Garrett Sussman: Inside SEO Week, I’ve been talking to this guy for a minute but he is like behind the scenes. I got Scott Stouffer today, he is the Co-Founder and CTO of Market Brew and they’re doing some really cool stuff when it comes to prepping for AI search, the direction that we’re going in. He’s the the Founder of Market Brew as I mentioned, so that – it’s like an AI that predicts Google and ChatGPT rankings before you actually publish. Graduated from Carnegie Mellon, Commodore 64 kid, still coding, kind of wants to just in the weeds. Scott, thanks for joining me today. How are you doing, man? 

Scott Stouffer: My pleasure. Yeah. Nice to be with you. 

Garrett: So this is your first week. Speaking of SEO Week, I am so hyped for what you’re going to share with all of us. But let’s dive in because so much has happened over the last 12 months, two years, five years with AI and search. What’s top of mind for you in AI search right now? 

Scott: Yeah, so, you know, obviously, I’m a search engineer who sort of spent the last couple decades building models of how search engines actually work. And so at SEO Week, I’ll be talking about something that’s becoming increasingly important in the AI search era, which is your brand’s mathematical footprint. So AI systems don’t retrieve pages the way that, you know, traditional search did, as we’re all starting to realize. And they primarily retrieve embeddings. That’s sort of the metric of units that they work on. Mathematical representations of content. And so we’ll dig into this. We’ll see how, you know, so the real question brands need to ask really is, is AI seeing the brand that you think you’ve built? Right. So that’s the session that I’ll be talking about. It’ll be about measuring that and aligning your content. So, you know, AI retrieval systems consistently choose you and your brand. 

What’s top of mind for me in search is it’s quietly shifting from ranking pages to retrieving chunks of meaning. So traditional SEO focused on, you know, ranking signals like links and keywords. But AI systems work differently. They convert content into vectors, retrieve information based on similarity and embedding spaces. And that means that a brand’s visibility now depends on whether the content sits in the right region of that mathematical space. It’s a really interesting concept, especially for a search engineer and somebody with a mathematical background. So if your embeddings aren’t aligned with what AI is retrieving, you don’t just rank lower. You may not even be eligible to show up. So it’s a fascinating dive into sort of the search engineering mathematical world of what you may not always think of SEO as. And I’m very excited to share it with your audience. 

Garrett: Yeah, I’m excited because it’s like engineering mindset, science-based. You’re not talking about like fluff and vibes and like, you know, our whole like LinkedIn world of this feels this way, that feels that way. No, like you’re going into, like you’re reading the patents, you’re kind of trying to reverse engineer all this stuff that we’re seeing online as it happens. Knowing what you know about what’s gone on over the past two years and your focus on vector embedding, semantic search, cosine similarity. What’s your perspective on the direction that SEO is heading going forward? What do you think the next six months to 12 months will even look like? 

Scott: I think we’re moving from optimization to alignment. I think that’s the big tagline. For SEO, for years, it’s been about improving signals page by page, but AI retrieval systems evaluate content much more holistically. Entire websites form these geometric structures and embedding spaces. And what really matters is what your brand center of gravity is, right? Like I call this the centroid. There’s a lot of different ways that you can represent this. So the engineer’s perspective is we call it a centroid. But it’s where your brand centroid sits relative to what the queries that people are asking, right? And so the next phase of SEO, I really believe is going to be much more measurable and mathematical, which is exciting because in the past, SEO, there’s this gray area where people just want to see hard numbers and figures. And I think we are moving more towards that in general in the industry. So instead of guessing why visibility changes, we’ll actually be able to measure how closely a brand’s content aligns with the topics that it wants to actually own. And so each topic has sort of a mathematical signature to it. So it’s, yeah, I think we’re moving from a optimization world exercise to an alignment exercise. 

Garrett: It’s so fascinating that you say that because I feel like it’s a counter argument to the direction that it feels like it’s going a lot of times because everything feels so probabilistic, so messy, less attribution. And so you’re saying, no, we’re going to get almost a direction of maybe not necessarily precision, but more scientific mathematics, what are the like problems or the questions that you’re most focused on to point us in that direction? 

Scott: Yeah. I mean, it can, it can seem messy right now. And that’s primarily a derivative of the LLM industry just starting to grow up. You know, we see all the, like some of these old tactics and SEO working now on LLMs. And these are just short-term hiccups with the LLM sort of going through the adolescent stage and understanding like, oh yeah, there’s a whole underlying SEO ranking system that we need to get right so that our LLMs can quote and cite and, you know, retrieve the context the correct way. 

You know, so the big focus that I’m looking at right now and in my team of search engineers at Market Brew is that why do some brands consistently get retrieved by AI systems while others don’t, right? That’s the primary goal of most CMOs and CEOs. They want to understand why they’re not showing up in LLMs. And so what we found is many websites, what they do is they accidentally build content that’s mathematically indistinguishable from competitors. And it’s really hard to see this today because it is a mathematical footprint. You could have a 2,000 word article next to another 2,000 word article and show it to a human and it looks completely different. But show it to a search engineer and his algorithms and it looks identical. And so what we’re really going to be focusing on is letting SEOs into the world of search engineering and all these mathematical algorithms so that they can see this is different than this. And then, you know, we go back to the centroid or what is essentially an embeddings cluster of content in the same region of that space that they’re trying to target. 

So retrieval systems, when they look at these pieces of content, they see millions of pieces of content. They all look different to a human, but they’re really, they kind of just are interchangeable because they’re all mathematically very similar. So the challenge becomes, how do you create a distinct digital footprint that AI systems can clearly associate with your brand? So that’s the thing that we’re going to be focusing on in 2026, 2027. 

Garrett: Oh man, I have so many questions and we have limited time, but because it makes me think about like, it makes sense that we’d see everything being similar mathematically because Google for so long has incentivized copycat content. Like it feels like there’s an incentivization, from a ranking perspective, if you want to be in the top 10 and people would build their sites like that. But you’re kind of saying that you need to be able to distinguish without going too far into it. What does that look like? How do you identify the brands that have distinguished themselves? I know it’s kind of like a sneak peek of some of the stuff you’re going to share.

Scott: Well, yeah. So it’s, it’s, it’s a little bit of both, right? So there is a mathematical structure that Google prefers and it changes per SERP, right? So depending on the searcher’s intent and the environment that that searcher is in, whether they’re on organic search or LLM search, all of these play into a specific structure or footprint that these search engines prefer. And this is still carrying on through the transition over to machine learning way back 10 years ago in search where they sort of – you start at the end, right? You have the quality rater guidelines. You give it to a human. You have them label spam, not spam, hot dog, not hot dog, for those of you who’ve watched the show. And so that is then taken into an unsupervised machine learning process. We figure out sort of what are the weightings to get to the end, right, to get those results that the humans have labeled. And that is still carrying over, right? So you still have this like signature, mathematical signature that is the good signature to have. 

The problem is, is that everybody sort of uses all the same LLMs, uses all the same embeddings, and then they put that on top of that and it all looks the same. And so what Google and what search engineers are looking for is we want to have as much variety of results coming back to our users. And to do that, what we effectively do is we’re just going to group all the mathematically similar content together in a group. And then we’ll use the underlying SEO factors to sort of pick from the top and say, you know, this is the representative sample from this group. So the goal, you as a brand, what your goal is, is to look unique mathematically. So you want to use the right mathematical structures that Google is preferring. And, you know, there’s ways to do this as search engineers. And, of course, we invented Market Brew to sort of deconstruct what that is. But there’s different ways to see what works and what doesn’t with many different tools. But then it’s applying, it’s figuring out what your digital footprint is for your brand and making it unique. And then making sure that everything that you write about has got that digital footprint alongside it. So it’s sort of part SEO creative content writing and part mathematical search engineering that’s sort of combining together to be like the modern version of SEO. So or whatever anybody calls it today. 

Garrett: It’s so interesting to me, too, because also we hear like a lot of these myths around like the idea of information game. And that’s like the information game’s an element of that, you know, kind of separate from the individual mathematical signature of a brand. You still need like that’s what we’re talking about here, right? Is like that mathematical addition. That’s not…

Scott: Yep. Yep. 

Garrett: Okay. Okay. So speak more to your like you’re an engineer. What is your core wheelhouse? Like how does it show up in your thinking when you’re building these tools and providing these solutions? 

Scott: Well, my background is in search engine modeling, right? So actually building systems that simulate how search engines retrieve and rank information. And so, you know, that engineering perspective changes how you look at SEO. Instead of focusing only on tactics, you start thinking about like the actual algorithms underneath. And so that’s what I’ve been doing for the last, you know, 15 plus years. When you look at search through that lens, you know, you realize that AI retrieval is fundamentally a geometry problem today. Content becomes vectors, sites form clusters, you know, visibility is determined by distance and similarity. And so that’s really the, I think, the search engineer smashing into the SEO world. That’s the magic of, I think, what I bring at least and my team of search engineers are trying to bring to the industry is it’s sort of like, you know, have you ever go to a museum where you have the engine cut out and you have all this, the, the gears turning and there’s a glass window and you can see how it all works, right? You can just, you spin it and you’re like, oh, this connects to that. And you can see inside that engine. Well, that is essentially what we’re trying to do here at Market Brew for search engines, right? And LLMs. We want to put that glass window up so that everybody who is this, this, you know, these amazing SEO minds that know how to sort of take that, that world and convey it to the human world of websites, they can see clearly what the search engine is asking of it. 

Garrett: Do you think that’ll continue to be accessible? Or do you think like the black box, black box nature of LLMs and the growth of search engines will get to a point where it is completely obfuscated? Like even with reverse engineering, we’re going to be limited in what we can figure out. 

Scott: I think we already are at that point. Even if you look at like the commercial products that developed at Market Brew, we still don’t have, you know, we don’t have the database of intentions that Google has. We don’t have the, uh, the dwell times. We don’t have like the stickiness factors, all this stuff. There’s a lot of stuff that is just, it’s only the, that search engine has, has purview to. But, um, I think at this point, I think we’ve, we’ve sort of, um, if we haven’t hit peak obfuscation, we’re, we’re very close to it. Um, and part of the reason why everybody is sort of turning to the friendly team of search engineers, which is basically saying, hey, look, you know, we’ve got like a very, very hard mathematical calculation that’s going on in the background. We don’t quite understand why LLMs work the way they do or why they’re retrieving your brand versus somebody else. And I think you’ll see a lot of the pivot in the SEO world to go from just we’re going to just throw content at something to, here’s here’s where we’re started mathematically, here’s the sort of anchor of what we want, we built this brand’s anchor and then as we write content we’re always anchoring it to. That we’re using AI to assist in that, right, because the AI is is the is our bridge to that. Even search engineers, we deal with like thousands of dimensions of embeddings, right, there’s no way that that any of it makes sense to us looking at it on paper. Or we can’t just, you know, reverse engineer it on a back of an envelope. So a lot of this is AI-assisted. So it’s a sort of an augmented SEO world now where we’re using AI to talk with the AI so that they understand what we’re trying to convey. So, yeah. 

Garrett: It is so interesting. I’m hyped for your talk. You’ve already kind of like danced around us and given us little peeks. But give us what might attendees hear specifically from your presentation that should make sure that they show up on Day 1 to see Scott? 

Scott: Okay, yeah. And I’m very excited to give this speech. I don’t really give many speeches, so this is a great chance for me to sort of talk about what we’ve been cooking up for the last few years here. But at SEO Week, I’ll show a framework for measuring what I call a brand’s mathematical identity. So we look at how content embeddings cluster across the site, calculate the centroid of a dominant cluster, and then measure how closely pages align with that identity. And that measurement, which I call a top cluster similarity, reveals a brand’s content, where a brand’s content is drifting and where it’s aligned. So you’ll be able to mathematically see in a visual format how it’s aligned with what you’re trying to target. And this will be very eye-opening for a lot of people to see how this is actually being viewed by search engineers and how we write these algorithms and how we interpret it. And it’ll turn SEO into something that’s much more predictable because you can actually see how AI systems interpret your site. So don’t think of this as something that’s like, oh gosh, this is just gonna be, like, too complicated. It actually, the goal of what we’re doing here is to try to simplify things and make it easier to understand why AI is choosing your brand over somebody else’s. 

Garrett: It’s so important. And with the feeling of volatility of the AI search industry, it’s like, to have a framework like this where the goal is sustainability, like long-term sustainability, especially for established brands, I think that people need to start putting these processes and these frameworks into place. Just finally, what are you excited about for SEO Week? What makes you most excited about the event coming up? 

Scott: I mean, I’m really excited about SEO Week because it brings people together who are thinking seriously about where search is going next. I mean, you guys are my people. I’ve been surrounded by SEOs my entire career here. It’s a kind of an interesting, weird world that I live in because I’m sort of very isolated. I’m not like a lot of the SEOs. I’m a search engineer, so I have a different brain. I have more of a problem-solving brain that’s more mathematical. But we’re in a moment where AI is changing the mechanics of retrieval, and a lot of the industry is still trying to map that shift. And I feel like this is something that I can help the industry do, which makes me excited about the event. And I just love, you know, events like this is where…it’s where those ideas start to click for people. I love seeing the light bulb moment go on for people who are, you know, they’ve got these creative brains and they have these, you know, they’re just looking for that light bulb moment, that little piece of information that Google is withholding from them for so many years. So I’m looking forward to showing how SEO becomes a lot more understandable once you start thinking about it in mathematical terms. And I’m also just excited to see everybody. I have a lot of friends. I’ve been with a lot of you guys for 10 plus years, and it’s just nice to catch up. So it’s going to be a lot of fun.  

Garrett: I can’t wait, man. It’ll be great to see you in person. So if you guys haven’t already got your tickets, April 27th to April 30th, It is coming up. Come see him, Scott, on Day 1. Find him on LinkedIn if you haven’t already. We’ll have the link in on the page as well. And, yeah, we’ll see you at SEO Week. This is Garrett signing off. Thanks for joining me, Scott. 

Scott: Thanks. See you.