is it's not really about keywords anymore. It's about presenting a picture of what your company does in a hundred different ways. And that is a very tricky deal. But if you get it right, then your brand is represented on the internet, if you like, in multiple ways, in multiple places. All of those things are connected together and so on. And I think just to answer something very specifically, I mean, one of my first slides I talk about this idea of the importance of brand, which is actually most people make their decisions based on they go into a shop, they haven't got time to look at all the different toothpastes.

(05:33):
They just pick the one that they always do. And that brand is, I think, equivalent to what I would call a prior, which is what beliefs and thoughts do people have before they start looking at bits of software or products or whatever you're selling.

Kiran Kapur (05:47):
That's really interesting. And what I like is it works for both B2C. I guess you're quite right. Nobody's got time to stare at toothpaste and try and work out the difference between all the different brands and what they're offering. But it also works very strongly for B2B, which of course is where you do your consultancy work in that B2B, I have to be top of mind.

Ben Rees (06:07):
And where it sort of gets difficult and interesting, top of mind in lots of different places for lots of different people. I mean, that's why I think B2B is, as I say, like an orchestra, because actually you've got all of different parts of a business, marketing departments, product departments, everything, and people might interact with all of those online. They might read a blog and they might watch a video and they might read your articles and all that sort of thing. All of that feeds into this representation of you as an entity, as a company. And that all works together to create, as I say, this kind of representation of what your company does. And then people will come into that in all sorts of different ways. They'll come in, as I say, with Google searches or with perplexity searches or all sorts of things. So in a way, it's quite hard to explain and you'll see that I slightly struggle, but because in a way, what I'm trying to say is the job of being seen by potential customers isn't just the case of optimising SEO.

(07:03):
You need to be optimising SEO and answer engine optimization. And ideally you want to be in GEO and you want to be seen in various LinkedIn and live things and all of those things together hopefully add up to you, somebody giving you a call and saying, "Yeah, what do you guys do? Can we talk to you?"

Kiran Kapur (07:22):
I do love your analogy about conduct an orchestra, not just one instrument. So it's a whole channel. It's everything that we need to look about. And we've talked about marketing being more than one thing for so long and marketers looking after brands which are overall, but still we tend to, in careers in marketing, get very specialised and very into specific areas.

Ben Rees (07:43):
Which is fair in a way. I mean, there is an issue at the moment with jobs and AI taking jobs and so on. And I think there is an issue where you have to keep ahead of these things and understand them and so on. But I suppose that's the one caveat there is that you do, although you're using AI a lot, there's so many things for so many different people to do that. I think there are roles

Kiran Kapur (08:05):
There. Okay. So let's go back to, you talked about brand and AI are linked, and you have a really interesting analogy about how beliefs are created for both human beings, but also for AI. If I've understood it correctly, and correct me if I'm wrong, one of the things we're trying to do with AI is get the AI to find us in so many different ways that it has a belief about who we are. Is that right?

Ben Rees (08:29):
Yeah, that's kind of a way of thinking about it is that the systems, so take a simple one like Claude or ChatGPT or whatever, well, they're not simple, but one of those systems, what it's trying to do is if people are searching for certain things or looking for certain bits of information, it has a representation of that sort of idea or that entity. It's easy if you think of it as a company, a company, it's got these sort of things. And the AI engines, you have to understand how they work under the hood to know what to do. So a really simple example is, is it better to have lots of different messages about what your product is or to be saying the same thing in different places? Is it confusing or not? And so that's a kind of interesting question because you are thinking, if you talk about content writing, should I have my content writers writing all sorts of bits of pieces that will add up or should it be everybody says the same thing?

(09:26):
And it helps answer things like that because really in a way it's neither. You need to have a kind of a spider or a network really where you have all of these terms which are linked together and so on and that reference each other and so on. And if you do that, then when people are doing searches, the systems like ChatGPT or whatever, they're searching through this network for corroborating evidence and is what this person's saying quite right or what's it related to? And there are other blog posts or sorry, other websites which corroborate it and so on. And so you've got this very complex process that AI uses and that these systems use, as I say, at the end of which hopefully they spit out you as an option.

Kiran Kapur (10:14):
You talk about it being a prior.You used the word prior building, which I thought was really interesting. So what do you mean by a prior? And then let's talk about how I actually build one.

Ben Rees (10:26):
Yeah, so prior, it's a Bayesian term. So Bayesian maths, which is just this idea that your belief about something is a combination of two things. It's about what you thought about that thing beforehand and then what evidence you see. So when you put an advert up and your advert says, "We're the best ever toothpaste or whatever," what a lot of marketers do think, right, let's get the messages right. Oh yeah, we're the best toothpaste because of X, Y, and Z, Y and Z, right? And they sort of assume that if I get that messaging right, that people will read it and think, wow, and buy your product. The problem is all of those potential customers already have a view about good or bad toothpaste. And so what actually happens, so Bayesian Maths is saying you are combining the previous knowledge people have about things with the new thing, the advert, and it's coming out with a sort of probability at the end of that saying, "Okay, I think they're good or bad." And so it's the way most people think, but if you go into the shop to buy toothpaste and there's a whole bunch of toothpaste, you've never heard of any of them, you've never heard of any of them, the probability of you buying them is still very low, even if everything on the back of the packet is great.

(11:34):
And the basian stuff is like, well, how do those get built up? What can you do to build up that sort of prior so that you are there in the brain? And the problem for small companies in particular, if you take a company like Microsoft, their prior in the world is enormous. I mean, there's often this joke about if you're at Microsoft or Apple, why bother advertising? They all know you, and that's because their prior is more or less 100%. But if you are the tiny, tiny little company, like me, a small little sort of company, I'm starting with more or less 0%, like 1% or something. And I think if you're a marketing department, if you're already at 100, I mean, I don't know what it's like to be in the marketing department at Apple or something, but I feel like job done, tick, job done, let's do something else.

(12:21):
Whereas if you're a small company or just getting going, that's where it gets more interesting where you have a little bit of traction and how do you build on that traction and how do you get into those kind of AEO answers so that you get not just in Google search, but you're also getting picked up on AEO searches, which are things like snippets and things like how do you get into snippets and that sort of thing. And a lot of the work I do is how do you stage by stage increase your visibility in those things when perhaps you've got quite a small team? I mean, a lot of my work is like, you've got a small team or it's just you or there's two of you and you are against this behemoth of Apple or Google, what do you do? What do you do?

(13:03):
And you have to find some way of gradually building up. And that's what the sort of Bayesian approach takes because it gives you this idea of like, okay, how do we get into the priors? How do we get into priors? And that's where you get into these things. I mean, a lot of the work I did, the research I did was as I say, how'd you get into those priors? And it is about you need to build this kind of cluster of information. You need to have lots of things which reinforce each other. So here's a blog post and a podcast and an email spin and all those things. And all of those things reinforce each other. It's not just one-off pieces and so on. And when you do that, the sort of AI models, they build up their knowledge based on pulling those different things together and creating a network of information.

Kiran Kapur (13:48):
You used the phrase snippets, so let's just come back. What are they and why do I want to have them? Because it sounded like I needed to have them.

Ben Rees (13:55):
Oh yeah, no, no, no. Fair enough. If you do a Google search and for almost, and it depends which country you're in because different laws everywhere. But in the UK, if you do a Google search for something, then you get different results. But one of the things you get is in somewhere it's usually in the middle, you'll get a thing which is a snippet or a little quote or line from websites. And so those snippets, and they're very, very important now. And they're things like, "Oh, I've just done a little definition about what GEO is or whatever your company does. I've just done a little definition and it's like three lines or seven lines and it's a really sunk sync thing." And those things can get picked up very easily by these large language models and they are used very easily. And so what you're trying to do is rather than say write a long form piece about something, you might write some snippets which kind of interact with other bits of your website and so on.

Kiran Kapur (14:54):
So we've agreed that we need a prior. Talk me through how I build that. We've talked about having an orchestra and I'm sort of getting some ideas through, but I think there are other things that I could be doing. Is it all about me just getting some PR out?

Ben Rees (15:10):
PR is definitely one of the good ones. I mean, really when I say that, we all know the long list of channels and activities that we could do. I've got a list of 30 or 40 that I pull out from day to day on a piece of paper somewhere. And the question really is, okay, given what I'm trying to do, which of those shall I pull out? And so if I've already got a great brand and everybody knows me, then I'm going to do certain sorts of marketing, trying to activate people who already know me say. But if I've got nothing, what are the activities you should do? And that's where I get to things like in-person events. So I did do the analysis of all my visibility and so on over the last 15 years, in fact, I've got a lot of data and you can see these correlations where I did certain sorts of work, like a talk.

(16:04):
A talk is the best example actually, where you then look at who was there and what they saw and so on. And you see that a talk, there's something about the magic of being there in person. It's something about the credibility. So if somebody is physically there talking, you're like, okay, well that's not a bot, is it? I can see them. And there's something about those sort of real things, which I think is great for building the prior. Now, of course, you might go to an event. As I said, I've got examples of my own background where we've done an event and we've though, oh, we only got one lead out of that. That's not very good. You then, six months later, nine months later, they come back to you. And the problem is you are trying to assign that success to this thing and you can't, you can't.

(16:54):
Whereas if you think of it instead as doing those events are building your prior, they're building this sort of awareness of you in the world such that when there is a trigger in six months time where somebody's, oh, actually I do need to find a thing, you've already built this kind of background awareness, this kind of prior knowledge of something. And the thing is, when I say prior, actually you could just substitute that for brands. And that's really why I think of them as the same thing. It's kind of your previous knowledge of who these people are. But like I say, with priors, what I do now is you can start doing things like what do I appear in? Do I appear in these various search types? And so on.

Kiran Kapur (17:37):
One of the things I noticed about your website is you do have an awful lot of data that you talk about. There's clearly a lot of data behind it. And I think again, that's so important because you say branding can seem very ephemeral. It can seem a bit, oh, well, it's the colouring department fussing about the right colour red or whatever. And I've worked on rebrand projects over the years and that can be what you end up doing. Now I'm terribly sorry, that's the wrong shade of red or that's the wrong font or whatever. And I think therefore branding gets pushed down this sort of it's only about the visuals, but you are pointing out that branding is a much bigger thing than that.

Ben Rees (18:12):
For me, I think of it as what do people know about you before they've even interacted with you? And actually, although we talk about things like colours and writing as not being so important, but the consistency. So one of the things that the AI work shows you is, so when I talk about different entities linking up and you think this and that, a lot of that comes from the consistency between things. And so if you are doing graphic ads and the tone or the impression that people get from those ads is completely different from your writing over here. So say over here, you're a very techy company and over here you're a very human company and those things mismatch, then that confuses these models, these enormous models that get built by the big companies and you're just making your life more difficult. And so the reality is you have to have consistency across different channels and so on, which means you have to have consistency across different people in your organisation, different departments.

(19:09):
So I mean the classic one is I always talk to everybody. I talk to the devs. When I was previous jobs, I would talk to the salespeople almost every day. What are you seeing? What are you hearing? And so that when a salesperson says something, that is backed up by what's on the website because we just had a chat a week ago and therefore that helps to create these kind of systems or these networks of information that the AI engines can pick up more easily because you are being consistent.

Kiran Kapur (19:37):
That's interesting. So one of the things I noticed you talked about in building a prior, you talk about having case studies and that's logical because that's proof that something actually works.

Ben Rees (19:47):
It's real.

Kiran Kapur (19:47):
But you also talk about having visible people, not just brands. Why is that so important?

Ben Rees (19:55):
I don't know about you, but I believe people far more than I do computers and AI and things like that. It does come from specific examples. Like I mentioned before where I've been to conferences and I spoke to somebody who, it was one of the people who started Marketo, I think, and I spoke to him and he explained what they were doing and why they did it. And I was sort of instrumental in pushing for Marketo... No, not Marketo, it was HubSpot in the end actually, sorry, wrong way, but for HubSpot to be taken up twice because really from speaking to person. So the power of that physical interaction was enormous. And then you think about, okay, well, I'm a tiny little company. I'm not online. I haven't got much presence and so on. If you go to an event with your founder giving a talk on something to a room, it's extraordinarily valuable, extraordinarily valuable, far more than spending...

(20:51):
I mean, I guess if you've got very rich investors who are willing to burn lots of money on this stuff, the best way of getting going from nothing really is to be giving talks and going to events and so on or exhibiting events because you are setting the groundwork for the future. You're building up your priors.

Kiran Kapur (21:12):
So I know you work with large companies, but also very small companies. So if I'm sitting as a marketer in a relatively small company and I'm thinking, "Come on, Ben, I've got a to-do list as long as my arm. I've been told I've got to create video and I've got to write blogs and I've got to be out doing podcasts. Now you want me to do in-person events as well, which chances are that's taking me out of the office or it's in an evening." Where would you recommend that I started?

Ben Rees (21:39):
The real answer depends how technical you are in the sense that... I mean, I've built a system which is tracking everything I'm doing. It's tracking my success. It's constantly automatically seeing how I'm doing in Google searches and things like that. And it comes back and says, "Look, you really should be doing much better in Google searches." And in fact, I've got examples where I confidently said, "Oh, here's some great ads. I know what I'm doing." Just relet for about six months, not a single click or anything, and it tells you why and it tells you actually they were searching for this and et cetera, et cetera. So I suppose the first one answer is having systems like that. I've seen people selling them and advertising them. I suppose maybe I should be more trusting. I'm not quite sure I trust the system, which I wasn't involved in building.

(22:30):
So you think you just show me things I want to see. But being able to build some sort of... Even if it's just a spreadsheet where you are tracking everything and you're keeping an eye on the details and what happened here and what happened there, I definitely think I've found when I get my hands dirty and I get into the real data underneath it and I know why my chart is showing X rather than Y, that to me is a really crucial thing. So if you're say starting out and you just don't have any team or anything, I've always spent time in Excel or whatever, truly understanding what's going on.

Kiran Kapur (23:12):
That's really interesting. From somebody that works within AI and is talking about agentic, what I was expecting you to say was and vibe code something, but you didn't, you said Excel.

Ben Rees (23:22):
Well, the vibe coding, I mean, it's a whole different topic in the sense that I definitely have created systems which help me do my job more quickly. And so many people do this, just automatically gather this information together and send me an email every morning telling me about what went wrong yesterday. So people do things like that and that's great. And I definitely have all that stuff. And then you just get into, when I talk about what you should be doing, you can then make that better and better. And I have agents running, doing stuff now and everything, but you need to know what they're doing and what's going on. These things are dangerous. They will give you false information constantly. I mean, the work that I'm doing this week, in fact, is putting in an infrastructure or a harness which says, okay, I'm doing these blog posts and I'm sending them out.

(24:12):
I don't want to manually check that I've done the right thing. So I've got all these tests, these sort of automated tests that say, Ben, you said that three months ago you were going to spend your time working on, I don't know, B2B in small companies, and actually I've just read everything you've written for the last two months and you haven't written anything about that. You've just been writing a bunch of stuff over there. So it's almost like a memory thing. It's almost like a memorise all the work you've done. I suppose to answer the question, if there's just say one or two of you, it is good to get a system like that set up to help you get on top of everything because it's just too difficult.

Kiran Kapur (24:47):
Ben, that was absolutely great. So what I've taken away is that I don't need to worry that AI is not going to take my job because AI can't stand up at an event and do a speech. Is that fair?

Ben Rees (24:58):
Yes, that's pretty much it. Or it can, but it'll be rubbish. That's the point.

Kiran Kapur (25:04):
Ben Rees, who's consultant for B2B marketing, and you can find more about Ben's work, his research, which he publishes and writes about extensively at his website, bjres, that's R-E-E-S.com. Ben, thank you so much for your time and insights. I really enjoyed that.

Ben Rees (25:22):
And thank you. I really enjoyed that.