Jabez Reuben: GEO, LLM Optimization & Listicles!

Jabez Reuben, how are you? I'm doing good, sir. I'm doing good. Thanks for having me.

What else is really cool stuff when it comes to LLM optimization, listicles, what sort of stuff you're working on in a minute? A lot of testings, I would say, and I'm enjoying it. So we are, I think we've completely moved towards just LLM service, LLM rankings for our clients. And we've been just doing a lot of tests earlier in the year, we were just publishing a lot of listicles, but soon LLMs got up to that.

They started filtering out straight up biased listicles. So we've started figuring out different article formats. Even listicles are still working, but you have to upgrade the way, update the way you publish listicles. So basically just trying different methods, different strategies, and also working on our LLM mastery events as well.

One common trend that I've seen this year, particularly across a lot of SEOs who were building their own niche websites, was that they've kind of switched and they're not creating their own websites anymore. They're switching over to publishing content with Parasite SEO, guest blogs or whatever on different websites. Have you seen the same sort of thing? I would do both.

Now, again, with LLM rankings, we've seen a lot of our clients, especially in a lot of competitive niches, clients getting outranked by fresh EMDs. Even small, not non-EMDs or partial EMDs, just small dedicated site, but they outrank for a big query. And that's what we're doing for our clients as well. Building satellites, EMDs, partial EMDs, and just satellite sites along with Parasite.

They're all working. When you say satellite sites, what does that involve? Satellite sites just mean it's non-EMD. It's just another regular small website built around my client's target niche, or let's say a shoulder niche, but it's not an EMD, but pretty much a niche site.

This is something I've been experimenting with recently. Because we found that creating listicles, publishing them on one website with AI works really well. And so we were like, okay, why don't we just scale that out and build like five or 10 websites in the same niche? Is that something you've tried as well?

So there's a very good case study, Corey published on EMDs and how he switches EMDs. Half of the stuff Corey shares, it's hard for me to digest as well. But I first, I did an interview with him in May and he gave this idea of just don't publish EMDs in the same category, just copy paste. You need variations.

And that's what we started doing. We just started doing EMDs or satellite sites, niche sites in the same niche, but with definitely a variation. So each site is targeting a different sub-sub category. And that's how we are trying to rank and see results.

That makes sense. So for example, let's say I'm in the AI niche with my AI communities. Okay. You might have one website for Hermes agent, then another website for Claude code, and you would just focus on different sub-niches.

So they're totally different. Yes. Yes. Makes total sense.

And when you are like, describe what you want, because this is something else, and I'm thinking about how to title this video. That's why I'm thinking about it. But like, would you call this like AEO or NLM optimization or GEO, or do you think it's just all the same bracket? I did a post a few weeks ago and just for the fun of it, because all these terms are popping up, right?

AEO, GEO, then I just randomly posted. So all my AEO, GEO and NEO experts, let me know what you think. What is this new category, NEO? I want to get into that.

I'm going to rank for best NEO expert. I don't know, man. Whatever you call it, whatever works, whatever makes people understand AEO, GEO, SEO, I call it like ranking expert. That makes more sense for me.

Yeah, for sure. I think it's just like still SEO, isn't it? It's just like adapting to what's changing now. What about when it comes to your actual approach?

So for example, you've mentioned putting out a website for a client and then building out, for example, the sub-niche websites for clients, targeting listicles as well. What else goes into the strategy if someone wanted to implement it? Okay. So the first thing we do is we take our client's target keywords.

Then we have our agent built out that scrapes all the backend network processing of various LLMs. So I can show you a screenshot. Can we do a screenshot? I can show you.

So let me share my screen. I'm in middle of all planning. So I'll show you what we gather. So we scrape data.

So we'll run the query through chat GPT, you know, all models and it will scrape all fan out queries and the reasoning framing. And, you know, now apart from fan out queries, there's also something called the verbatim where it's processing. I don't know where it's gone. Oh yeah, the framing.

So it will, you know, it reasons while it's processing the data, it's reasoning. And this can be seen in the right right-hand side panel. And sometimes you don't see it, but our agent just scrapes everything. And then we extract which all brands showed up in which models, what all categories showed up and who are the winners in those categories.

And we do the same across all models. So Google AI mode, cloud, Gemini, everything. So this gives us a lot of data to, you know, start planning. And then, then we see, okay, these are the brands with whom we need to make comparison articles.

If a brand is appearing a lot across answers, then we want to avoid including them in top positions in the listicles. But we do want to create a comparison articles with those top appearing brands. And then even with the sources that come up, you will see each source comes in chat GPT comes with a snippet as well. Like there's a small snippet that, so we take ideas from there also that this is how we will structure because that's the messaging chat GPT is initially pulling from that article.

So all these things come in handy into planning what kind of listicles we need to do, what kind of comparison articles, user guides and review articles. And then we plan out, okay, this will be the plan for the month. And also depends on the client's niche and budget, because if the client is in a tough niche and has a bigger budget, then we can go into building dedicated websites for them as well. Yeah, something I was looking at, I mean, that was a crazy spreadsheet, but I like the system you got there.

Do you find like there's actually different answers depending on the model? So for example, I can see you had GPT 5.6 sole in there and that was compared to a different version of GPT 5.6? It does, it does. So sometimes, how do I give an example here?

So I've just started comparing 5.6 sole and 5.5 instant because I think in the free version, it still uses 5.5 instant. So in 5.5 instant, there's less reasoning. So it's just straight up web search more, at least in the niches that we are working. And in that case, the fresh articles, the listicles that you're publishing, they get to influence the results faster.

Whereas in 5.6 sole instant, especially the thinking model, we see that it goes really hard into making sure the answer is backed by higher or bigger publications. So that kind of difference we are seeing, but there's definitely a difference even within ChatGPT's different models. Yeah. And you also mentioned about Claude, for example, and you're the second person I've heard say this, where if you're trying to rank inside Claude, you can't.

Or if your article comes across as biased in any way, it just kind of discounts you. Yeah. Yeah. 100%.

So this has been happening, I think, start of June, where we started noticing that listicles were being flagged as... they were straight up being flagged that, okay, there's one more brand, but it's appearing in a lot of listicles that seems to be sponsored or affiliate. So we will mention it, but you need to take it with caution. That kind of full warning signs or notices that it flags.

But then there's a way around it, right? So the first thing we started doing is don't put a target brand as the number one winner. Put it on number two, number three, number four. And then we rotate, I think, at least for the next few months, this will be our strategy in listicles.

We don't want to put our brand or a client's brand as best overall number one. It will be in a few articles, but then target two or three sub niches or subcategories that will be most helpful for the client. So let's say, if I'm working for an SEO agency in London, right? And they are offering...

they work with SaaS, they work with Ecom, and they work with local businesses as well. So we'll ask the client, okay, what's your priority order? Which category is most helpful for you? So if they say SaaS first, then local SEO, and then Ecom, then that's how we rotate them as best for SaaS.

So five best SEO agencies in London, this one is best for local, this one is best for SaaS, this one is best to work on enterprise. And we try our best that categories that are helpful for our client or the category that our client wants to target, we don't give those votes to anyone else in the whole campaign as much as possible. That's smart. One thing that just came to mind then as well is, for example, you could have a sister brand for that.

And it depends on what the client's comfortable with, I guess. If it's a big enterprise, they're never going to do this. But for example, if it was you trying to rank, then you could have multiple different sister brands of your same agency and then bank top three no matter what, and you just rotate who ranks number one. You can do that.

I would say still use... because there was also a big debate, people saying that, hey, why don't we just make fake brands so that if we are giving votes, we are not giving to our competitors. But then, you know, LLMs and Google, they're all getting smarter, and they can track fake entities. And the point is to put your brand in a known bucket.

You want to put your brand again... stock player so that LLMs and Google can identify you as a bigger player, not as someone who is just popping up with random names. So I would use sister brands. Yes, that's a great idea.

But I will make sure that I've established its entity and its consensus as well before I put it in that bucket. Yeah, no sense for sure. Like if I guess if they fact check the article, and it's like Bob's SEO agency that doesn't actually exist, then no matter all the hard work you've done becomes undone, doesn't it? Yeah, yeah.

I'm saying in some niches that are less competitive, you can still get by. But the more competitive you get, then it's definitely going to backfire. Yeah, I can see that for sure. Especially, I mean, one thing I was thinking as well was like, if you've got your strategy versus someone's, you know, another SEO agency strategy, I don't see how they can compete with this.

So you're reverse engineering the actual data from the LLMs, and then you're fine tuning it month by month based on what works and what doesn't, right? Yeah, that's what I'm seeing. And especially, you know, in models that are working in the free account, like chatGPT's instant AI mode and Gemini flash that are these models that are pulling a lot of, you know, live retrieval data, new or fresher content, you know, kind of overpowers a lot of time. So the ones who are reverse engineering and publishing new content will keep moving up.

But at the same time, you know, this whole thing of getting into the training data and building your grounding effect that comes into play. So if you are consistently still diversifying and building different kinds of articles, because there will be a lot of articles that probably your competitors cannot see unless they are, you know, tracking your brand mentions. So because a lot of art, we built off page topical authority, and we build a lot of different kinds of articles, not just listicles to build consensus and to get into the training data. So if you're doing that, then you may, you know, keep still keep outranking your competitors, and they won't be able to catch on to you unless they are doing your brand monitoring, because then they will know whenever you're publishing a new content and putting your brand there.

That's interesting. So you're doing listicles as well. And then you're using that strategy you talked about before, but then you're also doing other articles to sort of back it up and to reinforce it. And what is that like query fan out content or?

Not just that. So the simple goal is what we've been doing all this while with topical authority on your own site, you basically now do off page topical authority, you just build a detailed topical authority for your brand and publish it on third party sites on Reddit on different third party sites on and if you don't have the budget, then there are so many other, you know, free platforms just published there and build your off page topical authority, then you then get granular for you being aggressive, you publish listicles, content articles, buyers guide, review articles to you know, force into LLM and AIO view results. Yeah, that makes sense. Have you tried testing out trying to rank with YouTube videos like listicles inside YouTube videos?

Not yet, but I've seen plenty of results where YouTube gets into the citations very easily. Yeah, it also depends certain platforms perform really well based on the niche. So what I've seen, if it's B2B or business related, LinkedIn will perform well. If it is health, beauty, those kind of niches, Instagram will perform well and perform by that I mean, besides your regular websites.

So the different platforms, Instagram will be there. And then YouTube, I think is in a lot of segments, a lot of niches. What are you using for actually tracking the rankings? I've been using, we've been using radar kit from I mean, it's someone I know, like a friend of mine, he's built that tool.

And that's the only tool I have tool wise that I've used because he's good with implementing our suggestions. But honestly, I prefer manually, you know, testing our clients ranking. I mean, that tool is good, but I also keep testing manually. So using VPN, change the location in cognitive mode and just track random locations, random variations for our clients and send them screenshots from there.

But the tracking tool is helpful because in LLMs, there's no number one ranking, right? So you have to take out the average. So yeah, that's what we think is what seems to change so much, even depending on the user's history and what they're working on. So yeah, and also location, you know, so same query will have different answers in search from different locations.

So radar kit does that, it gives you location based answer, different location based answer. And I'm sure like all the tracking tools, there's so many tracking tools, right? So many, every other week, there's a tracking tool being launched. Find something which is easy to use.

I think there's so much investment in that area right now. It's like everyone's trying to create their own tools to get invested in and sell quickly. What about with the listicles that you create, do you build backlinks to them too? So this is like what we are doing.

First of all, we've almost reduced adding our client's actual link in those listicles. Unless we are getting, we are publishing on a really big site, only then we will add a link or if the client has never published a dedicated review article. Once we have a dedicated review article, then what we'll do is we will use that published link. So let's say Julian Gold, here's a number one review of juliangoldy.com.

Once that article is published, then other listicles, other comparison articles that we publish, we add a link to that article. And then, you know, kind of daisy chain and then kind of just give links to previously published articles. And also apart from links, there's another concept of citing, backing up your claims. So something that I published in my first review or second review, I like we back up those claims in future articles.

So those are two things that we implement when it comes to giving links or you're just adding more power to previously published articles. Yeah, it makes sense. How about like stepping away from that because that was super actionable when it comes to, you know, how to organize them, how to track them and everything else. When it comes to the SEO industry, how have you seen that change this year?

Because it feels like it's accelerating in ways I've never seen before. Yeah, yeah, it's something that you were talking earlier, like there's in terms of from client response, I think there is still a lot of formal element, which we are enjoying. Everyone wants to be in NLM, great, we come and buy our packages, irrespective of whether it brings business or not. But I mean, people want to be present in NLM results.

That's that part. But in terms of SEO industry, I think it's evolving, right? I had this interview with Kyle a year or two ago, where he said that, man, year after year, we've deal with updates, and we kept evolving. And I think that's the same stage right now.

I think that's what it will continue to be for more years to come in the future as well. There'll be updates, even listing the NLM SEO or AISC, what we've been doing that has changed in the last six months, the way we were doing. We keep evolving every month. And I think from that point of view, it will continue to remain the same.

Yeah. Yeah. I mean, one day there might be like a listicle update or something like that. I mean, it works now.

Yeah, exactly. I mean, we are already extensing that. So we already moved from, you know, I said, we were earlier just heavy listicles. Now we already started moving towards, okay, now the next kind of article type that's working better.

And I'm sure after six months, there will be another kind of article style that will be working. Links, I didn't expect to, you know, degrade this much, but it has found a new way to, you know, still work or find its way into SEO. So it keeps evolving. As long as you keep yourself updated with the latest trends, I think SEOs will remain in business.

Yeah, for sure. And you've got a lot of events as well right now, right? Like a lot of masterminds and that sort of thing. Yeah.

Yeah. So yeah, we are doing, we do a lot of events. We started doing events also for brand building point of view. But then recently I started doing a lot of masterminds and podcasts because I think that especially doing, getting on podcast and interviewing people right now, I gather a lot of new insights just by interviewing people.

I did one today with Charles and then just got blown away with some of, and we talked very, very niche. We didn't even cover all LMs. We just talked about AI more than AIO. And just in that, I was able to, you know, track some really solid points.

So I think doing events and doing these podcasts and masterminds, it has worked not just for our branding, but also just upgrading our skillset as well. Yeah. I feel like just because I started this new channel, you know, about a couple of weeks ago, and it feels like I've learned more in the last two weeks about SEO just from stepping back from my own work and seeing what everyone else is doing. Just by doing that, you learn so much.

It's amazing. Just for that alone, it's well worth it. And then also, I mean, if anyone's watching this and they're an SEO agency or freelancer trying to get clients, content creation is I think one of the highest leveraged things you can ever do. For sure.

For sure. And when you're interviewing someone, you are like fully focused right into that conversation. So you're fully absorbing that as well. So you learn more.

Yeah, for sure. You're like, you're paying attention. You're just trying to focus only on that. That's the other thing I find is like, if you're listening to a podcast in the background, sometimes you get distracted or sometimes you go off and do other things.

So you don't fully give it all your attention. Yeah, true. Same here. Have you considered, because it's interesting to see, like you said last year that you're more of a like LinkedIn agency, but it sounds like.

you're switching more towards LLM optimization using the methods we talked about before. Is that something that's just happened naturally or is it something that you planned? I would say accidentally we jumped into it. I mean, we were just doing the same old drill, publishing links.

And I was also trying to build a platform with a friend and which was like one of the main features was how do we influence LLMs and build consensus. But then when we accidentally did a test and saw results with LLM rankings, it got me excited and for the first time I just started doing testings and experiments on my own. And because of that, and because demand also increased from our clients that, hey, we want to rank across LLMs, we do a lot of white label for agencies. So agency clients started coming back to us that their clients want to rank more in LLMs.

So yeah, that's why we just now moved, at least for now, our current focus is all around LLM visibility, LLM rankings. And it's also fun. It's still new, right? It's still not fully saturated.

So it's fun to be in somewhat like, you know, in the front line of doing experiments and sharing new case studies, which probably others haven't done yet. You are working on a SaaS tool as well, right? For link tracking. Are you still working on that?

Yes. So I was just having my chat with my developer because I mean, this whole, we definitely want to add some, you know, like everyone else, we do want to add that AI powered or LLM tracker or something. But that segment, which was so tiny, was massively saturated already last year. So after my developer, if you can think of a really out of the box idea that's helpful and something that we as a company can use, then we can, you know, add that feature.

Other than that, it's just going to be the same old rail of our tracking tool, which I don't want to get into. It's just not worth it. Yeah. I mean, for the amount of work it involves as well, I always feel like if you're fully focused on one thing as well, you just get 10 times better results.

100. 100. And I think that's what I, that's one of the reasons again, I've enjoyed this year is because for the past six months, just the whole focus, everything has been around LLM rankings. And it's, I think probably the most exciting time of my career right now.

And obviously, like you're pretty well connected. I mean, you do a lot of SEO podcasts. I know you mentioned, for example, like Corey with Topical Authority and Charles Float before as well. Is there anyone else that you're really paying attention to in terms of like, you know, the content they're creating or the experiments they're running in SEO?

You know what? One of the first person who really changed my perspective of how we were taking LLM rankings was James Dooley. He really shared a lot of information, a lot of information and testings and data, and then also did a lot of tests with us. So James was really good in sharing information.

I'm still on the lookout for people who are doing a lot of testings. I know like James and you know, the bunch there, like James, Kastra, Corey, these people, they do a lot of testings. Then there's this other group that shares a lot of Google News, right? I love it.

I don't know exactly what you mean there, but for sure, like, I mean, James was on the podcast last week and the amount of knowledge he shares is unbelievable and he's got so much passion for it, which I love. And that was one of the most popular podcasts as well, I think everyone else just loves to listen to it. Absolutely. And he's super helpful, which I really respect.

And there are a few guys I've started following because they are doing like reverse engineering, like how we are backtracking data and trying to extract. So there's a guy called Suganthan, he publishes more on LinkedIn, and he's doing basically the same tracking. He's trying to extract a lot of data from the network logs. So I like what he's posting.

Steve Dorth also publishes some good content that I like following. But I am also on the lookout for, you know, people who are doing more on testings, like we did this test, this is what we implemented. And this is how it works. Joe, I was asking someone for more podcast guests the other day for SEO.

And the first response I got was that, you know, people are trying to be really careful about what they show in their stuff, because obviously they don't want to be any sort of target in the future, I guess, right? Yeah. But then also, like, if you become the target, you know that you are doing, you're on to something. Sure.

For sure. Gives a good publicity. We do have a dedicated testing group where people are testing different things. So from there, we are able to, you know, just share some ideas.

And I'm trying to just add more people who are, again, testing a lot of stuff. Michael Sisky is another guy, co-founder of Surfer SEO. Oh, yeah. He also, yeah, he's also doing a lot of data analysis and tracking and testings as well.

So he can be a really good guest as well. How do you actually do the data extraction? How does that work? So you can just create an agent.

I'm sure you've built so many agents. All you have to do is tell the agent that you give the agent Chrome access, right? So tell it to get into the log, the network, the backends, and extract all the information while you are processing this query. So tell it that, okay, run this query.

And while this query is running from the backend, whatever that you are seeing, all the sources you're seeing, all the fanouts you're seeing, all the verbatim that you're seeing, snippets, you keep extracting it. And then you keep, and then over time, if you are building it in a way where it's improving, it's making its own brain, it will start analyzing and getting better at extracting the information. When the first time I did this, I just couldn't handle this. I was like, man, what is this?

How do I interpret this? And then over time, it started interpreting. And mind you, the agents will also start adding their own, you know, hypothesis and analysis. So you have to keep pushing it back to, hey, don't give me something that you cannot back with actual screenshots or actual, you saw something in the background for real.

And that's what we do. And then whenever someone publishes a case study or someone else finds, you know, more information from the network logs, or this is what we found, this is, then we feed that data into this and say, hey, why didn't you track this? And how come this guy is tracking this information? So just keep improving that agent's brain.

For the brain, do you use like Obsidian or something like that to store the data? Just plain, everything is stored in what's called co-work. Very basic. I'm not very great with agents.

It's a very basic cloud agent that I made. Would recommend if you, if you want to store, for example, all the stuff that is learned like in the logs and everything like that, Obsidian is really, really good for that. That's what I've heard recently. It's very good.

Like I'll share my screen quickly and I'll show you what I mean. So this is what Obsidian looks like. And basically all of these notes, I mean, you could do this just for SEO, just for your tests, but all of these notes are literally just markdown files that we store, like everything that is, it's done and all the most important parts of our database. So for example, like every, all of that web that you saw, these are just markdown files together.

And then it will categorize, for example, like all the tools we use or the people we work with or the people on our team, you know, like we have little things like the tools that we use day to day that connects it all together. So then for example, you could have like each one of those logs you could have as a markdown file. And then also like each thing that it learns day by day, you could have, so it could like update its learnings or like its testing or whatever. And it can do that all for you.

Like none of that I created myself. Brilliant. Have you created some sort of a company brain as well? Yeah, that was it.

That was the, so for example, if I go into Hermes agent, cause it's connected to that, it will read through. Let me, I'll show you an example of what we did just earlier today. It's recording of course on this. All right.

So we actually, we visualized number one, the actual memory. So it looks super nice and you can see it the way an agent would see it, which is like this galaxy. And each one of these stars is a memory and they're all interlinked. And then the stars that burn the brightest are the ones that have worked on the most recently.

But then also, for example, if I go into Hermes or whatever tool I use, and I'm like, okay, based on what you know about me, based on my obsidian memory, give me a list of keyword ideas that were made, you know, that are personalized to me, have all our keyword data. So we've got that over here and then it's like, you know, here's what you're working on. Here's the sweet spot. So it knows we're working on the agent operating system and then it gives me all relevant keywords based on that.

And like a super long list based on our obsidian, which is super useful. Yeah. I definitely need to get into using obsidian because I heard recently in another group chat in Chiang Mai, and everyone was recommending obsidian. Yeah.

Plus it's just fun to look at the memory galaxy. You can mix so many just social media posts showing up that is what I'm working on. Everything on my Instagram is like, guys, look at this. By the way, if you want this.

Makes you like from a link builder to a full on like scientist. Yeah, that's what it is. It does feel like you're a crazy scientist at times, you know, experiment with some of the fun. Yeah, it's a weird journey.

And that was totally accidental, like the journey that I've personally made from going from like link building, which we still sell, to like actually AI community became the biggest thing that we created. I love it. Yeah. You're still enjoying that?

You still get excited about it? Yeah, I think it's easy to get excited about it. I was like, probably the most important technology has ever come out for us and I get to talk about it and experiment with it every day. That's fun.

Yeah. Yeah. And you're still on school? Yeah, we actually, they actually, they changed.

So the one we just won the school games again, like the last quarter, which was great. And then pretty insane. And then they changed the leaderboard now so that you can see who's in the top 10, literally get a ranking. for everyone in the world, regardless of category or industry.

So we're, we only just started in February last year. And when I started, I felt like we were too late, but we're ranked number nine in the whole world, like out of 186. Of all categories, of all categories. Wow, that's incredible.

186,000 new owners on Skrill, number nine. That's incredible, man. Well done, well done for that. It's scary.

And we get, I actually looked at like, who gets the most traffic, cause it shows you who's getting the most. Out of the top 10, we're like, again, we're the second highest in terms of traffic, without running ads or anything, which is pretty amazing. Crazy. And that's incredible.

Yeah. Yeah, I was pretty happy with that. But yeah, that's, I mean, again, cause it's a game, it's kind of like, I never had that with the agency, but if I knew, if, for example, at a league table and it was like, okay, you got, you know, Jabez, and then you got Alejandro, and then you got, for example, Authority Builders, and you could see NRR, compared to everyone else's, that would be so much better. Yeah.

And that's what we were talking, right? You keep doing that one thing, or like, as you were saying, you keep doing your SEO stuff, and then you just find that one thing, or that one service, or that one category that really starts exciting you, and you just double down on that. Yeah, you just go all in. Yeah.

You go all in that. And that's where, you know, once you, you know, conquer that mountain, then you will start like, you know, flourishing in so many other ways as well. Yeah, that's it. That's it.

You don't feel confused. You kind of, you just figure out what works, and you just rinse it. You're like, let's go as hard as we can. 100.

Yeah. 100. 100, man. All right, man.

Well, it was fantastic having you on. Really enjoyed this. Thank you for having me, Julian. It was great chatting to you, and I think we should do a chat with you, where you can do a podcast showing how we can probably build a brain and all.

Yeah, we'd love that. Yeah, we should do that, for sure. Do that. Cool.

Thanks, man. Thanks again for having me. Thank you. Julian, we can link to you in the description.

Link out to you. Yeah, I'll share the link to my upcoming conference event, and I'm also launching the LLM course, LLM Mastery Course, so I'll share a link to that. All right, can't wait to see the course as well. Cheers.

Thanks, man. Thanks, cheers.