Is Your AI Visibility Problem Actually a Messaging Problem?

📬 You like this podcast? You’ll love the newsletter.
Join the weekly 3-2-1 on AI search + marketing: subscribe

Most brands treat AI search visibility as a technical problem. Add the schema, clean up the structure, fix the crawl paths, wait to get cited. Then nothing happens.

David Kirkdorffer has been doing B2B marketing since the nineties and now works with CMOs, CROs, and CEOs on exactly this. His argument: the technical layer is a building block, not the answer. What actually gets pulled into an AI-generated answer is the language on your page — and if five different teams are describing your company five slightly different ways, you've given the model nothing to be confident about.

In this conversation, David and I get into why messaging drift quietly moves your brand into a different category, why product marketing may be the highest-leverage seat in your AI visibility strategy, and why large enterprises are often better at this than scrappy startups (for reasons that have nothing to do with budget).

What we cover

• Why SEO gets you on the shelf but doesn't get you quoted
• "Word math" — how small wording changes move your brand into a different semantic category without anyone noticing
• The hub-and-spoke model: product marketing as the source of messaging for PR, demand, partner, and sales
• Corroboration as a trust signal, and why consistency across surfaces beats volume
• Why big companies quietly solved this decades ago, and what small brands can borrow
• The digital footprint gap, and why challenger brands compete against a mountain of existing content
• How LLMs decompose a query, fan it out, and let retrieved chunks compete on completeness and semantic alignment
• Why efficiency is a retrieval criterion most people ignore
• The hiring problem: why "AI search optimization" job descriptions are mostly technical SEO
• What happens when your predecessor's positioning is still live across the internet

I'm an AI search visibility consultant and a fractional content strategist for startups and enterprise brands. If that sounds like the kind of help you're looking for, email me at cassie@cassieclarkmarketing.com. 

Or request your 7-Day AI Search Visibility Audit: https://cassieclarkmarketing.com/ai-search-visibility-audit/

Let’s connect:

LinkedIn → Cassie Clark | AI Search Visibility Consultant
Website → https://cassieclarkmarketing.com

Transcript

SPEAKER_02 0:32

Hey, welcome back to Found and AI. I'm Cassie Clark, an AI Search Visibility Consultant and host of a show where we figure out how bad it actually get found inside AI to write the answers. Today's guest has been doing B2B marketing since the 90s, which means he was thinking about discoverability before I had even used searching and telling you my answer for a second. Definitely not needed to get over your email spent and days with the CMOs, the CROs, and CEOs, all the exact problem we talk about here every single week. Why isn't my brand showing up? What makes this one different though is to where the conversation goes. Most CEO conversations kind of stay on the technical side with schema, structure, callability, that kind of thing. This one gets into that language layer. Like the actual words on your page, why changing them to move your brand into a different category without anyone noticing. We also get into hiring because if you've read any AI search optimization job description lately, you've probably noticed the same thing I have. There are all this technical STL in disguise. And while you do need technical STL, that is not all of it. Fair warning, we go deep. David will tell you himself more than once that he's gone on a tangent. David is amazing to stay with it. The tangents are the good part. Period.

SPEAKER_00 2:00

Hi, my name is David Kirchdoffer. I'm a fractional marketer. I've been focused on B2B marketing since um the 90s. And uh I live in Boston. I play guitar. Um and I'm currently doing a lot of work with uh businesses and CMOs, CROs, and CEOs, um, helping them with their brands discoverability in LLMs so that when prospects and and customers are trying to frame how to evaluate a product set or a category, their brands are showing up.

SPEAKER_02 2:40

So you and I connected a couple months ago. I mean, I don't know how long at this point. It was spring, maybe? Maybe? I don't know. You left you left a LinkedIn comment and I thought he knows, he he gets it. So when you say that you're talking with the CMOs, CEOs, CROs, why are you talking with them and not just not just solely the CMO? You've got the whole gang together.

SPEAKER_00 3:03

Well, it's different people in different organizations. So generally there's a there's someone who has a keen interest. Um so in some organizations, it's just different people. There's um it it is a uh so that's the reason why it's just happenstance different folks. Though I would say it's good that I can speak with any of those different individuals or different roles, because to me, the problem of discoverability, mentionability, and citability in LLMs is a really huge problem for brands, especially for those that are not showing up, uh, that are not turning out. And so it impacts the um a CEO in that my brand isn't visible, and this is the place where things are being learned. So I'm not in the conversation. How can we become part of the conversation? CRO, of course, is concerned because, well, I'm responsible for the revenue of our organization. And if we can't fill a pipeline with people who know who we are, it's gonna be tough. And CMOs, of course, for the same reason essentially. Um, so that's that's why those different roles would be concerned.

SPEAKER_02 4:18

So you're working mostly from what I understand from the top down, correct?

SPEAKER_00 4:24

Well, so I I view the problem of visibility in large language models to be a combination of two things. There is a technical component, um, which we might oversimplistically call SEO. So search engine optimization. And search engine optimization is a vital kind of building block. You're not going to do well in geo, and I call it geo because to me it's about the generative aspect of things, which is kind of really important to the overall understanding of the picture, in my view. You're not going to do well in geo unless you have a certain robustness or kind of completeness in your SEO. We can kind of come back to that. But the other half, which I see very few people talking about, is the LLM, the large language model side of things. And so LLMs do word math. And word math are these vectors and things. We've got you know, we'll kind of maybe understand it, maybe we don't. That's fine. Let's go with a simple idea. It's word math. Certain words add up to mean certain things. Change the words, and we change the meaning. So interesting. We can say the same thing with different words and still say the same, have the same meaning. But if we change the words enough, we jump out of one kind of meaning into another kind of meaning. And we all know this example. For example, we're we're writing something, and we want to replace the word uh I don't know, maybe we would like to replace the word compounds, because the word compounds has come into our language in a way that we've never seen before. And now we're starting to use it ourselves, right? And we realize, oh, I've used that word three times already. So you pull out a thesaurus, you know, your online thesaurus, is like, what's another word? And you get a list. And there are some words on the list that you discover kind of fit your situation, your application of the sentence and what you're trying to say. And then there are others on the list that really don't. And you're like, okay, well, they're there. I don't know. Can't imagine why I would use those words, but they were part of the chain of words and kind of shifting meaning from this end to that end, so to speak. So if we were to use one of the wrong words, let's just call it, and drop it in, it would very much change the meaning of our sentence. What does that mean to you and me? It means that the understanding would be different. We would have jumped from one semantic meaning into another semantic meaning. So we know this feeling, we know this how it works, because we intuitively just go to the right word, right? Same thing happens with LLMs. So when we have our hero section on our website and we're looking for, you know, language market fit, and we're we're we're early stage and we haven't quite found what resonates yet, or we're searching for what could resonate better, we're often changing words. And sometimes we change words and it helps clarify, and we haven't really shifted our category because the words have been within a tight orbit of meaning. Other times we legitimately seek to change the words a bit more dramatically to try and find where that hit would might be, right? When we do it knowingly, when we do it unknowingly, surprises. So why this, why that? And so why do so to come back to your answer? I look at things from a technical standpoint. Yes, the SEO matters, but I'm also looking at the words on the page. And so category and positioning and messaging becomes very important. And I know this has been a little bit of a long answer, but I hope everyone's you know bearing with me. I'm entertaining enough to keep it going. Um, that means to me it's your product marketing team that makes the biggest impact. If we think of a larger marketing team, just that it because this makes it easier for the example, right? We've got a full marketing team, we've got the full all the all the roles of marketing have a small team or a group of people, right? We could imagine a hub and spoke environment where the hub is the product marketing folks who provide the messaging, who provide the positioning, the the planks of the positioning to all the other groups, the the content people, the demand people, the partner people, the sales people. So that, you know, it's not that they're more important than all those roles. Those roles are all important too. That's not my point. But when we start with a common set of words and meanings, and everyone sticks to them and then do their piece of the marketing puzzle, speaking to their constituents who are the natural consumers of their work, who consume their work in the, I'm going to use the word containers of their work, and that container has the message in it. So clever words. Let's use, let's get real. So when product marketing provides to the PR team, this is the plank, these are the things, and PR teams create the containers that PR teams do, which is opportunities for executives to speak with people, press releases, various other kinds of announcement type of related activity, those are the containers, the things that they create, they push message into the market. The demand team have containers, they've got landing pages, they've got um ads, they've got all sorts of other things that they put their their, they kind of use that to communicate for their portion of what they do. But what's inside the contents in the container, very generic term container, I know it's got to find a better word. Um, if that can be consistent, then that helps with the AI discoverability because one of the core things that you all know is corroboration. So when the same message is is is perceived by the LM to be present in many, many places, that tends to help the LLM corroborate, feel like okay, there's corroboration for what this is. So that was a long answer. I'm so sorry. Let me wrap it up. SEO is important, words are important. SEO is a building block, but the the words it's the words on the page are become what is quoted.

SPEAKER_02 11:07

That's long answer, sure, but also very detailed. And I really like how you're starting with product marketing and getting everyone on board with the thing that we are saying, this is what we're saying. I've been calling that the strategic source of truth. Have you been noticing that brands don't have this? A couple of the brands I've been working with, they just everyone's working in their own different, their own different way. No one's really cohesive. How are you fixing that or how are you recommending that brands fix that?

SPEAKER_00 11:38

So if you work in a big company with a large team, the larger the company, the better they're doing at it, ironically. Because they have always had it. I remember when I was working at a company called Computer Associates, at that time it was a $4 billion company. And by today's standards, that might not seem like a lot, but by the standards of the day, it was one of the largest um software companies, the top 10 software company in the world. There were over 450 people in the marketing department, and so and there were four business lines, and each business line had product groups, and the product groups gave to everybody else in the marketing teams, the shared services as they might be known, the messaging to use. That was a way of the hub and spoke a little bit. And anything that was published, and I was in the kind of field marketing demand journey, right? Hundred people in that group. Um and I was anything I would want to publish or share or put outside of the walls, figuratively speaking, went through what we we kind of disparagingly call the brand police. And the brand police would evaluate on two things. Are you using the prescribed imagery that the company is using? We've purchased this imagery so that we can have a look and feel. And we're also evaluating on the message. Have you used the message platform that you've been provided so that you're not kind of making stuff up for the risk? Because then you're also going to go through legal, make sure that the certain legal um requirements were added to anything. Because if you made a claim and then you couldn't deliver on it, that was a bad thing. So this is what big companies do. So big companies go through that kind of thing because they they can, they need to, or whatever. And so the fly by night, any odd message getting out there doesn't happen so much there. Now, if you work in a big company, you might call that very frustrating. All that system, there's all those layers, there's all these surface level agreements, there's all this process, and everything's divided out. It's the smaller companies where you can just do whatever a little bit, and I say whatever, it's people who are trying to do the best they can. It's not to try and say um they're doing bad work. They're trying to keep up with the process and what's being asked of them. And in that speed, they also have pressure for results. And the and the the real pressure comes from a messaging standpoint, how a demand gen person or a partner person or a salesperson might take the core message and twist it, not twist it, but ultimately twist it, change it a little bit so that it serves the purpose of their particular need at that moment. I'm speaking to this partner, so I'm going to kind of like lean in more heavily. Leaning in is one thing. Changing the words and changing the meaning is another thing. Um, anyway, um, I've worked at big companies, I've worked at small companies. It's the small companies that have the biggest problem with geo, partly because of the lack of pervasive presence. They just don't have the footprint on the internet, right? Um, if you're competing against Gong, I I mentioned Gong because so many people know Gong, but any particularly well-known brand, they've got a lot of words written about them on a lot of different websites. And your little brand that's challenger brand that might be a better brand, sorry, might be a better technology for today's situation and the things that people um are dealing with today, has to compete against that mountain of digital ink. Um and so now I'm off a bit of tangent. So anyway, I'll stop there.

SPEAKER_02 15:48

Yeah, yeah, but I don't think like the competition for smaller brands, like it might seem like it's a big deal. Like, oh, why are we even bothering? But I've noticed that some of the smaller brands, once they get everyone on board, they're working with that strategic source of truth or whatever we're calling it, it doesn't take much to actually make a difference inside of these answer engines. Have you noticed the same thing?

SPEAKER_00 16:12

You can have an impact that's pretty quick. Yes. LLMs pick up a lot. So there's as most people will know, there's the training data, there's the real-time, you know, um augmented generation data, right? There are the the search indexes, all of these different places are part of the overall picture where an LLN may go to get an answer to uh to respond to the situation you present it. The situations we present it are generally questions, but not always. So some of those um sources pick up information within hours of a page being published or a change being made. And so that can so you have an ability to affect the outcome um quickly. And if you're on the bubble of being mentioned in an answer, then that might help you bust through, so to speak. Um but the other side to this, and again, this is language, it's the quality of the answer. It's the um alignment to the situation the LLM is trying to respond to, most often a question of some type. And so the quality of the answer is the other side of things that I try to make like try to share how and why that how that works and the impact it can make. Um, and I can talk a bit more about that if that would be interesting. Um but just having more stuff written about you doesn't always mean that your stuff, that stuff is the right stuff, relevant stuff. So it's before before going on a content creation spree and paying of anything, you kind of want to know what is what are the characteristics of content that LLM's care for. And that would help because you were like, oh, okay, sure. Now I'll spend the same money and the same energy, but I'll do it in a way that makes more impact.

SPEAKER_02 18:12

Yeah, I agree with you there. I was doing a preliminary audit for a brand a couple weeks ago, and I noticed I was really mapping those citation sources to the prompts that they were trying to win. And I noticed that most of the time the citation sources were from publisher websites. So, you know, brands can go partner with these websites, they might have to pay a fee, but they get their stuff out there. Um, and I'm not sure that a lot of people are talking about that particular strategy yet. Have you have you noticed the same thing?

SPEAKER_00 19:37

Um, I haven't because I'm not involved in that aspect of things with the clients that I'm thinking of. I do a lot of audits and I do a lot of training. So after the training, then they go forward and conquer the universe, of course, right? Um what you're pointing out is in many ways, we're doing the same things as we did pre-AI. And there's a good reason for that. And the simplest reason is that LLMs are like a mirror. And they reflect back what they find. And part of what they reflect back is um they reflect back to each individual who asks a question. Here's what I found based on your situation and context as you've presented it to me. This is what I found that seems to relate. And so having a lot of stuff is now in com let me just let me kind of jump back up a layer. I've got like 12 things in my brain I want to say at once, and they're just like a competing for tongue time. The internet is an information service, and AIs are the user interface to information. And we are in a competition as the individual brands at the idea level. Our ideas are competing, and the ideas that are most relevant to situations become noticed by the LLM. SEO, let's come back to that. SEO is like the Dewey Decimal system in a library. We remember the Dewey Decimal system. Without the Dewey Decimal system, we couldn't find a book to save our lives. And we just go walking up and down trying to find stuff. So it's vital. Without a Dewey decimal system, we can't find stuff. We can't categorize it. We can't know what is what. Where should it go? When a new thing happens and it gets put on a shelf, where would it go? That's what the Dewey decimal system is brilliant at, right? Right. But when we were doing our research reports, remember when we were in college, or perhaps you're still in college and you're listening to this, you don't actually cite the Dewey decimal numbers on the spine of the book in your response, in your answer, in your paper. You cite the words in the book. That's the stuff. The stuff is in the book, and the stuff in the book is what's presented on the HTML page that the user sees. So SEO can help get you be found on the shelf, but it doesn't help get you mentioned or cited. So the mentioned and cited part is the important bit. And I talked about the competition of ideas. We have to understand that LLMs will take our initial situation, many times a query of some kind, decompose it into what it feels we're asking for, understand the intent of what our try are we trying to get, and then we are all now familiar with fan out query or query fan out. And that's the LLM saying, well, I could do a search for this, but you know, when they say best holiday in Europe, I don't really know. Is that best based on price? Best based on weather, best based on what? So what I'll do is I'll look for best based on a lot of different criteria. Because somewhere in all of those different things that I find will be a piece of the answer. And we know this experience when we do a a Google query. We get a bunch of blue links. We take a look inside, grab the bit that was relevant to what we cared about, and they put it in a document, and then keep on going down until we find something meaty enough that we like. Elements are doing the same thing. They're grabbing chunks of information from lots of different places, and then they compete those chunks. And the chunks that are in the competition is on um completeness, contextual completeness, um, semantic alignment to the situation, many times a query. And so out of the hundred plus things that are found, three to four survive the competition of relevancy and alignment. And then those things become mentioned become part of how the answer is constructed. And by the way, all of this happens in milliseconds, which is you know wonderful. I mean, do a query looking for something on Google search or Bing or Brave. Or do it in an LLM, pick your fate. And it's so much faster the other way, isn't it? Right? It's like, oh good lord, it's so much so much better. But it happens in milliseconds. So the other thing that's really important is efficiency. And efficiency is how LLMs decide what gets is relevant. If it takes too long to find it, if it takes too long to understand what it means, if I pick it up and the context isn't complete within the piece that I picked up, I can't reconstruct it. I don't have time for that. So efficiency is um measured in different ways along the whole retrieval and synthesis process. Okay, I've gone on a bit of a twist here. Sorry. Um but it has to do with the words on the page, the context that we create, the relevance that we create. I've mentioned words like chunks. I haven't mentioned the word windowing, perhaps not as important as chunks. Um anyway, let me let me kind of pull back up. Help, help. I'm drowning here in too many detail. Um, how how can I what what what else is that would be useful? Where where should we go next?

SPEAKER_02 25:35

Well, you've mentioned that the SEO layer is super, super important. So you also mentioned that you've been talking with CEOs, CROs, CMOs, everyone. Have you noticed, and and I'm only bringing this up because I have noticed so many brands have been hiring for someone to do AI search optimization for them full time. And if you read the job description, it's mostly technical SEO. So have you how are you explaining that to you know, C-suite level that, yeah, SEO is important, but so are these other things.

SPEAKER_00 26:10

Well, I start by trying to say it's a question of language. There's two sides to it. There's what's on the page and being said about your brand and how you say it and all that. And then there's the technical behind the scenes, um, what happens on the HTML side of things. And in large companies, it's easier because they separate the roles. The SEO team in a very large company is maybe, I don't know, 10%, 15%, 20% of the overall marketing team. And that person is never choosing the category or choosing the the the positioning and messaging language, they wouldn't that wouldn't be part of what they would expect that they would do. And yet it's that messaging and positioning that actually has a big impact on SEO, sorry, on geo success too. Um, and so I um when when I see so and and because SEO sale um search engine optimization is such a vital component, it's not wrong that someone would have an SEO background. It's just incomplete and it's not deal it's not addressing the other side. And in all fairness to all the SEO professionals who might be listening to this, I'm not here to say that SEO work is not important. It's vitally important. It's it's it's ridiculously vitally important, right? It's just that it's not enough because we're competing in ideas, and so the quality of the idea is the is the bigger challenge. And so that comes back to well, that's just back to good old-fashioned marketing, right? I said, yeah, it kind of is. And so it's owned by the entire go-to-market who is creating a digital footprint of who you are and what you are. And the the other thing I would mention to folks who are in seat now working at companies, I'm in a marketing role, I'm in a CRO role, I'm in a CMO role, I might be a CEO who's been there for a while or replacing someone, is the work that was done by your predecessor, if there was one, and the preceding teams, individuals, is still out there on the internet in many cases. So LLMs will find information from everywhere, cool, but that information captures moments from every when. So if your brand is repositioned, and as many brands will, right, um, and you did a certain amount of work to try and update the key places with the new positioning. Well, we used to say we were this, but now we're kind of saying this. Could you update your thing? Thanks. That'd be awesome. Um, and you spend a bit of time doing it, you obviously triage to the places that might have the biggest impact within your information ecosystem. Because every solution and brand kind of lives within its own competitive ecosystem. So the things that you do with brand X and its competitive ecosystem of information compared to brand Y, that's in a different category, may or may not work. The same things won't apply the same way because they compete in a different ecosystem of available options and available information. Um I lost the thread of where I wanted to go with that. I'm sorry. But the the the the the way to describe the difference between geo and SEO is to say that SEO is pretty much purely the technical side and doesn't impact the positioning, the the the market messaging, whereas geo is impacted by that stuff too, which makes it to me a responsibility of the go-to-market leadership to make sure that everyone's coordinated on the same story and message that they don't um inadvertently um in in their desire to do well within their own KPI set, within their own audience, within their own kind of lane, deviate beyond a certain amount. Because when that happens, you jump out of one semantic meaning and into another semantic meaning. So this is all very nebulous to people, right? This all sounds very nebulous. Like, how do I do that? How do I get well it's a little bit more policing by the CMO, hey guys, that isn't the right message. It's it's it's checking on things, which feels like, oh, I don't want to be a big company guy. It's like, well, okay, that's fine, but don't think of it as being a big company guy. Just think of it as being a company guy. Um it's uh so that that's a big start, is just recognizing it's not necessary challenge, it's everybody's challenge.

SPEAKER_02 30:55

Okay, that's David. He was on LinkedIn. I will drop the link to his profile in the show notes. Here's what I want you to take from this episode, though. This is not a huge project, maybe about 15-20 minutes of your time. David's point about corroboration is the whole point of this episode. If one word can describe this episode, it is corroboration. These LLMs get confident about your brand when they see the same description of you in a lot of different places, which means the fastest thing that you can fix this week isn't your content volume, it's just your consistency. So if I were thinking about this, here's what I would do. Open a blank document, I'd pull the one sentence description of my company from five different places. I would look at my homepage hero, I would write that down. The LinkedIn company about section, write that down. I'd go look at my most recent press release boilerplane. I'd also look at sales one pagers or pitch text, whatever the last one is, and then I'd go look at one third-party surface. That could be a directory listing, a review site, or a partner page, just whatever's out there with my brand name on it. Then I would read straight through all five sentences, one right after the other. If they're saying the same thing in slightly different words, you're in good shape. Personally, I would make sure that the positioning in the sentence is the exact same across every surface, but if they're the same but with slightly different words, you're most likely okay. But if even one of those sentences puts you into a different category than the others, maybe the noun is different, it's a different pie or a different problem. You've just found the reason an AI engine can't describe you confidently. That's the semantic drift David is talking about, and it's happening inside your own marketing team. One of those sentences off might be that it's off across every surface that you touch, so it's something to go look into for sure. Most of this you can fix in an afternoon, and it costs you absolutely nothing to log into those accounts and fix it. If you want more of this, the small specific things that actually help with your visibility, that's what I send every Friday in the visibility report. The link is in the show notes. It's free, and it's the one email I'd want in my inbox if I were doing this kind of work. And Cassie Clark, I will see you on Thursday until then. Stay visible.