Jeff Rain (2:39)

Yeah, look, this is the fundamental change and so it's really important to spend a second just really grappling with what it is and what it means in some of the obvious and also some of the non obvious ways. I think we all know how we searched the Web yesterday, I won't say today, because today is already different. But if we go back in time, if we had our Michael J. Fox DeLorean and go back in time, how would we search? We Google. There's no two ways about it. God bless Microsoft and God bless others who are trying to capture the old traditional search. We, we would Google something. And what does that mean? We all know we get those blue links and we go spelunking around the Internet ourselves. It's a DIY project. You give it a couple of words and you go spelunking. Now, unless it's in an industry that Google has decided that they want to control the first third of the screen, but we'll let Jeremy Stapelman and others worry about that fight.

Jordan Cooney (3:45)

That fight.

Jeff Rain (3:45)

Yep, let them worry about that. But you did it yourself. You spelunked around the entry Internet and you've, you looked at the pages that they delivered and in an otherwise potentially infinite list. And you, intrepid researcher, searcher, were in charge of figuring it out. So that's, that's traditional search. You search and then you discover the answer. But now it's ask an answer. Now you prompt in an AI search or LLM, whatever the parlance you want to use, and you get an answer. It's not the answer, but it's an answer. And it looks pretty discreet. It ends. It doesn't have an arrow that says there's more of these links over here. There's not an alternative, unless you're in some test cohort from OpenAI or somebody. So you get an answer. And we're humans, we're trained. We've been trained for tens of years now at this point, that when someone texts you, you respond. You're now in a conversation. So. So one thing that's changed is instead of giving you raw materials to search for from Google, you are getting an answer, and that answer is presented to you like a chat. And OpenAI stumbled on this with ChatGPT, but it's really important. It's presented to you like a chat. So it feels like not only an answer, but in some cases, the answer. Yeah, so that's two.

Jordan Cooney (5:23)

Let's talk about it.

Jeff Rain (5:24)

Yeah, so there's a couple more, but we can pause and talk about this one or we can kind of.

Jordan Cooney (5:28)

Yeah, let's dive into this answer concept for just a second, because I think the answers is where most of us are spending a lot of our time. And I have to ask you, because I think this is all going to come back when we talk about accessibility and ensuring that these LLMs can, can see what your, your, your website's about. We got to talk about hallucinations. Right? We got to talk about when these answers aren't always exactly as we thought that they would come out. Right. It is a machine at the end of the day, which is where like, I think a lot of the traditional SEOs might go. Hey, the rankings are great because you just allow a human, although they are spelunking, as you said, and I love that the Internet, you still get to choose your adventure, right?

Jeff Rain (6:14)

Yeah.

Jordan Cooney (6:14)

And, and so, but here you necessarily get to choose your adventure because to your point, you're getting an answer. So let's talk briefly about hallucinations and what the implications of that is in an ask and answer world and how that, that fundamentally, I think is maybe changing the way brands are thinking about setting up their content and websites.

Jeff Rain (6:35)

Yeah, look, and I, we could talk about hallucination or we can talk about it. And that's, that's a, that means something to people. It means I'm seeing something that's not real, that's not really there. And so there's a question of is it, is it not really there, is it wrong or is it not what the brand or the business wants to be portraying? These are all different shades of what is essentially the same problem, which is. That's, that's not what I really want to say. Yeah, because, because think about it. The, the answer engines, ChatGPT, Claude, etc are now meta influencers or mega influencers. We tell them things and then we don't decide. We don't control what they then say to their followers, their prompters. We have to rely on them to interpret it. And so sometimes you would call it a hallucination, but maybe it's just, it's just a little bit wrong. It's just a little bit off. What it means though, is as a brand, you have to control what you can control. You have to put forward content on you, on your controlled properties that is as specific as possible, in, as transparent as possible, in as positive a light as you're able to muster so that the, the AI search bots can retrieve that information and present it back. Otherwise you are really relying on the information coming from elsewhere and you're not in control. And so, you know, one person's hallucination might be another person's disinformation.

Jordan Cooney (8:11)

Right.

Jeff Rain (8:12)

So as a brand, you have to try to control the narrative as best you're able.

Jordan Cooney (8:16)

Yeah, I think that's One of the unique differentiators here between search and, you know, the new AI discovery is this realization that there's a lot more control than we actually believe there is as marketers around the narrative. All right. And we dictate that whether it's on our websites, whether it's with our partnerships, whether it's in these other ecosystems, but there's a tremendous amount of control that we do have around the narrative and what we expect these models to respond with on this ask and answer journey that we're on. Right. Does this differ though? Does the expectation of a user differ based on who the user is? Is a buyer have different expectations than someone who's just perusing the Internet in, you know, like what, what, what are your thoughts on, on the, the differentiation that is, is. Is out there for the journeys that different users have?

Jeff Rain (9:20)

Yeah, this is actually really, really good topic. The journey that they're on and how the journey is starting now versus how it used to start. Right. And whether you said it before, it's a little choose your own adventure. If we used to be on a choose your own adventure, you'd scroll Google and you'd find the right spot to go to. Right. Of course, you know, that information was presented to us in a specific way, but we'll leave that aside for now. Now you start your journey in this conversation and the party that you're having the conversation with, whether it's Claude or perplexity, has some memory, has some knowledge of you, and we actually think at everything machines, that this is the beginning of true one to one marketing and that the journeys now, the journeys that will be taking place within the confines of a chat window are in effect the completion of the one to one marketing journey we've been on for 25 years. We've been getting closer, but we've never really gotten there and we're starting to rub up against privacy. But now we get to imbue our personality and our knowledge and the things that we care about into our little chat assistant and we let it go search and bring back just what's right for me. So if I'm a buyer in a corporate environment, you know, and I'm looking for HR software for my specific niche from the company that I work for, in the location I work for, with the type of employees I have, my assistant can know that it's going to know that, you know, from all the things I've done with it before and when it's going to look out and fetch the information about this new HR system. That I want it can do so in a very, very personalized way. That's extremely different than any level of personalization that we could get to on the open web as a casual browser. And again, this can be true in a personal way. If you're looking for a mountain bike, perhaps you and your chat bot have already had conversations about what it means for you to have a bicycle and the kinds of trails you like. So we think this journey is going to ever more personal and that ironically, it took a third party, it takes a third party to really do the personalization. Instead of just buyer and seller, it's now there's this intermediary who can help truly personalize it.

Jordan Cooney (12:00)

That's right. And that, that personalization aspect I think is where this gets really tricky. And I think that's where a lot of SEOs spend most of their time trying to kind of decode what's happening.

Jeff Rain (12:12)

Right.

Jordan Cooney (12:12)

Because the intimacy of a prompt is so much different from the vanilla aspect of a keyword. Right?

Jeff Rain (12:22)

Yeah. Look, I mean as much as you can get out of what used to be one keyword, then we inched up to whatever 1.1 average keywords. I don't know what it is now.

Jordan Cooney (12:32)

Yeah, the slow growth of keyword.

Jeff Rain (12:34)

But we were measuring it in tenths of a keyword. Right. I don't know about you, but I've never really typed in a tenth of a keyword. Which means the vast majority of people were typing one word. You know, brand searches. I used to work at Coca Cola. Like, come on, it was usually Coke.

Jordan Cooney (12:51)

Yep, exactly.

Jeff Rain (12:52)

But at the end of the day, we're moving now from that being the prompt, this one word and Google having some knowledge of who we are to sometimes a paragraph to paragraphs to if then statements being given to these bots and them having memory and knowledge of us. And because of that, we just are able to dive right in to the buyer's journey at a level that which we, we dreamed of, we've dreamed of for 20 years. And the best practitioners of account based marketing are approaching it. But this is a much wider, much broader move and we think it changes the way brands and businesses need to think about what they're publishing and how they're publishing it and how deep they're publishing so that the answers can be as custom as the user really wants it to be.

Jordan Cooney (13:50)

Absolutely. One of the things that this kind of brings us to is the reality that in order for us to truly understand where things are going in this ask and answer world, we, we likely need to change the way we measure what's happening. And there's a lot of news around this. Just even yesterday, Bing just announced, Microsoft announced that they're going to have an AI performance section. We're recording here just on, you know, the day after this, this announcement, Obviously this, this will be, you know, shared publicly and I'm sure people are going to start digging in and researching and unpacking all of the data that's behind that. But the concept of measurement is under extreme change in this moment and most folks are rushing to this concept of AI visibility. Yeah, but candidly, as we talk about how everything machines steps into this equation, it's very different. You guys are looking at the space very different. You're looking at this from an infrastructure and accessibility perspective versus a visibility perspective. Can you tell us more about that and why you're looking at that direction?

Jeff Rain (15:06)

Yeah, it's a great question. And we think that there's really a few parts of this stack and this is going to evolve. Heck, this is so fast moving, it'll evolve before the end of this podcast. Right. Yesterday, Microsoft makes an announcement right now that something has happened. But we think there's several layers that you need in your answer. Engine optimization, AEO or otherwise called GEO Toolbox, you need to monitor and you need to be constantly and consistently monitoring. There are many ways to do that. We work with a company called Gumshoe. We work with another one called Otterly. But there's profound and there's Scrunch and there's dozens really in different methodologies because you need to know how you exist right now in, in the eye and the mind of these LLMs are you showing up and you need to do this regularly because this is a bit of a quantum product, quantum problem instead of a, you know, a physical problem. There's probabilities involved here. The probability of you showing up even under the same prompt are not always the same. So you need to keep measuring, but that's not enough. You need to understand how your content is being published to the world. How does it look to an LLM search bot? And that's where we come in. We decided to focus here to have as much impact as we could have as broadly as possible in the industry. And so no matter how you want to measure whichever tool you want to measure your visibility for whatever fits your business, you will always need a better way to communicate your content to the bots. And that's where we've, we've built the everything cash. This is a product that is meant to really be a thin knowledge layer in between what you would normally and what you've been telling humans for these many years, publishing a website and what you need to now tell these bots. And then we think, and just to complete the stack, at the very least, there is a third layer, which is content enrichment, which is publishing more content, richer, deeper content, more specific content than perhaps you would ever imagine doing on a website aimed at humans who would never get into the kind of specific content that you will happily publish for a bottle which has infinite appetite for your content. So these three layers are distinct. Some people, some providers are going to combine them, some will pull aspects of each. We've decided to focus our capability in our expertise on this publishing aspect of it, building the cash for our brand partners.

Jordan Cooney (17:57)

One of the things that I find remarkable about this, and it's, and it's one of the complicated changes that's taking place in the entire LLM space and that is that like historically Google and search engines, they basically created a standardization around the process for this, right? You'll build a site map. You should have follow some of these general rules, right. And you'll have the, the ability to allow a search engine to access your content. Right. And SEOs and marketers were forever trying to measure this, this benchmark of like this progression of oh, did we get crawled? Did those pages get indexed? Are we getting any rankings? Like, you kind of followed this sequence of work. But what I find really remarkable about the way you're thinking about this transition is that it's not just about the sequence of work. Did this model get access to what we have? But you're also thinking about the organization of it. So going back to this whole measurement concept, you think about like what is, what are the assets that actually matter within this ecosystem of a website and how do I point this machine to the right places on the site? Will this become standardized? Will this become like organized to a, to a, to a fault maybe? I'm curious to get your, your future thinking on that because I fundamentally think that right now we're just scratching the surface on this. But like I also want to stress test like how far does it go?

Jeff Rain (19:32)

Yeah, it's a great question. And I go back to first principles. When we don't know something, we can't know the future. We can try, we could predict, you know, you can bet on anything, Akashi, but we don't.

Jordan Cooney (19:45)

Fortunately these days you can.

Jeff Rain (19:46)

Yeah, I know. And it's really frightening to me to be perfectly blunt. Yeah, but we go to first principles, why did SEO get organized in the way it did? I believe one of the reasons is Google won. Google won. Google won 90x percent of all search volume. And so their rules, the way to win there was the definition of winning. Now it didn't start like that. I used altavista, I used Excite back in the day. If there's any young people listening to this, go check out Dogpile. These were. There was a competition but Google won. And they did it the good old fashioned way. They built a better mousetrap. I don't think we at everything machines don't think this is going to be a one horse race. We think this is going to be a multi party party. We think there are lots of people who are coming to this party. By definition it's a little bit too big to sit out and we'll see how it happens. Right. You know, has Apple decided they're going to sit it out and just take the money from Google and run like they did with search on the browser? Maybe, maybe not. They have an entire app store that they're going to have to figure out how to protect. So how are those things going to interact? Meta. They're not going to sit this one out. They're building glasses and they want you to ask meta with a swipe on your temple. Right. Amazon. Amazon's not going to sit this out. Now what's going to happen to anthropic and perplexity and OpenAI? Well, when the bubble pops, the dominoes start falling. We'll find out what happens with the pieces. But we think this is at least a three party system. Yeah, this is at least going to look like a European Parliament. Much more so than it will look like an American Congress.

Jordan Cooney (21:35)

Right.

Jeff Rain (21:35)

Or I won't even, I won't get political.

Jordan Cooney (21:39)

What does that look like, Jeff?

Jeff Rain (21:42)

Like a single party system. But it's not. It's. We don't think it's going to be those because it's too important. It's because these front doors, these new LLM searches seek to be everything machines, hence the name of the company. They want to go all the way down to the purchase. They want to control you through that or at least guide you through that. And you can't afford to lose that. Battle it for the front door. If you're Meta, if you're Amazon, if you're Google, you just can't afford to lose that. So that's an important first point. SEO evolved in one way because Google won. If we don't think one party wins. If we think this is three, let's call it three, 33%, you know, shared companies, who's to say what the standard becomes? Does doing one set of rules optimize you on one system? Does doing another set of rules optimize you on the others? Right now we're seeing a little bit of that. We're seeing, I think it's fair to say Google attempting to use its current leadership role, or it starts with ma, ends with opoly in search to try to set the rules for how answer engines will be optimized. Oh no, no, do what you're doing. Keep doing SEO. It's working perfectly for you and it's working for Gemini. That's their self interest. So we'll see. It's a wonderful question. I mean, I'd be interested to hear your thoughts. You've been doing this longer, frankly than I have.

Jordan Cooney (23:29)

Well, I think there's a couple components that you brought up there, notably the one that I think we're going to go into in our next topic here is the fact that there's a expectation of a conversion event. And so if the goal in this new ecosystem is that there's an event that takes place and maybe conversion was an overly specific aspect of that. But if, if you want a user to do something and you need them to engage and do something, then the byproduct by which you engage them with this, this ask an answer ecosystem is solving for a very different approach than the spelunking that once was just search results.

Jeff Rain (24:15)

Right?

Jordan Cooney (24:16)

And so, so I think, I think you brought us right to the place that many marketers, many SEOs are thinking, which is, hey, the ask and answer world. An LLM response gets us to a conclusion. If I'm having a conversation with an LLM model, I am by, by nature refining my way to a specific outcome. And that specific outcome is satisfying me as a person. It is giving me what I want. It has found the running shoes that I need. It is help me solve my health care problem. It is defining my specific personal finance issue. It is doing the thing that I need done. And that is a very clear outcome based experience. And in that we're measuring a very different aspect. It's probably not rankings, it's definitely not visibility and it's really, it's really some form of user satisfaction. But what I found really interesting from the everything machines perspective is that you all are looking at the layers that are happening before the user engages that there is so much happening behind this machine to get us to that satisfied answer that the LLM is not only processing our content, but it is looking at third party sources like Reddit and other places. It needs to categorize and organize our data and information about our brand. And it's doing so much energy and work to get to the satisfaction of that specific user. And you're probably one of the first company that's really trying to unpack that. So tell me more about this cash. Tell me why the cash is so valuable and how it gets to that conclusion event of a satisfied user.

Jeff Rain (26:00)

Yeah. So if we think about previous statement was about one to one marketing, we are getting ever closer to one to one marketing. And a cash is a publishing canvas for getting towards one to one marketing. If we are a sophisticated business these days, we might be looking at our 10 Personas or our five Personas and that's how we're, we're doing personalization. But when you realize that those five or 10 Personas are asking these five to 10 key topics or the asking about these questions, they tend to do. If we, we look at that grid of those ten topics and those ten Personas and that grid of a hundred representing some, you know, large percentage of prompts and conversations that we show up in, we then realized we could expand that from, from 10 to 100, from 100 to 1,000. We can subdivide each of those boxes on the checkerboard and each one of them has its own nuance. So each combination of a given Persona and a given topic has its own nuance. And it's that language, it's that nuance in that language that is going to allow a brand to provide better raw materials to the AI search bot to improve their likelihood of showing up. If I know that a given demographic in a given area with a given use case is talking about their problem with certain language, I as a brand can and should attempt to mirror that, talking about my opportunity, my products, my services, and how it satisfies for that use case in that very specific way. Previously, we would never imagine doing this. We would be thrilled to present this information to an influencer or to an agency and have them concoct some creative around that and present it once in the ephemeral stream of content. But we would never think of publishing it in an evergreen way. That's the idea of a fac. Frequently asked questions. We are really moving away from that idea of a FAC to specifically prompted content. Yes, you know, spac, I guess, but that would, that means something else. Yes, but specifically ask questions and that's a bad, you know, that's a bad one. So sack, you know, sacks of content, but that's what it becomes. If you can constantly and iteratively put your brand and products into the perfect context for prompts that are on their way, you are giving the raw material to the LLM to do its job better. And it's going to reward you for that by coming back and visiting more. And if it rewards you and is coming to you as the authority for this information, it's putting less weight on your competitor site. It's putting less weight perhaps on those blogs. Maybe it's putting less weight even on going to Reddit because it's getting high authority, highly detailed information right from the source.

Jordan Cooney (29:31)

That's right.

Pre Visible Sponsor (29:33)

Time for a one minute break to hear from our sponsor, Pre Visible. So you're looking for SEO help and you got a couple of options. You could start replying to spam from agencies that claim they can get you to rank number one on Google. You could pay an hourly rate for a consultant who will inevitably nickel and dime you with hourly charges. Or you could work with a cookie cutter agency to quickly launch a strategyless project with low success rate. None of those sound very good now do they? Well, that's where Pre Visible's integrated consulting model comes in. Pre Visible draws From a collective 40 years of SEO and digital marketing experience to unlock your organic growth opportunities. They build custom solutions that combine strategy, technical expertise, content and reporting to effectively operationalize SEO for your business. Pre Visible's four stage approach ensures that your SEO programs thrive by starting off with a strategy first approach. Then they support you in your efforts to create quality content, help you identify technical issues, and most importantly, to work with your cross functional teams to integrate your SEO strategies to make sure that your SEO budget actually drives results, not just your agency's bottom line. So join brands like Yelp, eBay, Canva, Atlassian Square, all who rely on the SEO consultants at Pre Visible. For more information, go to Previsible IO that's Pre Visible. P R E V I S I B L E.

Jordan Cooney (31:01)

I mean, what I one of the things that I found fascinating looking at the data you know, of LLM traffic or even notably, you know, like we were just discussing earlier, Bing's AI performance dashboard that they released yesterday, regardless of either of those two data sources, which arguably are likely the most factual sets of data we can look at about what's happening inside of an LLM, because none of these companies have exposed a reporting engine to us, right? They haven't given us a Google Analytics. They haven't given us a Google search console. I mean, Bing is now arguably giving us a taste of this. And the thing that fascinates me the most is that when you look at the pages for large established brands that receive the most traffic or now in some cases are most cited, like in the Bing AI performance report, the pages that often show up are the ones that I would least expect. They are support pages that explain a very specific use case for a user. They're learning center sections that help a user walk through an entire process. They are like some of the most functional pages that I'm sure received visits in the past. But by no stretch of the imagination that I anticipate these being the top 10 on a report.

Jeff Rain (32:27)

That's crazy. Yeah. When you said you weren't expecting it and then you said those are exactly the pages that I expect to see. Yeah, those are precisely the pages I expect to see. And those are precisely the pages that need to have a Cambrian explosion of creation that we imagine. Machines will create a lot of it. There'll be human in the loop. There'll be different flavors for different. But even further beyond that, publishing chat logs with anonymized chat logs with actual customers to really get into the heart of it and publishing motivations for the production of that product. Why did you produce this product in the first place? What niche were you hoping to satisfy in the market? Tell the LLMs everything. Again, plugging the name. This is why we call the company Everything Machine. These bots seek to become everything machines for their prompters. And you need to feed it everything. There is going to be some declining marginal return for that next marginal page or that next marginal answer we think is even more important. Forget pages. What we're trying to do is produce questions and answers in conversation that are going to help the LLMs learn. Yes, and it's exactly the support page. It's exactly the very, very specific case study you published with that customer in that particular niche that is going to get cited. And that's what we want. And how do we do that infinitely and not overwhelm human visitors? Our answer is that you don't. You create a parallel structure. The Internet is splitting in two and for a good long time, lots and lots of humans are going to come visit web pages, but more and more traffic is going to be consumed by search bots, by these AI driven search bots. And that and the cache is a solution for this challenge.

Jordan Cooney (34:37)

So let's dive into that. Let's Talk more about the cache. Let's talk more about. I really want you to give our listeners just a brief overview of how this everything cache works, why you've built this first version or, or aspect. And then, and then, and then I'm going to ask you a few follow up questions about like the whole, the whole discussion here that we've been having, which is why LLMs are accessing it and all that. But just give us the lay of the land of what this thing is.

Jeff Rain (35:06)

Yeah, look, it is a mirror, is a true and faithful mirror at the beginning of your existing content. So you have a website, you have pages, you need to, we would argue you need to replicate that and mirror it so that the bots can easily access it because they're not going to fight through your JavaScript, because they can't. So all of these beautiful, dynamic things that you designed on your website for humans, a puppy with big eyes looking at you, beckoning you to buy insurance, it doesn't matter to Claude, it doesn't matter. The lovely colors and the logo and how it melds down love, you know, beautifully into the navigation window, it doesn't matter. What matters is clean, fast markdown wrapped in JSON ld that is lightning fast. That is pre rendered, just get it in, get it out. And so it's just completely different experience. And so, you know, the world right now is struggling. It's like, do we, do we compromise? Do we build a camel if you will, and optimize for everything? But everyone who's ever tried to satisfy multiple stakeholders knows that you fail in some regard. You can't have it fast, good and cheap. You can optimize for people and for Google Search Index to send more people to it, or you can optimize for these AI search bots and these prompting environments. So we say split in two and we're SEO neutral. So we start, we scrape your site, we produce this cache and we do it. And it's designed to be SEO neutral. That is a very, very important first step. It does not compete with the original content. There's no index protection. All pages include no index, no follow meta tags for Google, Bing and everybody else. Canonical links. Every page that we create points to the original source URL as canonical. There's clear attribution and there's cache identification. Pages are clearly marked as cached archived content. So that's the first important part is what it's not, it's not competition for your search optimized pages and your human pages. So we're not going to get sighted. We don't get sighted. We are there to communicate, to wave our hands in the air, almost wave them like we just don't care and bring those search bots to us. So then what do we do? We are there to reduce noise, we're there to enhance the semantic structure, put it in structured data and then as I said before, to be SEO neutral, um, I mean it starts and a lot of it is done by just stripping out JavaScript, CSS and tracking, tracking scripts. You know that first and foremost does, does a lot of the work, get rid of the ads. The, the LLM doesn't care that there's an ad on your page.

Jordan Cooney (38:19)

Right, right.

Jeff Rain (38:21)

Getting the navigation and the boilerplate content separated, getting forms and the interactive elements to be documented, but not preserved. And then semantic enhancement. These are the biggest aspects of it really. Using HTML5 semantic structure, having really clear heading hierarchy is validated and correct. Then there's a checklist. But these are some of the key things. Having the structured data semantically enhanced with as little noise as possible to serve it up.

Jordan Cooney (38:58)

So one of my first key questions on this is if you think about this cache, this ecosystem that we're serving up for these models, is the intention here to have the model leverage this for training or inference or both? So the training aspect is, hey, this is deeply ingrained into the model and now they kind of use this as a source over and over again to, to, to define or respond and answers. Inference is like, hey, we're trying to figure this out and understand this. Right? Concepts like grounding are part of the inference process that an LLM uses. So I'm trying to understand like where, where does this cache sit? And is it, is it one, the other? Both.

Jeff Rain (39:42)

We believe it's both. And all right for an inference for web search, for executing that subtext sub task of a web search and then over time for training. Look, we can't claim that right now, you know, we, we don't have nearly the volume and there we haven't existed through the training cycles. Sure. It is reasonable to believe that if we are able to put language in front of LLM bots in their training processes, that perhaps there are concepts that can be trained. And I've had these conversations with brands before who I say they're experts in their field. You're a hospitality brand, you're an expert in hoteling. Why aren't you teaching the LLMs what it means to be a family friendly hotel in Paris for a family of three from Atlanta? So it's Interesting. Is there training value? We believe there could be. We don't know yet at the moment. Inference and really doing these web searches is where we're going to see the most benefit and we're seeing benefit that we can measure. So we are live with clients, we are watching their visibility, we are watching the number of times they are cited. Their content is cited because it's citations that are, that are indicating, hey, you're, you know, we care about this content and that how often you're mentioned by name in these responses. And we're, we're seeing good results. How we'll measure how it's impacting training, we haven't given a thought to that yet.

Jordan Cooney (41:31)

Fair enough. Fair enough. I am curious on this concept of organization because when I, when I've seen these caches and we talked about this, you and I, I found it really remarkable that you guys are organizing this content for, or this, this, the structure of this cash, excuse me, for the brands that you're working with. Why is that and how have you been thinking about the future of that organization, of the cash?

Jeff Rain (41:59)

Why is that? I mean, our, our goal is to help that business and its products, its brands, be represented as they would like, as they hope to have them represented inside these responses. That's the goal. You know, if we, if we worked for an LLM, our goal might be different. If one of them wants to call us and reorient our work, our phones will ring. But our goal, and we believe what is extremely valuable for marketers is to ensure that they can communicate to this new audience. So that's our intent, is to organize it by brand, by the economic actors who are interested first and foremost in it showing up correctly as they believe it should. Is that different from the truth or from reality? We'll leave that to the model owners and the search prompt owners themselves.

Jordan Cooney (43:11)

Fair enough. Fair enough. You brought up a concept earlier and I want to dive into this last main topic that I think is super relevant to our audience and that's this, this notion of one to one marketing, that we're now in this era of one to one marketing and that what we build and the experiences we have on our websites are genuinely there to serve that one human being behind them. You know, this, this goes in many ways in contradiction to much of the foundational practices that many search marketers and performance marketers have been taught. Right. We've been taught to go after search volume. We've been taught to go after massive trends. We've been taught to seek out audiences in Our content or Personas in our content. Personas, audiences, massive search volume, none of those is one to one marketing. So tell us why the importance of one to one marketing is the future of how to best serve these models.

Jeff Rain (44:22)

Gosh, I might argue with the premise. I think it's rooted in Persona. It's just, it's a huge unlock. And so our frame of reference just has to change. And we always, you know, and this is different marketers with different schools of thought and different practices are going to differ on the tactics. But you know, we are, we all are, we all are trying to get as close to one to one as possible. We just know the limitations of the system that we're operating in, right? And so we internalize those limitations and we say, well, we can't run a million campaigns for this candy bar for this spring, but we can run a hundred. That's doable. We can manage that. And then Meta changes something. Google changed something. I was like, oh, maybe we can try 500 campaigns or maybe we can do dynamic content optimization. Okay, let's try that. All of a sudden we expand our view and what we thought was a constraint is no longer a constraint. And I think that's what we're seeing here. The thing that was previously a constraint. In order to drive volume, I can come up with certain Personas and I can deliver content against those Personas, but if I try to parse it any further, I won't be able to deliver the volume at the top. But that constraint is now different because the context for those individual queries prompts is now baked in. It's baked into the prompt. It's always there or it's increasingly frequently there. Now there's still an argument on the other side of what is the marginal rate of value. But we'll figure it out. But it's a lot. It's going to be a lot. And it's also just different structures. But you know, I've been in all sorts of marketing roles and we almost start thinking about this as how, how would you build infinite landing pages if there's that perfect, perfect landing page for every user out there to land on? How, how would you go about building that? And that that's sort of how we think about everything. Cache. It's the tool to build these infinite or nearly infinite landing pages, which again in some get you back to the audience. Now I will say we're perfectly rational humans. We know not everyone is moving to AI prompted search immediately tomorrow, right? But it's happening. It's taking on more volume and it has to be looked at very, very carefully in the overall mix. It's not the same as running a television campaign. It's not the same as running a display campaign. It's not the same as doing your search campaign. It's its own channel and it's going to have slightly different practices.

Jordan Cooney (47:30)

I love that. I think it's remarkable to have that perspective. I think a lot of people are still looking at this channel as a byproduct of traditional search and I think there are some similarities. But to your point there, there are some very unique differences here that we all have to, to grapple with to, to ensure that our campaigns are working effectively.

Jeff Rain (47:55)

Yeah. Here, look, Jordan, I'm not a search guy. I haven't been doing SEO and SEM for the past umpteen years. I've done all sorts of things and I've interacted very closely with the search world. But I have a little bit of a beginner's mindset here and my colleagues are a little bit more steeped in this industry than I am. But my beginner's mindset goes back to first principles and that's how we're trying to orient everything, machines and the everything cash is. If we go back to first principles and we think this is going to be a multi party system, if we think personalization is sort of baked into the idea of having an assistant, if we think that the learning mechanism is content and asking questions and giving answers, then how do we reorient our publishing mechanism from here? If we know they can't read js, if we know they're hungry for more content. Okay, well what do we build? So it was a little bit of a beginner's mindset because frankly I'm a little bit of a beginner and it's been really fun, it's been really exciting. These last to call it six months as we've been on this journey and you know this is going to be another decade that, that I think we're in. So there's no, no rest for the wicked, they say.

Jordan Cooney (49:20)

That's right. That's right. And I, and I love the beginner's mindset. I think, I think many of us that have been in this space for a long time need to reevaluate and that, that reevaluation can, can pay off in, in a big way in terms of how we, we implement, execute and, and leverage the technology we have around, around this, this no longer monopolistic environment that we've been in for the better part of 20 years. Right. It's a multi pronged approach. Now when you look at these LLM models.

Jeff Rain (49:49)

Yeah. And when we think just this is. And some of this gets a little philosophical, but when you think about the con, the context of receiving this answer, it comes to you. As a chat. We have been trained. I used to work at Skype many, many moons ago. But for years now, we have gotten used to what the text messaging format looks like and how we behave. In a text messaging format, when you prompt, you're expecting an answer. You receive that answer, you consume it, not looking for anything else. You're no longer your mind. And I'm sure someone smarter than me knows this. Your mind is in a reception mode. I'm reading what I was, what was just written to me and I, I just take it as it is. And so there might be more, there might be a follow up or clarification, et cetera, but I'm in this linear mode that is fundamentally different mind space than querying Google with a keyword and then you being the active agent, yes, the entity that is now spelunking. Okay, Google has returned me a list of stuff. Let me go figure it out. Versus oh, this chat creature just delivered me an answer. Let me accept it, integrate it, and then figure out what to do next. I don't know if this is, you know, we'll have to talk to some professionals here, but it's a fundamentally different way of finding new information. And I think we're just businesses, especially B2B right now, but eventually and ever more increasingly B2C, we just, we're gonna have to do things differently.

Jordan Cooney (51:40)

No question. So this is a great moment for us to move into our lightning round. Oh, I'm gonna ask you five questions from our discussion today and you give me like a 30 second or so response. Want to just get kind of that real clean, direct and I mean, Jordan,

Jeff Rain (52:00)

you've been on this long enough with me to know I don't give 30 second response.

Jordan Cooney (52:03)

I know.

Jeff Rain (52:04)

I will try. I will try.

Jordan Cooney (52:06)

Love it.

Jeff Rain (52:07)

Love it.

Jordan Cooney (52:08)

Okay, let's dive in. Jeff, what's one thing B2B marketers still measure that matters less every month.

Jeff Rain (52:21)

Oh, they're search traffic referrals from google.com 100.

Jordan Cooney (52:27)

Yep. Yeah. Where should they be going then? Where? I mean, it's such a, it's that, that has been such a well of, of, of knowledge for, for marketers for a long time. Where do they go now?

Jeff Rain (52:40)

We wait until someone builds the tools to actually figure out how many people started their journey somewhere else? This is the hardest thing. There's not a clean replacement. I mean, look, Bing, Bing just threw something out into the world yesterday. We just talked about it. More tools are coming. We are looking very carefully at how we measure the traffic from bots and give that back to our brands, our clients, so that they understand how frequently they're being interacted with from another, from this other entity. If that's where the volume is going, if it's leaving google.com and it's moving to these four places, how do we get that representation? Because people aren't just clicking through on citations and mentions. It's just not happening.

Jordan Cooney (53:35)

Yeah, not happening.

Jeff Rain (53:35)

It's like one to a thousand. But we don't, we don't have an answer yet. But it's a really, it's a very good topic.

Jordan Cooney (53:47)

Love it, Love it. Hope for a whole nother episode. All right, next question. What's the biggest mistake brands make when they treat LLM optimization like SEO?

Jeff Rain (53:57)

Well, I, I think the biggest mistake is they assume the bots all behave the same way and that they assume they behave like Google Pop and they don't. Yep. Yeah.

Jordan Cooney (54:10)

100. I, I think that that is a, a very clear shift that the market needs to make. So love that.

Jeff Rain (54:18)

And they have economic incentives not to work in the same way. Of course it's, it's not like it's not a coincidence. There's, there's reason and there's market rationale for them to work differently. Now that might converge, but it might not like it even further apart.

Jordan Cooney (54:35)

All right, if you could change one thing on every Enterprise website tomorrow, what would it be?

Jeff Rain (54:43)

You mean besides the Pay Jeff button? Look, I mean, we think every Enterprise website should, should have a cache. And now that's maybe a little bit of a self serving answer, but we think it's true. They are not, they are currently not communicating as well as they can to an extremely important other constituency, that is the search bots. And if anyone's a nerd, you know, out there in your, your listener audience who's watched Star Trek before, you're, you're communicating to all the humans. So Captain Kirk's here in your message, but you're not communicating to the Vulcans. So Spock's getting half of it because he's half and half. But, but the Vulcans aren't getting any. They're not getting the message clearly. Why aren't you talking to the Vulcans? Yep.

Jordan Cooney (55:31)

Love it. Love it. All right, what's the Clearest sign a company's content is invisible to an LLM bot.

Jeff Rain (55:40)

Oh, readability. We actually have a readability score that we run. We have an audit which cause we're really original, we call the everything audit that will go through the site and will break it down in a deterministic way. Looking at your universal signals, raw HTML signals, discovery and rendered signals, this is becoming more of a science. And so the higher your readability score, the more visible it is to the bots themselves. And that's the start. And it's growing. We're learning every day. We're on our third version of it and we're are getting better every day.

Jordan Cooney (56:21)

I love it. You guys are always on brand with everything, so maybe too on brand. I love it. All right, deal with that another time. Right? Right. Last lightning round question here. In two years, what will B2B teams wish they would have never ignored in the current LLM shift?

Jeff Rain (56:43)

Since I don't think they will ignore readability and building caches because I think they will good. They will see those things as required and necessary. What I think they will wish they hadn't ignored is the increasing truthfulness and transparency that will be rewarded by lems on a forever now forward that it is now ever more difficult and will not be rewarded. Attempting to persuade, but rather to present the truthful and right information. And you know, persuasion is one thing and that's a neutral term, but you can't trick anybody. We're moving away from being able to pull a fast one on the buyer because they will have an extremely rational agent as their partner in crime in figuring out what's right for them. And if there is, there's lots of reasons why there might be kind of non rational or irrational use cases. And I just like that brand better and that color looks better on me and I just love Nike and that's just what I love. But it won't be because of fooling anyone. And I don't know that that's particularly obvious right now. We try to to repeat it and to make it clear because we think that's the future. The future is in this perfect one to one match which is this agent, this intelligent agent, finding the right things for my, my human, my person. And it's not going to be fooled.

Jordan Cooney (58:33)

Love it. And that's a great place for us to wrap up this episode of the Voices search podcast. Thanks to Jeff Rain, co founder at Everything Machines for joining us. If you'd like to contact Jeff, you can find a link to his LinkedIn profile in our show notes or on voicesofsearch.com or you can visit his company website, everythingmachines.com if you haven't subscribed yet and would like a daily stream of SEO and content marketing knowledge in your podcast feed, hit the subscribe button in your podcast app or on YouTube and we'll be in your feed every week. Okay, that's all for today, but until next time, remember the answers are always in the data.