How to Get Recommended by AI (Not Just Rank on Google) - Secrets Most SEOs Won’t Tell You

How long have you been doing SEO for? What's your background?

>> Uh 17 years. So yeah, I I started in SEO when back when it was easy uh just doing it freelance uh in my 20s, but it was mostly like a way that I could just play more video games and not have to work a real job. And so eventually I I joined with someone who was a little bit older and and had a business and became a director of SEO for that new agency and and did that for 11 years. But um just recently decided to go out on my own again and it was it was because of AI. Like Chachbt came out. I saw things were changing and um the founder of that agency just didn't see eye to eye on me. He was kind of like I think it's kind of a fad and let's let's see what other marketing agencies are doing first. I don't want to, you know, stress about something if it's not going to be a big deal. And I was like we don't have time. I got to I got to do this the way that I think we need to do this. And so I just just started the agency I am running now. uh last May and and now we're yeah really focused on AI.

>> The previous director said generative AI is a fad.

>> Yeah, I don't know if he used those words, but that was definitely his like his feeling towards it was just like yeah, let's not let's not jump on this too quickly. Who knows where it's going and and I I just told him I said I think by 20 I said by end of 2026 I think I could do the job of our whole agency by myself the way that this is going. And I I still feel that way. I mean, yeah, we're doing we do we just do a lot more work for our clients at this agency than I did at the last agency. And it's yeah, it's because AI just lets us move a lot faster now.

>> So, are you focused more on getting um citations within LLMs or within automating SEO workflows with AI?

>> It was good question. Those are the two things that we're really focused on. But yeah, so I we still do all the normal SEO stuff. Um, I mean, yeah, like I I agree with you. If you had a ven diagram of like SEO and GEO, most of the stuff is in the middle. Um, but my assumption, and I I'm curious if you've changed your mind on this or or if we disagree here, my assumption is by the end of 2026, essentially every search will involve an AI response. Um, that that Google's main search results page will look more like AI mode than traditional search results.

>> So, That's um that's that's been a debate on on this podcast. I've I mean I've I've said no in in the past and um I probably would still say no. I don't I don't think that they're going to switch to AI mode as a default by the end of this year.

>> And I I don't think it'll be AI mode. I think it'll be some mixture of like um

>> AI overviews are already in every search though.

>> Yeah. Yeah. Yeah. I think it'll be I don't I think it'll be something that's more similar to AI mode than their traditional search results. Um I but yeah, they're just playing around with so many they're experimenting with so many different formats that I I think it's it's it's unlikely to be exactly what AI mode looks like right now. But but we're operating with that assumption. So when we're doing SEO work, we're thinking how do we rank in AI, not how do we show up in the 10 blue links.

>> So So actually that's a question that I have. How different are those activities for ranking an AI versus the 10 blue links?

>> It's the activities themselves are the same. It's just it's just the um that the terms we go after are are different. I mean, it's a lot of what you talk about with with compact keywords, like we're we're almost 100% focused on bottom of the funnel search terms, but now we can just get a lot more granular. Um, because instead of trying to rank for humans, we're trying to rank for AI and and the AI just searches in a much more granular way. They search for keywords that are way more specific than a human would search for. And it doesn't really matter if you're rank one or you're rank 10 as long as you're in like the first 20 or so. You're seen on the same playing field as everyone else in that result set. And so yeah, suddenly there's just way more terms available to each of our clients. And so it's yeah, the most effective um tactic that we've seen is just creating mostly blog content, sometimes pages, targeting a ton of hyperspecific search terms.

Are you scaling search terms? So like I don't know, let's say like you have like a thousand pages, desks for scenario, desks for offices, desks for hospitals, desks for kindergartens like

>> and yeah, scaling those types of pages. Is that what you're doing?

>> Yes. But but not not at large scales. We're not doing thousands of articles. Um and we are

>> How many do you think do you think you're doing? or do you does it like scale with other activities like you can't you won't allow those keywords to scale unless you're also scaling back links and scaling other types of content?

>> Yeah, we we don't do a lot of backlinks right now. Um we have a few clients that we're we're actually paying to build back links for, but typically we just try to organically build backlinks. Um but yeah, what we're looking at when we think about how quickly we can scale, how quickly we can create content, it's how big is the brand? Like how much branded search are you getting? how many backlinks do you already have and how many pages do you already have on your website? We we don't want to take a a brand new website and build out a hundred pages in the in the first month. Um, but yeah, for most of our clients, we're it's yeah, we can safely do a couple dozen pages a month and it's it's not going to set off any alarms and and we're not automating the the um content. We're still we still have a human that's working with AI to develop each piece of content. This method of marketing is so effective, I had to make sure it wasn't against Google's rules before I kept doing it. It's a form of SEO I call compact keywords. Whereas most SEO focuses on putting up articles to answer questions, how, what, when, compact keywords focuses on putting up dozens of pages that sell to searchers who are actually looking to buy. These pages rank on Google and convert so much better than normal that when I discovered this years ago, I couldn't believe this was allowed. It's less work, too. The average Compact Keywords page is only 415 words. Compact Keywords is a 13-hour deep course on getting sales with SEO. A customer recently said, "Each lesson is dense with information. You're giving years worth of experience boiled down into 15 to 30 minute lessons with no filler or fluff. I feel like I'm gaining a new superpower. Compact Keywords is about setting up an SEO funnel that brings you sales for years and years and years. It works with AI. It's less work than traditional SEO and it makes way more money. You can get it now at compact keywords.com. Back to the podcast.

If you were to start a new brand, and I love this question. You're you're to start a new brand and you want to start getting cover. You want to start getting surfaced in LLMs within the first, let's say, 180 days. You have half a year. What is what does your playbook look like?

>> It's tough. It's it's it's easier with large language models, but it's still tough, right? Because you still need if you want to show up in large models, you still need to show up in search engines. and especially Google is going to be really picky. Uh now we found chatbt not picky at all. So so we had uh let's see one of our first clients they had they hadn't even launched their website when they signed on with us. So we're start with a brand new website and he was he was like he was ready to go aggressive. you know, we always feel that out with our clients first and and he just want to do like whatever gets me in chatbt and so we just started putting out listicles and we were doing like two listicles a week putting him at the top of all of them and chatbt picked him up within like a month and then he was crawling up um you know, just really hyperspecific terms still loves him he's still struggling with Google so I think like

>> can you talk about that what does so what what does that look like um that Chachi BT loves him but he's struggling with Google. Does that mean if someone does the same prompt on ChachiPT or like let's say this take you take the same keyword on Google that maybe he's not performing well with and he and and you turn that into a prompt using the same language. Does that mean that he's getting recommended right away in Chacht even though he's like on page two of Google? Like what does that mean?

>> Yeah. So he's in a really competitive space. Uh and so when I say Ch loves him, he's not at the top of his industry. He's he's going it's uh it's lead genen. So um so there there's some big players in in in the field. Um but he he was quickly getting to like 20% of the time we're doing prompts for his terms. He's in the mix being recommended where on Google, yeah, he's still a brand new website. We're still it's still how it always is with Google in those first few months. Um so yeah, we're not tracking keywords, but yeah, tracking the same prompts across like Google AI mode and and chat GBT.

>> You have a system. So you do you track you track prompts but you don't track keyword performance.

>> Yeah. So I we use two pieces of software to track results. Um Google search console has always been my favorite since dawn of time. Um and now and now we use Peak AI which is which is one of the many. It's like profound is I think the the leader in the field. Um Peak's our favorite. It's it's like profound but much less expensive. Um, and so and and obviously tracking in large language models isn't nearly as precise as tracking in like Google Search Console. We don't know what people are actually typing in. And even if we did, the search volume would be one for every prompt because no one ever types the same prompt twice. Um, but yeah, so so we I don't do a lot of rank tracking. I just I don't find it super reliable. Um, is so yeah, we use Google Search Console for for organic tracking.

What do you think the biggest mistakes people make are when they are trying to get more citations or just any citations within LLMs?

>> I think I I made this mistake. So just backing up, I think when you start to think about, okay, I want to get cited by large language models, I want to get cited in AI, you you kind of break down how it works, right? And so for anyone who doesn't know, someone puts a prompt into chatbt or Google's AI mode, the AI is then going to do something called query fan out. It's going to look at your prompt. It's going to break it down into three or six different searches, and it's going to search its own index, whether that's Google or who knows what index CHBT is using at this time. They they're definitely taking some from Google, but but not all of it. And um and then they're going to look through hundreds of possible pages and they're going to choose to site the ones that they think are most relevant to this prompt. And then the summary you get back or the response you get back from the large language model is essentially just a summary of those pages that it's cited. And so it seems pretty clear if you want your brand to get recommended, it needs to be recommended on those pages that are cited by the model. And so I think what most people do is they they go after that first. you can you find the pages that are being cited and you say, "Okay, how can I get my brand mentioned on that third party website or at the top of that listicle?" Turns out that's really hard to do and we put a lot of effort and we still do it because there's there's things that are worth doing like sure go leave a comment on that Reddit thread that's being cited a lot. Easy to do. Um get a few reviews on that that directory so you can move up higher in their listicles. Easy to do. Um, and then we'll reach out to like the top 20, the top 50 articles and say, "Hey, could you recommend us and we'll recommend you in our article and and maybe we get a handful of those, but you run out of the easy ways to do that pretty quickly. So, what we found that's most effective, what we do 70% of the time is we just create content to replace those pages that are being cited. We just we look at, okay, this is what the large is searching for. these are the kind of articles they're citing. Let's just write better articles or let's write more specific articles. Articles so specific no one's even thinking to write them because no one no human's looking for this search term. And and that when we're like really target targeted in that way is much more effective. It's like it's stupid effective. It feels like early days of SEO. We can just go and like just grab those placements because there's so little competition.

So >> what what would a SER look like? I mean because is it is it like okay number one is is your site targeting the keyword. Number two is YouTube. Number three is like a Reddit post that you made. Number four is an Instagram reo that you made targeting the keyword. Number like are you are you creating properties on Medium uh to to do this with? Like Yeah. Are you creating are you just spinning up WordPress sites that kind of look real and doing this like Yeah. What are you doing?

>> Doing all of that. Um, it really it depends on the industry. So like, yeah, there's some industries where Reddit is being cited in almost every single response. There's industries where Reddit's being cited less than 1% of the time. So like, yeah, any of these, you see all these like LinkedIn posts and videos now like Reddit like they love Reddit or they love LinkedIn or they love Medium. It's like maybe it it depends on your industry. There's like so we it's one of the first things we look for. We bring in a client, we put them into peak, we start to track if Yeah. If the models are looking at LinkedIn, if the models are looking at Medium, we're going to post there. Now, in our experience with those two sites in particular, really hard to get noticed on a brand new Medium account or a brand new LinkedIn account if if you don't already have some following or traction there. It's like we'll do it because it's easy enough to just spin up a Medium version of the same blog post. Um, but where we've seen the real traction is where people already have a following. Instagram I've never seen cited a significant amount of time. Uh nor like Tik Tok. Um now those will show up in Google search results so we'll still do them. Um but yeah, YouTube is really big for a lot of industries. Um I'm trying to think. Uh and then yeah, spinning up a second website like you said, that's something that we're doing with a lot of clients. So we have a handful of clients that already had a second website. It's just yeah it's a game changer because like like I said 70% of what we do is writing blog posts but now we get to double that with a second website. We can we can write the same topic on both website recommend the same brand and now it's just reinforcing that.

So you said you watched the podcast.

>> Yeah.

>> Did you see the episode that I made on a I called it parasite properties and it's like a property that you build up over time. You post to it. It doesn't even have to be very consistently. You know, maybe you're posting to it once every two months or something or or 3 months and it's not even a deep post, but then when you need it to to snag that keyword uh because you need more variety in the SER or something, oh, look what we got. It's a secret weapon.

>> Yeah. Yeah. Yeah. So smart. Yeah. I've heard you talk about that a few times. It But I think even with with large language models, it's it's even more effective than it was in in traditional search. So, so you're you're creating these parasite properties. Well,

>> yeah, we've just started to create some, but we when we do it for clients, we tell them like this is a long-term asset because like you said, like those first six months, they're going to be hard, especially in Google. And so, you're not you're not building it to just immediately start to to influence the response. But, you know, 6 months a year after you build it, you'll have something really powerful. So, so what do um when you are building these like these second websites to influence I mean cuz honestly if you do it well it's also going to influence searchers. What does that sort of thing look like? So typically it's going to be like an an industry resource, right? It's going to be a blog. Um, we'd start with like a few foundational like guides answering like the biggest questions in that industry and then just build out supporting content around that. Um, and then eventually we're doing so at first we're just trying to get some kind of authority, right? We're just building the kind of content that's that's likely to show up, get some clicks, get some links. We're building those foundational links. But then once you're showing up, once you've built some trust, then we're focused on those bottom of the funnel hypersp specific terms, I can actually show you. Let me pull up I have like a list of the most common uh like topic templates that work really well for influencing large language models. They'll sound very familiar because they're very similar to um

>> kind of topics.

>> Yeah. Yeah. So um the cost of X in Y or the cost of X for Y.

>> Ah yeah. Yeah. So you're so you're going after and I so like this brings it brings me back to the first question like you're going after all of these very similar templated keywords. Is that correct or like and you're scaling dozens of those a month?

>> Yeah. So so typically yeah we'll come up with like 12 different templates. Um, and not all the topics are templatized, but yeah, most of them. So, we'll come up with like, yeah, a dozen different templates and maybe we'll do like one from each template.

>> What are like like if Okay, if I was to look through one of your one of your clients blogs, what like uh let's here let's just like make a pend let's stick with desks. Desk desk works. So, okay, you have like an ecom client who sells desks. The best desks. And that's such a good brand name. The best desks. And uh yeah, so what would the what would the blog look like targeting like desk keywords because you said you're doing a lot of SEO with a blog?

>> Yeah. So it it really depends on the client and and this is one thing I found just running this agency right now is it's it's easy to create content that ranks. It's hard to create content that ranks and the client is happy with. So we do have a few clients are like just do whatever it takes. just go ahead SEO the hell out of my website and and just do what it takes. But most clients, they want it to represent their brand well. And so that's where most of the work is. So some of our clients, we you you would go to their blog and you would see that same template over and over again because they just it that template's working and so we're going to hammer it until it stops working.

>> What would what would a template look like?

>> Yeah. So, yeah. So, yeah. So, the, you know, how much does a mahogany standing desk cost? How much does a, you know, a 10-ft standing desk cost? How much? You would, you would just see a bunch of those if that's

>> you. That's interesting. So, you would structure that as a blog or would you structure that kind of like as like an FAQ subfolder?

>> Yeah. Yeah, we yeah, we might do Yeah, we might do single FAQs if it was a if it was a short one, but yeah, with the for the for large language models, it's almost always the longer posts that get surfaced. It's it's really like fact density is really important. And so, we're really big on just like just flooding the page with just facts, figures, tables, anything that makes the large think that this is like, you know, a wellressearched article.

>> That's interesting. Do you ever um do you ever kind of like so David Quaid connected us and uh and David came on the show and he shared his uh people also ask strategy and actually I had Mark Williams Cook on the show two days ago and he has like a tool to get people also ask questions. Um and I I thought David's people also ask strategy was so cool and what he does is he farms people also ask questions. He gets th the questions that are relevant gives them to perplexity has just like a markdown file like plain text file generated of just answers to these questions and the answers are one paragraph that's it. It's like it's like four sentences or something. uh he takes takes those he has like the FAQ subfolder with like let's say it's like I don't know like people also ask within uh jackets it'd be like or leather jackets leather jackets FAQ slash the question and the the question is the H1 it's the page title and then the body text is the answer that Perplexity made but what's super cool about this is lots of people don't target these longtail people also ask questions and so the pages tend to rank pretty well. But then what he does is once the page is ranking then he builds it out with the extra information. So he's not putting all the work into it right away not knowing if it's going to rank only once it starts ranking he goes and puts that extra work in and then he uses that as an asset to interlink to his pages which is also very cool. I I thought that that technique was like I I love that. I don't know if you're doing anything like that.

>> No, me too. He he brought that up on our conversation. He just talked he talked about putting up pages with just two sentences on it and then just seeing if it gets traffic or even impressions. Yeah, I thought that was so smart. Now, most of our clients don't want us doing those kind of tests with their websites. Um but I think yeah, that's that's one that I will definitely steal. I think it was it was super smart. But I think there could be a way to though to structure it in a way like in a way where like it doesn't hurt the brand because I understand like being a client you'd be like this is this looks spammy or scammy or weird like I could totally see a client being worried about that but then I could also see it like if you really focus on like the UIUX making it look like hey we have this FAQ trying to help our niche and it's also it's only linked to in the footer. it's not linked to in the top navigation. Uh, you know, because you you don't you don't want to throw lots of like weird SEO assets into the top navigation and reduce click-through rate on the the links that actually matter,

>> but you could you could do something like that in like a resources section of the footer.

>> That's where like

>> Yeah.

>> You know, I Yeah, I think you're totally right. I think there is a way to do it. And I think like pages like that can totally rank well in Google, but they don't get cited very often by AI

>> because they're not bottom of funnel. Like because even if they even if they do get cited, it's like it's top of funnel, so you don't really need to click through.

>> Yeah. Yeah. It's it's cuz and and we have lots of cases like that where like, okay, our clients getting cited, but the AI is not recommending their brand. And so that's that's the key is like how can we how can we get cited where the the citation they pull out also contains a recommendation for the brand so that the the response contains that recommendation as well.

>> Yeah. So so you basically like the options that you're faced with is like how do we get our brand onto all the the the pages that are actually ranking for this keyword. All right. That's not the best. Let's create new content.

>> Yeah. And Yeah. And that that works. I I don't think it's going to keep working well because I think people are gonna catch on. Like,

>> what do you mean?

>> The the main reason it works so well right now is because most of the terms we're going after have little to no competition because people aren't getting as granular as we are. Even, you know, we're going up against sometimes competitors that are 10 or 100 times the size of our clients and they're not bothering with these super niche keywords. They're going after the big ones. But it's it's getting easier and easier to create content with AI that's high quality. And that's like because the hard the easy part is making the content that ranks, right? And you can totally automate that. And like Google's not going to catch you unless a human actually sees that you're creating this content. The the algorithm itself very easy to fool. I think um it's only when a human sees it and then you get a a scaled content abuse penalty that you have to worry about. But I think the hard part is is creating content that sounds like it was written by someone at the company that that sounds like it has some personal experience. It has the brand. And that's that's the part that we're putting the most effort into solving. And I'm sure a lot of other companies are too is building this brand ambassador that represents the company that has all the information that the people inside the company would have and then can then create content as if they were a human in the company. And I think you unlock that now you can really scale this and it's it's no different than content that would come from a human.

>> The brand ambassador being like the second website or like a uh Instagram property or you said not Instagram YouTube like or YouTube property

>> or like a person who is going and posting across social media platforms.

>> So yeah, sorry I should have explained that. So, our brand ambassador is a a project inside of Claude. And so, this is it's like a it's it's a version of AI that represents your brand. And so, we when we bring on a new client, we do a long interview with them. We gather any resources they have, social media posts, PDFs, everything, and we organize it into usually about a hundred pages of information. And then we we uh structure it in a way that it's easy for Claude to reference what they need. So, we'll write instructions for clouds like this document has every case study they've ever done. This document has a description of every product and service. And then when we draft content, when we come up with strategy, if we do social media posts, it knows it can pull, you know, here's all their internal pages and the keyword you should link from when you link to these pages. Um, here's here's a message or here's everything that the founder of the company's ever posted on LinkedIn. So, you can you kind of understand their opinion and their voice. And so I and I I talked to other agency owners that are doing the same thing because I think this is like it's the last problem to solve in automating or turning everything over to AI in terms of marketing. And so I think I think there's a few of us of there's a few agencies like ours that are kind of at the forefront of this and I think everyone else is probably going to catch up and this is just going to be the game. It's going to be how do we create highquality content at scale that targets the terms that AIs are looking for? And then I think a lot a lot of these terms are I don't know the internet's going to be a crazy place in a few years. Like there's just going to be so much content and it's it's actually going to be pretty good content, but it's going to be created by AI. I think about that with like Reddit specifically. Like it's like Reddit's already flooded with AI, but like some of the AI responses are pretty good. Like you're going to have a bunch of AI bots talking to each other, but it's not going to be like it's going to be high quality content, like high quality conversations going on with AI, and I don't I don't know what that means for the world.

When you're when you're putting out this AI content, it's being human proof read or it's just going through different layers of uh bots.

>> Yeah. Yeah. So, we we do research with AI and then we run that through a series of prompts and then Yeah. one of our account managers is reading through it, sending it to the client some and and usually like the first one or two articles we write for a client, they have to spend a lot of time going through and just and we tell them to like please like be super critical. tell us everything that's wrong with this article, everything that you would change because we want to we want all that feedback and then everything they say on the article, we'll incorporate that into the brand ambassador. And so the next article comes out a little bit closer. And usually after like two or three rounds of that, it's being written exactly how they would have written it. It's it's getting very few things wrong and then it starts going pretty smoothly. But we still right now we always have a human involved. I I think there there's probably going to be brands that want us to automate it and then we can do it a lot faster. But um but yeah, right now we're pretty careful.

Really want to I I want this template. The template that you were pulling up before.

>> Oh yeah. Yeah. So yeah. So the first one is Yeah. The cost of uh my favorite one is how to choose the best X for Y because really everyone knows like like Yeah.

>> Well, yeah. So self-promotional listicals work really well. Um, a lot of people like Billy Ray now think that that they're on Google's radar.

>> Well, so

>> yeah.

>> to be fair, um,

>> to be so Harper, I don't know if you saw the episode Harpre saying he's been on the show before,

>> several times before, and we we did an episode about how Shopify was abusing this, but like they were doing hundreds of these, and that's when it becomes a problem. If you're doing like a handful of of of them, then it's really like that's that's acceptable. It's it's the the people who were getting hit had these like crazy patterns of like obvious SEO spam and they fit there was like lots of other problems with them. Like they're all I mean they were all patently and lazily AI generated with like no brand voice. Uh but then there was literally hundreds of them. There was many reasons why these articles were trash.

>> Yeah, I was just watching a live stream with Joy Hawkins and Darren Shaw and that that was her thought was that it's like the high quality ones are are fine and I spoke with her on Twitter actually and and both of us saw the the websites where we saw it stop working was on like new websites without much authority. So yeah, I think it's the same

>> stop working on Shopify. It's it was Shopify.

>> Yeah. Yeah. Yeah. I think yeah I think yeah in that case it's yeah you have to wonder if a human saw it or if it's or or if the algorithm is actually picking that up. Um yeah I I think you can go definitely too far with it if you're if you're doing at at scale or you're automating it. Um but we also just find that a lot of our clients don't want that. They don't want listicles on their website that promote them. And but my theory is it's not that AI loves listicles. It's AI loves to search for best X in Y. And that tends to service listicles, but that's why how to choose the best X for Y is is so great because you can still rank for those same terms, but it doesn't have to be a listicle. It's it's a how to choose article and then of course you also want to recommend the brand at the top of the article so that gets in there. But yeah, we that's the template we use more than any other one. um competitor alternatives if if if the client doesn't mind uh you know talking about how they're better than their alter their competitor. Let's see how to solve some problem that their product or service solves of course and then position yourself as the solution.

>> I love those.

>> But those those I I like to target with pages.

>> Like I love finding those like problem

>> those problem based keywords where somebody who is searching it searching this they are having this problem right now. They are dying for a solution and they will convert very easily and really fast and and it's very lots of times these keywords are extremely undertargeted too. So it's very easy to rank for them and like you said this is getting surfaced in LLMs.

>> Yeah. Yeah. And you can be hyper specific with with the problem and and use case demographic. Um buyer guides. So um reviews. So like we'll do we'll do the brand reviews because you can just grab that since you're the authority on your own brand. But then we'll also do like product reviews. So like you know standing desk reviews or um yeah large love to search for that. also awards. That's if if you ask for like best pool cleaning company in Dallas, they'll look up like pool cleaning company awards and there's very few content online for like pool cleaning company awards in Dallas. You can just grab that keyword and and for some reason large language love searching for that.

>> How are you putting up uh Oh, sorry that

>> No, yeah. Yeah, those are all the templates we use.

>> How are you how are you putting up um like review and award pages? Is it all within your own site or are you doing like I don't know taking like v testimonials or reviews or making videos about reviews and and then targeting them with a bunch of different platforms.

>> Yeah, we'll do videos if if we can get the client to help us with videos, we will. That's it's very rare we can. Um but one thing we've done for a handful of clients recently is do notebook LM videos. Just Google's notebookm will just take a a page that's ranking well or getting cited a lot. Throw it into there, make a if you haven't used it before, it's like it makes like a PowerPoint presentation with an AI voiceover going through it. It's like it's it's pretty good for AI. It's it's not great. Um but yeah, if you're going after a low competition keyword, you can throw that up on YouTube and get another citation there as well.

Something that that's been on my to-do list honestly forever. It's it's embarrassing how long it's been on my to-do list for is to just take the video testimonials that Compact Keywords has, which are literally on the landing page of Compact Keywords and then put like a bunch of them on YouTube with titles like Compact Keywords review, compact keywords rating, compact keywords testimonial, uh, is compact keywords a scam, is compact keywords legit, like put these put these up on on the on my YouTube with those titles. um with just the the videos that I already have and uh because that stuff ranks and shows up so well. I did that as an experiment. I I took one of the testimonials from the landing page. I put it up on my Facebook page with compact keywords review and like eight hours later it was showing number one for compact keywords review. And uh it's so important because people go and search for reviews of your brand and they might not bother to look through your website. A lot of them probably won't. They won't they actually might not see these reviews. They might go to Google and then they see it on YouTube and they see it on Facebook when they wouldn't have seen it otherwise on your landing page.

>> Yeah. Yeah. Reviews are huge. That's one of the first things that we we go over with clients. most of them get it, but yeah, there's there's a few companies out there they they feel like they're bragging by putting reviews on their on their homepage or or on their website at all. Um, but yeah, and and what you pointed out there, I think for branded searches, especially putting out uh or and I know you you're doing press releases for branded terms as well, right?

>> I love press releases. Yeah, that that's um so I'm still like training somebody to to do this. Um, but like yeah, I'm getting a sales copywriter who like they they because the thing about putting out these press releases for branded terms is you need to also know how to write good sales copy. Like you can't write just you can't just be like some like random top offunnel SEO copywriter. You actually have to know h you have because because at the end of the day what do you hope to happen with one of these branded press releases? You want somebody who is looking up the brand to come across the press release which is syndicated on another platform, read through it and decide to convert. And so you need somebody who knows how to do sales copywriting. And then from from that perspective, teaching them how to do SEO is really easy because you just take the brand name and you put it in the heading and you put it in the subtitle and you put it in the first sentence. It's naturally going to be in those places anyway. And uh yeah, so so and works really well with LLMs. LLMs love press releases.

>> Yeah. And they love sales copy, too. It's it's so effective just bragging about yourself.

>> Yes. Oh my gosh. Because when you give like when you describe your brand in a in a clear concrete way and and you're doing it describing the positives, that language gets mirrored.

>> Yeah. Yeah. They'll just repeat it right back into and they they love uh they love numbers, they love awards. And so that's that's our strategy really is we try to get a recommendation as close to the top of the article and then right after we recommend the the brand, you know, and it's on our own website on the brand's own website, but right after that, we try to get like two data points, a number, an award, two numbers, something like that.

Can you So can you give like a Let's make a brand. Let's let's uh Okay, I have a mug here. These are the best mugs. Uh, let's call let's call it um let's call it Ed's Mugs because I'm not feeling creative today. So,

>> we're calling it E Ed's Mugs. And it won uh it won the TJ award for the best mug in 2026.

>> TJ's best mugs. Yeah. Ed won by Edmug. Yeah.

>> Yeah. So, you would So, you would put out uh you would put out a blog post like this.

>> Yeah. So, Yeah. So, we'd have, you know, uh we'd have a blog post, you know, the the best mugs for drinking coffee, the best mugs for tea, the best mugs for secretly drinking whiskey at work. Um and and so the right at the top of the article, we want to address the the query. Now, with actually with with best listicles, you don't want to give away who's the best in the first sentence. You more want to like say how you came up with your list. like we we went we we went through a tons of reviews and and and this is our completely unbiased uh process for determining the best mug. Um and so in that case it there's an implicit recommendation by putting edge mugs number one. But let's say we we had a different article um yeah the cost of of mugs the the cost of uh coffee mugs the cost of tea mugs cough whatever. Um, so at the top of that article, we would we'd want to spoil the answer. Essentially, we would just say the average cost for a coffee mug is $12.95, but there's a lot of things that factor into it. The best mugs are made by Ed's Mugs, winner of TJ's uh, you know, best of 2026 mug award. They've sold 100,000 mugs in the last three years, and then and then go on to do the rest of the article. The thing about the um the branded the branded H1s and page titles or the branded articles is that those are useful for when somebody or an LLM is actually researching the brand. So that's like, okay, you've you're you're considering going with this brand and now you're trying to decide, is this brand legit? And then it's so so the LLM is doing all these all these searches and then it's like, oh, okay, Ed's Mug won the TJ award. Oh, that's legit. Okay. It's going to say, "Yeah, Ed's Mugs won the TJ award two years in a row."

>> Yeah. Yeah. Yeah. And that's Yeah. That's where press releases are are super effective. And yeah, that's that's like that's I've had to change my thinking there because before large language models, I actually didn't do much branded content. And maybe I should have been, but I was always just focused on like how do we get new traffic to this company? How do we get people who have never heard of the brand coming to the company? But like you said, with large language models, yeah, first they're going to do the generic search and they learn about your brand, but then often they'll do a follow-up search specifically on your brand. And so yeah, that brand content suddenly becomes a lot more valuable. So we do like like case studies never really cared about case studies for SEO because humans don't search for case studies, but large language models love to search for case studies. And so now we do a lot of those pages um or FAQ pages. Like we do a lot of FAQs or single question pages, but just like one big page of FAQs. No one searches for that, so they didn't rank, but large language models love to do that. And so we'll just we'll just fill it with sales copy on the brand just all the things we want the large language model to say back to the human put it in the FAQ. Um and then yeah those brand review pages is another one where just yeah this branded terms can be really powerful.

>> What so um what are what are the like uh I have two questions that I want to but I want to yeah I want to ask about FAQs. So how how do you build out um these FA these branded FAQ pages? What is like are you doing a subfolder which is just like root domain forward.com FAQ and then forward a question or like how are you doing it and then what do the actual pages look like? Is there a hub page for forward slaq?

>> So when we're doing

>> sorry and where are they positioned where where's where is the FAQ page even like linked to interlink to?

>> Yeah. So if we're doing one big FAQ page, we're we're doing it just to show up when the large language model is searching for information on the brand. Large love to go to the FAQ page. They'll just they'll just navigate

There from the website to get more information. And so we're not even trying to to rank for common search terms there. We're just assuming the AI is going to going to visit it because they like it. So we're probably going to link to it from the main navigation or the footer. And then it's just that's just filled with like softball questions just and and maybe common objections that that we want to give. We want the large language to give a very specific response to to these questions and we're just going to kind of feed it to them.

If we want to rank for a given question, just like a common question people might have in that industry or for that company, then we would do a separate page. And I think I think the strategy you laid out makes a lot of sense and and and the one that David Quaid um put forth of just making a simple page for each question, answering it, and then going back and filling it out. Right now though, we would just do a a separate blog post for each question, assuming it's a bottom of the funnel question. Otherwise, we we typically wouldn't even bother.

It could be interesting. I just I had this thought that of something that you could do is you could you could okay so you have in your footer like FAQ for your brand and then you click FAQ and then it's a hub page with all of this the common the common frequently asked qu all the frequently asked questions it's redundant all the frequently asked questions about your brand right and and you give you list the questions and you give like a one-s sentence answer for each question but then there's maybe like a button underneath each answer that says more. And then you can go into a dedicated page for that question. And I guess this really only matters if you're like an established brand and people might really want like deep deep answers. It also might be completely unnecessary. This is just like an idea that came into my head. Um but yeah, that's like but I like the idea of putting up all these FAQs so you completely control the narrative and give a very specific answer for this.

No. Yeah, I think that I think it could be powerful in the right situation. We we've had a few brands come to us recently that have like a PR problem. Um there's just bad information about their brand out there or there's uh in one case there's like a a Chinese company that's illegally knocking off their brand and and they want to kind of like spread that message. And that was really hard to do in traditional search, but it's it's a lot more accessible in AI because like you said, like you know, if we had an FAQ that talks about um you know, should should I buy the cheaper version of this instead, the short answer might not be enough to convince the AI, but if it clicked into a longer answer and we had a bunch of data and supporting evidence, then it's more likely that we're going to sway them to to take our side in response.

Yeah. And so, you know, you build out the pages for the the FAQs that only really matter.

Yeah.

Yeah. And that's like

Yeah.

It's like Yeah. And that's that's another huge opportunity with AI right now is is you can target the information that people don't know they don't know. Like the things that they're not even searching for, the the the things that you should be considering, you know, like like when someone goes into hire an SEO, they don't know what questions they should be asking. Uh but but you can get those answers into the AI response even if someone's not asking it just by you know creating content like yeah what what you should know before hiring an SEO which people have been writing that topic forever and it's I don't think it's ever done well because no one searches for what should I know before I hire but but large language models do and so there's like yeah whole new uh avenues into influencing the the AI response now.

So I want to go to your uh content creation process and using cloud.

Um,

All right. So, I I want you to speak to somebody who they haven't really used cloud much at all. Um, they maybe they have like a cloud account and like they've just really been using chatbt, but you know, they keep hearing people talking about using cloud for SEO and to create content. How will you tell them to actually set up Claude to make great SEO content about their brand? What is everything that they need to know?

So, yeah, the first thing they need to know is you need to teach Claude everything that's important about your brand. And so, we have what we do is we we have a long questionnaire we go through with every brand. And before we get on a call with the client, we have AI do a bunch of research to try to fill out as much of the questionnaire as possible so we're not wasting the client's time. So you could do this for your own brand, right? Have Claude give me a comprehensive questionnaire, everything you might want to know about my brand, and then go do research and try to fill it out so you aren't wasting my time. And then and then you just like dictate the answers to all the other ones, right? Don't waste your time typing. Just just voice the text, fill out this questionnaire. That's a great starting point for a brand ambassador. You'll want to add more later, but now it understands at least the basics about your brand. You're going to create what's called a project in Claude. And these exist, they're like they're called gems in Gemini or or projects in in JBT. Um, put separate that information into distinct documents like products and services, customerf facing information, however you want to subdivide it, frequently asked questions. And then in that project, you also have instructions and you can in the instructions just define what each document is and when you want Claude to refer to that document. Cool. Now you have a basic brand ambassador set up. You can use Claude and it'll speak on your brand's behalf. The next step is to pick a topic and you want to you want to provide some foundational content on this topic so it's not coming out generic. And so there's two main ways you'll do that. Either you provide that information and again you can just ramble about the topic for like two to five minutes into a mic transcribe that cool here's the foundation or you have AI go and do some research you know find find what people are saying online about this find some some supporting evidence and you can do both we we actually will do both typically give it the transcript and say go get me a bunch of research that backs up everything I just said in this transcript right and then you throw that into Claude with some guidance on how to create the article So, you want to give it just we have a we have like an SEO writing guide that we keep up to date and this is like Claude's Bible for how it's going to write SEO optimized copy but because it understands your brand because it has all this supporting information. It has your voice talking talking on the topic. It doesn't come out generic, right? It comes out as if something you had written and and really researched. And we also we do I know what you do Edward too is we just say like don't use these words. uh don't use uh it's not X, it's Y. Don't don't use like all these common uh phrases that that AI uses.

Yeah,

I we still have to manually remove M dashes at the end because no matter how many times I say don't use M dashes, it still uses M dash.

So frustrating.

Yeah. So, yeah, it's a it's still a process, right? It's not just like click a button, go. But um what I find is we we write articles a little bit faster, but they're way better than when we were like paying humans to write them a few years ago.

I use AI a lot for my copy. People don't know because like I I talk so I I'm like people think that I'm like super anti- AI. Um I get very frustrated with frustrated with AI. I think all of us do. I was extremely frustrated with AI using it to create a page today. But also like my copywriting is way better. Like it's it's like it's it's

AI gives way more detailed copy than humans do. Like you might write

sometimes too detailed, right?

Sometimes too detailed for sure. And you actually have to make it less detailed to sound more human.

Yeah. That's why we we use uh Claude Sonnet for writing because if we use Claude Opus, it's like it's too smart. It just and it sounds like very smart. It sounds like Spock or something like it. We need it. Yeah, we need something with a little bit more uh I don't know. Sonnet just has more natural language, I guess. I we we find AI photos can can do well, too. Um, yeah, I I wonder I I would I haven't seen a study yet. I don't know if you have on on AI photos versus real photos. Um, and and what's your take there? Do you think Google is actually like seeing, oh, they have photos, this page should rank better, or do you think it's like people are staying on the page longer and it's it's sending positive sign signals? I think people are less likely to pogo stick when you have real relevant photos and if you're making a bottom offunnel page which is about actually converting people I I think it's having a real photo builds so much more trust and increases the likelihood of conversion so much more. I'm not even thinking about Google at that point. I'm thinking about I'm thinking about because if you are making the the truth is like if you are making a page that is optimized for the keyword that will also convert people, it's sending all the right signals.

Yeah. Yeah. That that's our thought too is we're we're adding photos for people and we're kind of hoping Google's going to notice that people are finding the page more useful because they're staying on it longer converting. But yeah, typically the the most common photo we add is an infographic. So, we'll just have and I'm all I don't I I love SOPs. So, if if I like I like the infographics because I can write an SOP for it. I can have the AI find like claims in the article that would make a nice infographic. I can give my team the instructions to take that to Gamma, make the infographic, plug it into the article, and now we have these big like catchy graphics throughout. But yeah, like real images are definitely better. They just take a lot longer. they take a lot longer if you have a brand like when you're working with a brand you could also just instruct them like hey take more take more photos we need them or like a SAS for example you can reuse the same so this is a little bit spammy but it works is uh you can re like you can you can reuse the same image across bottom of funnel SEO landing pages and it just has a different file name but it's like the same the same image the file name is is optimized for the page and the alt text is optimized. It depends on the CMS. Some some like some CMSs you or you need like a plug-in in order to have um multiple alt in order to have alt text for the same image different alt text for the same image across different pages. Uh yeah, but but like um just using the same image with different file names across pages and having the alt text be optimized to that page. So, like, you know, your your SAS could do lots of different things and it could serve lots of different purposes and you have the same image of the SAS. Maybe it's like the main screen of the the software and you just have the alt text and the file name specific to to the page and to the use case that you are targeting, but it's the same exact image, you know.

Yeah.

Um,

No. Yeah, that's smart. I think for a lot of SAS companies, we'll just do we'll we'll find like an animation style that fits with the website and then we'll we'll kind of train like a a custom GPT to create images in that animation style and just generate them for the

that's what a lot of and a lot of people like doing that with their SEO pages, but my whole take is like I'm trying to convert somebody and what's more likely to what's more likely to actually get somebody to convert? a AI generated image that doesn't actually show the brand and doesn't actually show like let's say it's software doesn't actually show the software or the software itself because like here you are saying that this software can be used for this show screenshots people want to see screenshots and that's going to build so much more trust and increase the likelihood of converting so much more and like it's if you're targeting a keyword that your your software doesn't satisfy properly And then you're like, "Okay, well actually no, our software doesn't actually satisfy this keyword." Well, so we can't show a screenshot. What do you do in even making that page?

Yeah. Yeah. No, that's a good point. Yeah, I think you're right on that one.

So, yeah, I I I love I'm I'm a huge fan of like real photos versus like I'll use stock images though for for top offunnel content all the time. I'll use AI for top offunnel content non-stop. But when it comes to actually creating bottom ofunnel SEO landing pages meant to convert people, meant to get customers, to get users, to get leads, I'm Yeah, I'm using as many real photos as I can.

Good call.

Yeah, you're using cloud co-work non-stop.

Yeah, if I don't have it working on a task, I feel like I'm being unproductive. I try to have it always doing something in the background while I'm working. I mean really I I think what's more impressive is just Opus 4.6. I don't know if if used used it much but it was Opus 4.6 and then CHBT's codeex 5.3 both came out on the same day and then and then the stock market came way down right all the SAS stock crashed and I think for good reason. It's it feels like another catchy BT moment to me like the the step change in capability we got with 4.6 six is just I didn't expect this is what I expected like a year from now in terms of capability. I think it it shook all of us and it's I'm I'm just able to trust it with tasks I never would have trusted before. I give it like I mean yeah so I guess just yesterday I was coming up with uh content ideas and so I scraped like the thousand most common topics that large language models are citing for this client. I give it that I give it our the most common templates I went through with you and then I give it like 2,000 URLs from their website and just like figure out what topics are working figure out which topics have already been covered on their website and it just created this beautiful presentation going through just thousands and thousands of topics and URLs and cross referencing all of them like and you know I check it all or I check I spot check it and it's it's accurate like it's it's not making mistakes the way that previous models would. It's there's just a whole new world of things we can do with AI now.

So man, so first of all, I would love to do just a dedicated episode of you showing how you're using cloud co-work sharing sharing your screen. Yeah, that would be that would be great.

Yeah, that would be cool. Yeah, the the D I think you had the same issue on the episode you did um where uh clockwork just takes so long that may maybe we might have to have like the Martha Stewart method where we we start one and then I show you one that's already run and then

yeah yeah

uh also yeah using it I wouldn't say that SAS is over but like so I'm literally starting another venture right now where I'm paying like a a hungry young founder to do my method of SEO for a new vibecoded SAS. But guess what? We are not picking a niche where people want to vibe code their own SAS. We are and and funnily enough those niches also the the niches um where people don't want to vibe code their own SAS also have poor SEO. they they have like a lot of bad marketers and so like they're just really easy to conquer and um and and so like yeah this I'm just like advising and funding the the project and I own most of the equity and uh the young hungry young founder is really smart and getting paid very well and we're picking we're going into a niche with poor SEO where people don't want to vibe code their own solutions and then we are vibe coding the solution and then we're talking to actual ual users a asking like okay what are these people trying to accomplish here uh and how can we like Steve like how can we Steve Jobs it and give them something that basically what they want but they don't know that they want it, you know what I mean.

yeah I think it's super smart I mean I think you have to like you have to be prepared to be disrupted very quickly right so you're either your software has to evolve very fast or it's like a this software is going to exist until there's some AI agent that can just do this out of the box.

Well, to to your point, um like with another with another SAS that we have, we are making two months worth of iterations in an hour using Google's anti-gravity because

awesome

because the technology is like two month like we've worked with so many dev shops they will take two months to do this and we are doing it in an hour with anti-gravity. Yeah, that's yeah, it's it's crazy the world we're entering. I was just on uh Marie Haynes community and she it it's crazy how quickly she's building software to just do really simple things. You would never build software for it. It's like she was realizing it takes her a long time to take a blog post from a Google doc and put it into WordPress. And so she just went into anti-gravity and coded software that will take her Google Docs and publish them to WordPress for her.

That's cool.

Just custom software to do Yeah. Just simple stuff. That's also why I have no interest. Even though I have like such a big marketing audience, I have no interest in making like a martekch

or like more marketing. I just don't want to compete with like with with other mark. I don't want to compete in a niche where all of the competition is savvy. I mean, that's not like actually true. There's lots of dumb dumb marketers within but but you know what I mean. Like I don't want to if I'm doing SEO, I don't want to compete with Hrefs. I don't want to compete with Seamrush. I don't want to compete with all the geo tools. Like I don't want to I don't want to play. There's no there's no need there for for me. There's no need. All right. I want to I want to ask um because you know we've been talking about AI SEO so much AI and SEO multiple forms of it. What skill should SEOs double down on that AI cannot replace? It's uh this might sound like I'm walking around your question, but I really think that we're getting to a point where the the one skill that matters for everyone above everything else is just your ability to use AI. And and and it's it's I I really think as knowledge workers, fast forward 18 months, two years from now, we're all just managing a team of AI agents. And like right now having good judgment is important, right?

Critical thinking.

Yeah. Yeah. I like that is really important right now because the AI is going to give you something and and you have to evaluate it and it's going to get things wrong. Um like I can't have AI do keyword research for me right now. I I can have it give me a bunch of keyword ideas, but then I have to say like, nah, that's too competitive. That's not the way people search. Um but I don't see that lasting much longer. I'd like 4.6 Opus 4.6 actually has pretty good judgment. Not good enough to do my job yet or your job. Um, but like real judgment and like taste that thing that we thought AI would never have. Only humans have good taste. Like I think we're starting to see that 4.6. And I think it's just a matter of maybe a year before it does have that kind of judgment. And so I don't think you can hang your keys on or hang your hat on that. I think uh but just being able to articulate what you need from the AI, just knowing how to use tools like Claude Co-work. I don't think you need to be like out there with using Cloudbot and like on the bleeding edge of technology to stay relevant, but like at least be using AI in some capacity on a daily basis. I think that if if you can use just modern AI tools efficiently, effectively, I think you're going to have a a your career is going to look good for a while. Three, five years from now, I don't know, maybe all of us are replaced. But I think right now the the really just the master skill for anyone working on a computer is your ability to use AI and and specifically AI agents about Okay, so generative engine optimization which we've been talking a lot about without using that term.

What geo advice is there any geo advice that's popular that you strongly disagree with?

Oh so much most of it. Yeah. Uh yeah, schema. Yeah, structured data being like really important for large language models. No. Um, yeah, like uh making markdown versions of your website. It it was it was wild to me that Cloudflare is supporting that now. And that did make me do like a is does Cloudflare know something? I don't know. Um, but so far

I had that thought too when I saw this. I'm like, "Oh, should I be making content on this?" And I was about to cuz like it was going to make for good content and when I researched it, it was like, "No, actually it's like a nothing burger." Yeah, I that's the same reason why I haven't done a video is because like I don't really I I don't want to just say Cloudflare's wrong here because I I think they're smart and they know what they're doing. Um so yeah, I just I don't have an opinion that I could uh confidently put forth there. Um yeah, things like oh just just posting a lot to social media. And this is actually something that I was promoting eight months ago was just like, hey, the AI is looking everywhere. Post everywhere. But the the truth is the AI is not looking everywhere. I mean like maybe when maybe during the training phase like and and so sure the next model maybe is trained on your data if you're posting everywhere but just like in real time when you use large language models it looks at the same Reddit posts over and over again. It's not it's not scouring Reddit and finding a bunch of threads. It's like it's like looking at the same 20 threads every time you prompt it. And so and and in most social media sites it's not looking at it at all. And so if you really want to affect their the response, you shouldn't be just like spraying content everywhere. You should be like laser focused on the the places that the AI is actually looking at. Um yeah, this idea that AI is like really good at understanding your content and and knowing if it's like trustworthy and and like written by an expert. Like no, it doesn't. It'll it'll just believe whatever you tell it. So, like just brag about your company. Just tell it why you're the best. Don't hold back. Yeah. I I mean really it's it's most of the advice on geo right now is pure speculation. It's it's there's a lot of grift out there and uh yeah. Yeah. I would if you have any other tips. Yeah. What else are you seeing in terms of advice that people are giving on?

I I agree with all of that. Uh yeah, I think I um I think though people get obsessed with getting cit getting like citations for the sake of citations and it's kind of what you said, focus on what converts because it's like you can you can have a bunch of top offunnel citations and like some people say, "Oh, that's going to you know, they're going to say when people are doing research, they're going to see your brand everywhere." And I just think it's so much more effective. It's just it's so much more effective to make bottom of funnel content. I'm just always

100%. Yeah. The goal should not be to get cited, right? You can you can

The goal should be to get recommended.

Yeah. Exactly. Yeah. It's all about being recommended and and Yeah. So, if

if the model's not coming back and recommending your brand, then it doesn't matter that you got cited.

I still think No, Go ahead.

No, no, please. I was going to say a lot of SEOs are panicking about um the fact that people aren't clicking through to their website. You know, website traffic from Google's going way down and impressions are going up. And I I don't think that's necessarily a bad thing unless you're a publisher. If you're a publisher and you rely on like ad clicks or affiliate sales, I actually think you'll be fine long term. And I could talk about that separately, but I think if you

please talk about that. That's super interesting.

Yeah. I think Yeah. Okay. So, there's two situations. You're you either make money from website visits or you sell an actual product or service. So, we'll take the hard one first. If you make money from website visits, yes, it sucks right now. Your your revenue is certainly down. But think about what you were selling. You're selling ads. You're selling affiliates. You're selling influence, right? Your your website is valuable because you can influence the purchase behavior of humans. you still have that power to influence purchase behavior. If the large language models are citing your website but not sending humans, it means they're using your website to determine what product or service to recommend. And so just like you were selling ads or you were selling affiliate products, you can sell those recommendations, right? Like if if I was paying you to be an affiliate for my product and I get listed at the top of your listicle and I and now I'm getting sales, I'm still getting a lot of benefit being at the top of your listicle. As long as the AI is citing that page, it doesn't matter if people visit the page, the AI is going to come back and recommend my product. I'm still getting more sales. And so I think I think that's what we're going to see. I think we're going to see a lot of these publishers change their model to just sell the it doesn't need an affiliate link. It doesn't need an ad. you just sell the recommendation itself.

Question is how much will how much will I guess you you have like prompt data and stuff because you know you need to have brands will require some data in order to to pay. So that that's where a tool like profound or a tool like peak that we use is uh is is yeah really necessary so you can see when people search for and you have to track a lot of prompts so you get like a repres representative sample but then yeah you can see which articles are being cited and if if they're citing this website a lot and then recommending our product it obviously it's it's having a big impact. It's Yeah, obviously it's not as good of as tracking as as if you were doing affiliate sales um or ads, but I I think it's it's definitely more precise tracking than like a billboard or a radio spot.

I'm a I'm a fan of the second solution, which is making if you're a publisher just making your own product.

Yeah. Yeah. And if you sell an actual product or service, I don't think you should care about a loss of traffic. Like if as long as the AI response is recommending your product or service, I think that's better than than your website showing up at the top of search results. People are way more likely to take an AI recommendation than to just buy from whoever shows up at the top of Google.

I believe that it was like LLM recommendations actually convert four times better than organic search.

Yeah, it depends on the industry. I think hrefs found it was like 21x for their brand. So

Wow. Yeah. Yeah. Yeah, it converts. It converts a lot better

if you um we talked about this at the beginning. Uh but like I I really want like just more exact tactics. Um especially I mean you said you don't build many links, but I'm sure you you would in this scenario. Um so if you had to like rebuild visibility from zero starting like today, what would you do first?

Yeah, So, I mean, yeah, get get just a basic website up obviously and then foundational links. So, if there's directories you can get yourself into, you know, create social media channels for yourself and then just we have a brainstorming activity we go through. There's probably a lot more people that are willing to link to you just in your life than you realize and and you kind of have to scrape the bottom of the barrel to think about just every person, organization, entity that you've ever had a relationship with that has a website. Think of some way that and re really we have four ways that you can reach out for a link. You offer to put a test you offer to give them a testimonial on the that they can put on their website with a link back to yours. You offer to write them some content like a blog post with link back to yours. You give them some offer some discount on your offering that they can share with their audience if they put it on their website with a link back to you or some kind of reciprocal link like maybe they have a partners' page. Now, you don't want to do that at scale, but at first, you know, doing that a couple dozen times isn't going to get you penalized. So, yeah. So, getting that foundation set up is important. And then I would create like your your most high value pieces of content first, just like those pillar pages, those in-depth guides. put a lot of time into them, make them really good, and then just start building like really niche specific content around those guides. The guides themselves are not going to rank at all at first, but you have all these really niche ones um targeting keywords that have very little competition. And so, you have a little bit of authority from the foundational links, but then you really start getting authority just from showing up in Google's search results, getting clicks, and not having that pogo stick effect. So you said you're creating an ecosystem of content around these these these bigger guides. Is that what you said?

Yeah. So I think you I think you want the bigger guides on there just to establish like topical authority just to make it clear what the website's about. Um and and to have something if someone comes to your homepage, you want something there that now I'm assuming this is like a a resource website, not like a a website selling a product or a service. In that case, I think you should build all your transactional pages first for your products or services. But yeah, if it's just like a industry resource, then having some kind of hub pages that you're that are linked to from your homepage to show to make it really clear to Google what your website's about and what kind of content you're covering.

Do you care about Google versus users when you're starting a site?

I mean, yes. I care about Google and users.

The difference the difference being like turning out some generic long form guide just for the purpose of Google.

Um but like if a user came across it, they just look at it and be like

uh you know.

No. Yes. The way

have you ever had that reaction when you go to like you see a piece of SEO content and you just you go gh uh

Yeah. Yeah. I mean yeah it it depends. So we have we have websites where we put a lot of effort into each piece and we have websites where we put in not a lot of effort. I think either way the first few paragraphs should be for humans. It's like because most humans aren't going to scroll through the whole thing. They just want the answer and and you can put that at the top satisfy them and then the rest of it is less important. Now I mean the better the guide is, the better it's going to perform. So there's like if if you have the time and the resources, sure, make it like the best guide online. Um, but yeah, depending on how much competition there is and how many resources you have, maybe you phone it in after that and just and just write for the AI, which means just like really high fact density, lots of tables, bullet points, quotes, facts.

I like that. I I like the uh the high fact density. Um, I like the the high fax fact density when it comes especially, excuse me, especially for uh bottom of funnel pages for sales pages like um yeah like creating comparison tables for example

and you can completely control the narrative and you control the positioning like it's uh I'm a fan of that.

I want to end with this question. What are your biggest SEO failures and your biggest SEO geo successes?

I think yeah, my my biggest the first thing that comes to mind with SEO failure is assuming that Google was going to get smart. I I and this was maybe like seven, eight years ago. I just I felt like Google had everything they needed to to do what they aim to do, which was to truly understand content and rank the highest quality content. And so I fell into this uh like eat camp maybe before eat was the term um of just like we're just going to we're just going to create really high quality content. We're not going to do things that like trick Google because all that's going to backfire. Um, and I I stopped building service area pages because I'm like, Google doesn't want to see one page per neighborhood. That doesn't make sense. We should just have one page. Um, I really felt like that was coming. I really felt like Google was going to get smart. And then a few years later, I saw service area pages were just doing better than ever. And since then, I've I've decided to not listen to any Google PR and to just do whatever is working right now. And and I also kind of understand why Google doesn't understand quality. And I I I think it's primarily because it would just be way too expensive to actually analyze all that content. Um, so yeah, that that was that was a little detour I took in my SEO career that that I wish I hadn't. Geo wins. Yeah, I'm trying to think of I I'm thinking of like categories there. There's a lot of right now there are pages online that are being cited like crazy by AI like single pages that completely control the narrative around an industry and the the owners of those pages don't even realize it. And so we've had a few times where there's like this one page where 40% of the time someone asks about this product or service, it's looking at this page and it's just saying whatever that page says. and we'll reach out and say like, "Hey, could you recommend us here and there?" And they're like, "Yeah, it's $150." We're like, "Cool, $150." They have no idea what they're sitting on. Like, that's that's probably like a $10,000 mention right there. And they have no idea. Um, so yeah, we stumble upon upon those a couple times a month.

That's awesome. A couple times a month. That I love that. That's extremely directly actionable. That is great.

Yeah. Yeah. We We don't get it from most clients, but it's Yeah. It's It's really funny what happens and and I'm sure like we'll be in a new world in a few years. People will be a lot more privy to this, but like right now, yeah, it just feels like the early days of SEO again. There's it's it's easy in a lot of ways.

What What are these like uh underpriced geo assets look like?

They're they're just it's just some blog post. It's it's sometimes it's hard to know why the AI loves one specific uh post over another one. I mean, they're usually on authoritative websites, websites that have some clout in the industry, but it's it's just they wrote some listical or they wrote some guide three years ago and it probably never got that much traffic and and they don't realize that it's it's getting picked up by the by the AI.

Can you talk again specifically about how you're identifying these underpriced assets?

Yes, we we're using Peak. So, so we put in about 50 prompts per client. We run each of those prompts once per day in chatbt, Google's AI mode, and Google's AI overview. It pulls back all the pages that were cited from each of those. And then after we've collected about a month of data, we can look through and see which pages are being cited most often.

And and so you said that these are often on authoritative sites. So is it hard to get in touch? Who do you get in touch with to to change the content to recommend your brand? Yeah, it's it's not like it's not like on like CNN.com or like like what you would consider authoritative um generally, but authoritative within the industry. So it it usually is like some industry blog um that has been selling back links or has has been doing affiliates or ads and um and so yeah, they they usually have like a publishing channel. They have some way of reaching out and asking.

They just have a contact page.

Yeah. Now now it also happens a lot of times where we get no response. I I would say like our our overall response rate is probably about 5%. Um, and so that's why sometimes we we stumble upon someone that's willing to sell it and sometimes we don't. But yeah, we uh have you have you have you heard of Noble this uh

Okay. Yeah, I I I talked with that founder and and found what he was doing really interesting. I actually have a few clients that are trying out their service now. He's yeah they're automating the outreach process that we do manually. He he says they get about a 2% response rate. We get about a 5% but

what does your what does your message look like?

Hey, we really liked this article we saw on your website. Um what would it take to get this brand recommended in that article? We can pay or do a reciprocal recommendation. Just keep it simple. We just have

And you're you're just sending this to the contact page or do you send this to the author as well?

Yeah. So, we'll look for an author if we can find one. If we can't find an author, we're usually looking for like a sales and marketing email because that's if if they're willing to do this, it's they probably have a system set up for it already. And and then if we can't find that, we'll go through the contact page. And then most of the time, do they do you provide the direct language that you want used and they just accept it as is?

Yeah, If they're in the habit of doing this, they it's just treated kind of like as a service like, "Cool, where would you want the link? What do you want it to say?" And and then often we're maybe we're adding a reciprocal mission on the client site and they're giving us a language or we're creating a brand new article so that we can add it and then we we'll we'll give them some control over what that looks like. But obviously we have to protect the client's brand.

Yeah, I had I actually have one more question. Um cuz you were talking about training versus query fanouts. And from your experience like how much is it possible to affect the training data? How much you even care about that? Um cuz like most of the searches are using query fanouts. like how how do you think about that distinction between I guess kind of like doing geo with training data versus what the LLM is going to pull with a query fan out when it's doing web searches.

Yeah, that that's a big looming question on my mind. I I think like as an industry we'll figure that out over the next year or two. When I started thinking about geo that was when I started thinking about geo search wasn't very prevalent like you know catchyt search was a separate product and so the whole question was how do you impact the training data where we're sitting right now though training data seems a lot harder to impact a lot harder to influence and a lot less important than the searches themselves uh I think the the one the biggest advantage you get from showing up in the training data is what we're now calling primary bias like the if if the large language model already knows about your brand from training data, it doesn't have to find it in the search. It'll just automatically search for your brand when people ask about that product or service. To do that, you have to be a pretty big brand. Um or or rather you just have a have a lot of content in the training data. Or it might have an opinion on your brand and then it might lean one way or the other. But at the end of the day, it's the search and what it finds in the search is making up like 90% of the response. And it's way easier to influence that part.

I agree, TJ. Thank you so much for coming on the show and for listening to the show. Thank you for both.

Yeah, Thank you for for doing this. You've been a big inspiration to me and it's been a blast talking to you.

Been a lot of fun. Where should everybody uh find you and watch your stuff and listen to you and read your posts?

Yeah, So, I I post everywhere, but uh my biggest following is on Tik Tok. I think I'm TJ Robertson52 on there, and my website's tjrobertson.com.

And you I think you have a YouTube as well.

Yeah, I think I'm TJ Robertson Digital on YouTube

because David Quaid came on your show. I don't know if you have a show.

Oh, yeah. I I I post sporadically on there. I mostly just been doing shorts, but yeah, I did post the conversation with David Quaid on there.

Yeah, Yeah, that's cool. Thank you again. This was uh this was a lot of fun. This was good. And I I really want to do um whenever you whenever you can. I'd love to do another episode where you share your screen, show cloud co-work, or if you want to share your screen and show like any other workflows that you have. Those are so much fun to do.

Yeah, that'd be awesome. Yeah, let's do it.

Thank you, TJ. Uh, for everybody watching and listening, if you um if you watch this on YouTube, well, this episode, yeah, like we said, 963 963 days in a row doing this show. No days missed. And yeah, you watched this on YouTube, thank you so much for watching. If you listened on Spotify or Apple Podcasts, thank you so much for listening. And I will talk to you again tomorrow. Bino.