THIS might disrupt the $2.2B SEO industry completely (Generative Engine Optimization)

All right, we're going to talk about the new SEO in town now. It's called Generative Engine Optimization, GEO, or you can call it GEO, whatever you want to call it. This hat tip I want to give to Aleyda Solis, and I saw her post about this, and this is a 19-page document. Neil and I are going to walk through this and we're going to give you our reactions to this. So Neil, this is brand new to Neil right now. He's going to be seeing this in real time. I actually just started reading this 5 minutes ago. So we're basically going through this with you in real time. So those of you that haven't subscribed to us on YouTube, please subscribe on YouTube. And we want to talk about how this is going to change SEO in the future. And I'm going to go ahead and share my screen now, and then I'm going to read through what's on my screen. So those of you that aren't watching, I'm also going to explain what is on the screen as well. So first things first is when, if we scroll to the top over here, so it says the invention of traditional search engines 3 decades ago marked a shift in the way information was accessed and disseminated. Across the globe. While these search engines were powerful and ushered in a host of applications like academic research and e-commerce, they were limited to providing a list of relevant websites to user queries. The recent success of large language models, LLMs, however, has paved the way for better systems like Bing Chat, Google's SGE, and Perplexity.ai that combine the strength of conventional search engines with the flexibility of generative models. We dub these new age systems generative engines because they not only search for information, but also generate multimodal responses by synthesizing multiple sources. So I'm going to stop for a second and I'm going to continue to scroll down. So they're basically saying, hey, we have a hypothesis here that GEO is going to be the next version of SEO basically, right? And they're going to share some of their methodologies and they're going to share some of the results, which I'll get to in a table. We're not going to read through this entire 19-page document. But what I think is interesting where we can start off first is like, when you look at this picture over here on the left side, this is what a search engine looks like when you— a SERP looks like, right? So you might search for things to do in New York and you get number 1, here's the number 1 result, Central Parks in New York, right? Number 2, Statue of Liberty, right? Number 3, New York-style pizza. So you get the typical 10 blue links, right?

Yeah.

Now, they're saying it's going to change a little bit with a generative engine. It's going to be like, hey, here are all the things you can do. Check out this New York pizza. You could do this over here. You could do this over here. So it becomes more of a conversation. And they're basically saying, hey, look, we think this is going this way. So we should optimize our websites for these GEOs. I'm going to pause for a second. Neil, what do you think so far?

Yeah. So this is a new term I've never heard before. But you and I have already both talked about this, and I've even put out articles and tweets about this in which you can optimize for ChatGPT. You can optimize for things like Bard. Now, ChatGPT's index is from September 2021, but imagine if you're using a lot of the signals and they start asking you questions like, what are the top ad agencies? And believe it or not, we've picked up clients from ChatGPT recommending us. Now again, their index is from September 2021, the last time I checked. Over time, it's going to be more and more up to date, just like Bard I think is only what, 2, 3 weeks old or something like that at any given point. And if you can start getting them to recommend you, if you can get them to start talking about your companies and your products within the suggestions when people are searching or asking questions, you'll end up picking up more business. It's a new form of SEO or what these guys are calling GEO that, you know, I think will pick up. And funny enough, Eric shifted Single Grain, right? The branding around the agency to be more about AI. And we've already picked up customers who are requesting us to help them get more recommended by AI.

Yeah. And here's the thing. So they have a couple of generative engine optimization methods, which I'll read off in a second, but When we do talk about the new AI stuff we're doing with clients, their eyes, their ears perk up because it's something new, right? And if you're an agency owner listening to this right now, ultimately what clients want at the end of the day, and Neil and I, we're happy to share all this stuff. We think there's business for everyone. But look, at the end of the day, they want to work with someone they feel confident with and they want to work with someone where they feel like they're going to be proactive about bringing them new ideas. And that's your job here, right? And so that's why we're here to call this stuff out. So This paper, by the way, this 19-pager, there's a lot of mathematical formulas in here. There's some programming formulas in here. It's more of a scientific paper, I would say. But they said here, to improve the impression metrics, content creators need to make changes to their websites. To this end, we present several generative engine agnostic strategies referred to as GEO optimization methods. Mathematically, every GEO method is a function where W is the initial web content and— I'm not going to read the math. So, okay, so here's what it says. We promote and evaluate a series of methods. 1, authoritative, modifies text style of the source content to be more persuasive while making authoritative claims. 2, keyword stuffing, modifies content to include more keywords from the query as would be expected in classical SEO optimization. 3, statistics add-on, modifies content to include quantitative statistics instead of qualitative discussion whenever possible, wherever possible. Site sources, quotation additions, add relevant sources, easy to understand, simplify the languages, fluency optimization. So these are not necessarily a lot of new things. This is just good writing at the end of the day. So—

Dude, check out that graph. Share your screen. You're already sharing your screen, but click on that graph, the link I just sent you in the chat.

What chat? Oh, here. Let me see.

So we already looked at factors that are causing ChatGPT and Bard to recommend sites. Again, keep in mind, if you want to optimize for ChatGPT, you have to wait for their index to pretty much just update. But the old stuff that you did pre-September 2021 really helps you. We looked at over 82 ranking factors, and what we found was relevancy, which has a lot to do with keywords, and brand mentions were the 2 big factors on them recommending you. Brand mentions being, you know, if someone has an article talking about the best headphones and you get mentioned. Or if, you know, a lot of PR sites are mentioning your company like, oh, NP Digital won AdAid Performance Marketing Agency of the Year. So then when people are looking for a digital advertising agency, they're like, oh cool, here's NP Digital. They won Performance Marketing Agency of the Year. So we're seeing those 2 things, relevancy, which has a lot to do with keywords and not just the keywords using on your own website, when other people mention your products and services, what keywords are they using in those articles, as well as brand mentions. Those are the 2 factors that we're really seeing that help get you more recommended by Bard and ChatGPT.

Okay, so this one's a little different, I think. So just those of you watching, go back to YouTube, watch Neil's graph, but also let's look at this page over here. So there's this table over here. I'm not going to read through everything, but basically you can see for no optimization, The average is 19 in terms of subjective impressions. Now, they're saying the most important things would be around quotation addition, citing sources, statistics addition, fluency optimization, technical terms, authoritative, easy to understand. Overall, what they're saying is the best performing methods improve upon baseline by 41% and 29% on position-adjusted word count. and subjective impression respectively. And so basically they're saying if you do these GEO methods, whatever you want to call it, it should improve. It basically would average out to about 30% improvement in your overall, let's call it your impressions.

Yeah. I think there's a lot of factors, right? I don't think it's going to be one or the other, even in the graph I shared. It's not even about getting a 30% improvement or 10%. It's pretty much winner takes all, and you got to figure out how to get up there. It's just like with traditional SEO. Yes, you want to rank number one, but ranking number two and three and four still gets you traffic. With what's happening with AI, it's I think it's not going to be purely a winner takes all, but it's going to be more of you know it's the top spots pretty much take up everything. But in this case, it'll be you know. Almost winner takes all or someone like that. And the reason I say almost is they may recommend 2 or 3, 4 places, like what's the best, what are the things to do in New York? Or, and if your pizza shop gets mentioned, you're a winner as long as well as the Statue of Liberty that you know is probably going to get mentioned, right? But, and same with like an ad agency. If they recommend Eric and I, we're both winners. But in some cases, they'll only recommend 1 or 2 products. And I think it's going to be more of a winner-take-all compared to what search is today.

I'm going to read the conclusion here and we can work towards wrapping on this. But so it says, conclusion. In this work, we formulate the new age search engines that we dub GEO to help put the power in the hands of content creators to optimize their content. We define impression metrics for generative engines and propose a benchmark encompassing diverse user queries from multiple domains and settings along with relevant sources needed to answer those queries. Those queries. And so finally, we propose several ways to optimize content for generative engines and demonstrate that these methods are capable of boosting source visibility by up to 40% in generative engine responses. Among other things, we find that including citations, quotations from relevant sources, statistics can significantly boost visibility. Further, we discover a dependence of the effectiveness of GEO methods on the domain of the query. Our work serves as a first step towards understanding the impact of generative engines on the digital space and the role of GEO in this new age of search engines. So look, this stuff is really early right now, and I think it's interesting. This stuff can really change in the next 6 to 12 months or so. So take this with— I think it's good to be cautiously optimistic about it, but also understand that you need to be adaptable. So that's what I would say. And this document, just so everyone knows, this was a collaboration with Princeton University, Georgia Tech, The Allen Institute of AI, which I believe is Paul Allen's, and IIT Delhi. Is that a big school, Neal, in India?

Yeah. It's harder to get into than MIT.

Hey, there you go. So, this is a legit document. It's from pretty smart people.

Awesome. Well, make sure you rate-review this podcast and let us know what you guys think. Please give us 5 stars. We'll see you tomorrow.