141 - What AI Thinks About You - Dixon Jones Reveals What ChatGPT Really Knows About Your Brand

Speaker A: SEO traditionally was all about trying to get the customer to your website so that you've got them to a landing page that you can then, you know, talk them through the journey and convert them. But now people are jumping past that and they're getting their data from all those links that came in, basically all those pages that were linking to your website. Link building has not gone, uh, away, but the context in which somebody is linking through to your website or talking about your brand doesn't have to be a link anymore. Uh, that's enough for the LLM to read that and decide. Right. That's, uh, a good point about the brand. So now for me, uh, we're seeing the search engine optimization is, uh, in Wei K's terms, in what AI knows about you. Welcome to the SEO Leverage podcast, where we talk about search, marketing and conversions.

Speaker B: Welcome back to SEO Leverage. My, uh, name is Gerd Melak, and today I'm very happy to have one of my favorite SEOs out there back on the show, Dixon Jones. Welcome back.

Speaker A: Well, yeah, thanks for having me back and hopefully, hopefully I can shed a little bit of a light on a increasingly murky future.

Speaker B: I'm sure you will. So everybody who doesn't know you probably has been living in a stone, first of all. Below a stone, first of all. But definitely listen to episode 25 when we are talking about Dixon's background, uh, Majestic, uh, as a tool, we talked about a. Remember, and entities first and foremost and your tool inlinks. Today, the reason why I reached out to you and invited you back to the show is a different one. Uh, I want to give a little bit of context. When we saw that we really wanted to know with ChatGPT where things were ranking and where our clients came up, I had a conversation with our developer and said, look, why don't you build like a tool that just fires prompts against the API and we see who comes up where, how often, in which position, etc. Because it wasn't really a tool out there I could really rely on. I didn't like what was out there. Like, were a couple of options. I wasn't convinced by their pricing or by how they did it. M. And then I found yk.

Speaker A: Yeah. And I said to my developers, YK or YK or. Yeah, it's hard to pronounce, hard to spell. I don't even have the dot com.

Speaker B: I've come to realize that good tools in SEO need to have a hard to pronounce name. That's.

Speaker A: It would appear that way. Yeah. Another Example, uh, it's funny that, I mean, because. Because when we were developing the tool we bought, um, we own AI scanner.net you know, and we were going to use that and. Yeah, and the look on your face, you know, for anybody that's on the uh, on the non visual side of this, you know, the look on Gert's face. Yeah, that's it all really. Uh, and. But the reason we ended up with Waycay is, is we said we just want to say what it does on the tin and we want to say, you know, it's what AI knows about you. And someone said, well, let's, let's call it that then. So, uh, I've come to, I've come

Speaker B: to get used to the name and start almost to like it.

Speaker A: Yeah, cool. Honestly. Yeah, that was a surprise to me that people actually like the name. But that was.

Speaker B: I think it definitely stands out. And I always say it in my head, what I know, I know about you. I always say it in my head when I type in the name. Yeah, it definitely sticks and it definitely brings the message across.

Speaker A: Oh yeah, uh, it's hard to pronounce. It's like, it's like, you know, is it Semrush M or Semrush or is it HRS or uh, Aha. Refs or whatever. Spark Toro. You know, Rand went down that route as well.

Speaker B: Just the choice. It's definitely a good start for a successful tool.

Speaker A: It seems that way. There is a science behind that actually, um, which relates to the whole LLM stuff. What we've been been seeing when we were built Waycay was uh, as we were testing on lots of different websites, websites that had words that could be misinterpreted by LLMs were. I got a really good example just um, last week. Uh, there's a big agency in Los Angeles called Hennessy Digital. And I've known Jason Hennessey for about 20 years and, and um, and his agency is doing really well. And I sat down with him in Phoenix and showed him Way K and showed his site and his site is hennessy.com. now that's spelled with a double S E Y. So it's different to Hennessy Cognac. But, um, uh, the LLMs got really confused. They said that it was about cognac. It said it was about high performance tuning company. Um, and um, it was okay when you said what do you know about Hennessy? In the context of search engine optimization, then it got it right. But if you said what do you know about the brand Hennessy, in relation to Hennessy.com, his spelling of his website, the LLMs went right off, got the wrong founder, the wrong date, the wrong product, uh, absolutely everything. And um, this is a site that does well in SEO. And uh, so it just goes to show how important context is to LLMs, um, that you can't just go and say, you know, how do I rank for search engine optimization in an LLM? When you try and do it, it doesn't work very well, um, because the LLM comes out with different text each time. And that's really why we built Waycade, because we found that what we can do is say, well, okay, the text comes out the LLMs each time very differently. Uh, but the meaning is the same generally. Uh, and so if you've got an algorithm like inlinks is one which will break that output from the LLMs just as well as any, um, uh, web pages into the underlying topics or entities and concepts, then we can check that against the entities and topics on the client's website or the competitors of the client's websites. Uh, and that's where the magic happens because, you know, uh, I, my guess is, you know, when you, when you're thinking of building the tool, you're building a tool yourself. The, the problem that you get is any kind of uh, flow or systematic, you know, ability to see whether things are getting better or what you got to do with it. Because most, um, people have access to the API calls in uh, the LLMs, but they don't have access to a natural language, um, parsing, uh, routine that can turn it into a knowledge graph. And that's where, ah, Wake A works out to be quite effective, I think.

Speaker B: Yeah, I definitely remember talking about to my developer and said, stop whatever you're doing. We don't need this anymore. It was really based on a necessity. I had more and more clients ask me, can't you tell me us the tracking? Can't you fire a few queries against ChatGPT, tell us where we are. I remember I put together Google Docs where we just open up all the main tools and put in the same prompt and analyze what comes back as an important thing that seems to be associated. But this was obviously a purely manual approach which is not scalable at all. And it doesn't contain now, uh, really the meaning and context and all those things that we really like to look at. Uh, but yeah, definitely. Congrats. This was definitely like one of those moments where say, the industry just got better. Yeah. Finally have a tool we can work with and can do something suitable about it.

Speaker A: Well it's really important for everyone isn't it? I'm not surprised that the clients um, are asking uh, because yes the zero click thought process is very real and there's a few studies have come out recently, um, uh, about the fact that AI overviews uh, are changing the way that uh, people are using search. And I think that there has been an immediately incorrect approach for SEOs to try and deal with it. So the SEO way of dealing with it is to say well we want our links, our links in the AI overviews, we want to be cited with a little and so that people can still click on those links in the AIO views. And that's where I was thinking in the first place. But I think that we've, we've missed the point of zero click search. So I've just got, just behind me I've just got this brand new computer. It's 2000 and something pounds and it's going to run local versions of an LLM, uh, of LLMs and you know going to start doing weird and wonderful things I hope if my, if my skill set is up to the job. Uh, and um, and uh, so it's a very specific machine. You cannot um, just go into you know, PC World or whatever and buy that machine because it won't, won't run LLMs, you know, it'll just you know, crunch under the strain. So I used of course you know, Gemini Deep Research to try and plan actually my next six months work trying to do all this stuff. So what are the things I'm going to need to do to create this machine that's going to be my agent. Uh and um, what do I need to learn? You know I've got to learn Python again. Uh, and uh, you know, and uh, ah and n8n and all sorts of other bits of this and and um, so I did lots and lots of research and I said what kind of computer do I need? They said very specific. So we went, it's given me this list of you know the GPU and the, you know and the, everything that's got to be in there and I did all the research in Gemini. Uh and then I used another LLM to go and search for uh, for, for a machine on Amazon. The best machine that fits that. Uh, the, the spec, um, got the spec, bought the machine uh and I was a little bit nervous. It's an expensive machine you know. Um, and, and is my, is my Pythons are My Python schools going to be up to the job. I don't know, uh, bought the machine and only then did I think maybe I should check the reviews. And ah, I checked the reviews, no reviews because that machine's never been bought. Uh, three thought well I should, you know, what brand is it? It's called Gen6. That's the brand, you know, who's heard of that? You know, so I literally had done all my research using chat, GPT and, or Gemini and uh, bought them, bought £2,000 worth what, two and a half thousand euros worth of uh, machine without any click to the website whatsoever. And that I think is where SEOs need to understand the real fundamental difference between AIO and SEO. If we're going to start messing around with different letters, it's still SEO, I'm still calling it SEO.

Speaker B: I think this is really, I have a little bit of a similar example and similar experience. I was looking at helping our clients with some conversion rate optimization. I did this in the past and wanted to just check what kind of tools are out there and new tools and pretty much typed in one and alternatives into Google. Now actually I need to start from it from the other way. I signed up for a trial for a tool I didn't know and then I just stopped and said hey, what just happened? Because this took me 30 seconds and I went back and said I did a Google search searching for alternatives. Uh, I landed on a Reddit thread, I sign up for the tool all in 30 seconds. Yeah, when we, when we started with the idea of what we call platform leverage where we say hey, let's, let's go to other platforms and make sure your brand is mentioned there. Not only for AI obviously to, to pick you up in a certain context on a regular basis. Close experience with a hit, something has fundamentally changed and we just try to push like in a traditional way and say okay, let's still make this click work and let's still work with a uh, 50% lower click through rate and try to, try to get this across where the entire landscape is changing.

Speaker A: It is, yeah, absolutely is. And I think there's, there's, there's a couple of other things involved in there because if you don't need to, okay, if you're going to a SaaS product, of course you've got to end up on the website but if you're buying a, a car or a, you know, or a holiday or whatever, you don't necessarily um, you know, have to end up on the website to do that. And I Think um, one of the things that we've been seeing when we've been testing and playing with the WAYC product is that some websites or some brands in particular have had, you know, this knee jerk. I'm really worried about letting ChatGPT crawl my website. So one of the things that we saw was we tested Microsoft as a brand and when we ran WAYC over Microsoft and we used I guess Google and um, Google and IBM as competitors, I can't remember what the other competitor was. Uh, and um, what was interesting is that Microsoft, uh.com does not let Gemini crawl it. Um, and I can just see that as being a logical, you know, thing for Microsoft to do in the higher echelons of Microsoft, but without any thought to the consequences of that. Because when you ask um, the LLMs, when Wake RC at the LLMs, what do you know about Microsoft Microsoft.com, uh, in the context of its brand? Absolutely. It comes back with very rich results. You know, it scores in the, you know, in the 90s. You know, everything is understood but all of the citations are anything but Microsoft.com. so what they've done in that process is really lost control of the narrative. Um, by not letting chatgpt, well in this case Gemini crawl their website, you then it's just the rest of the world's opinions that are going to make up what um, what the LLMs say about Microsoft. And uh, we're finding this all the time, you know, this, this um, this disconnect between um, what LLMs know about a business and where they're getting that information from. And it's not always your website anymore. So SEO traditionally was all about trying to get the customer to your website so that you've got them to a landing page that you can then, you know, talk them through the journey and convert them. But now people are jumping past that and they're getting their data from all those links that came in, basically all those pages that were linking to your website. Link building has not gone uh, away. But the context in which somebody is linking through to your website or talking about your brand doesn't have to be a link anymore. Uh, that's enough for the LLM to read that and decide, Right, that's uh, a good point about your, about the brand. So, so now for me, uh, we're seeing the search engine optimization is uh, in the, in way case terms in what AI knows about you. When you come to a topic level, uh, then you start to say, right, what is uh, what do we know about Gert Malik in In, in the context of podcasts, for example, and it'll come out with, uh, with uh, information and it will come out with a bunch of websites. And then you can compare Gert Malik's podcast to, you know, to, to other SEO podcasts, uh, as well, and it'll say, right, you know, this is what we know about Gert Mallet's podcast because of these web pages over here. And that includes this SoundCloud one and this, you know, this Spotify one or whatever. And then these other biz, other podcasts will, will be getting talked about by websites that you haven't yet seen. So now you've got a new set. Uh, the sort of. The new link building is not necessarily about getting links, it's about getting talked about in the places that the LLMs are using for their training data. And that is a shift, really. It's not. And of course you're going to end up with zero clicks. You don't even need a website actually, necessarily. I mean, it helps, obviously.

Speaker B: Absolutely. Look at ChatGPT and the code leak that was there where they tried to close the Shopify sale, for example. Yeah, inside ChatGPT. I mean, this already shows where this is going, right? So it's this AI mode, it's this ChatGPT complete experience. They bring in the YouTube videos, they bring in the images, they bring in the, the full experience.

Speaker A: Yeah.

Speaker B: I was Talking to a B2B business before, for example, and I want to ask you, by the way, about, uh, about one of their challenges. But they are, they are essentially seeing that a lot of, um, a lot of the connections they have between their brand and their individual topics are just really fuzzy, right? Yeah, really fuzzy. They just have like a small bio on trade shows they participate in, etc. For example. And this is my question to you. They have, for example, have a particular challenge where they are doing a lot of different things. And this goes back to your ambiguity with the name as well. They are doing very different things. How does an AI or how does a website really optimize when they're doing things that are completely unrelated under the same brand name?

Speaker A: I think they do themselves some dangerous disservices, but the core SEO principles remain the same. They need to make sure that their, um, different products are connected together in, in logical, um, in a logical way. So if, let's say they, they make, I don't know, tent pegs and, and uh, and car wheels, you know, uh, you know, completely different products and they want to be, you know, they want to be Seen for both. What they don't want to do necessarily is to create, they're okay creating an overlap between the tent peg pages, uh, and the wheel pages, as long as the sentences that connect them make sense in their own right. In other words, the LLMs are reading and breaking everything down into individual tokens as our search engines. Uh, and they can say, you know, don't put a tent peg in your tire, uh, and it can link over to the tire page. That's that, that's fine as long as the sentence makes absolute sense. Where there's a problem, and it may be a problem for a brand like that is that they've put all of those products into a menu and a menu doesn't have context. So you've now got a bunch of disparate, disparate products in a, in a menu link. And that menu link doesn't have context. So now you've just got the word, you know, 10 pegs on the tire page in the edge. And uh, what machine is going to understand why there is a relationship between those? It makes sense in this text when there's a proper sentence around it. And this is where the whole of LLMs are, uh, going is that you need to have a narrative, a story that is clear and repeatable. So, so you can have one for 10 pegs and you can have one for, um, tires. It's not so easy as having one for tires and one for, you know, car parts. You know, that makes much more sense. Uh, but if you go into completely different areas, then, then it does create a bit of a challenge for them and they need to make sure the content is rich and clear, um, and where it can be differentiated. So one of the things we found out is that, um, slightly different to your point is that it might be good on your About Us page now to tell people what you don't do, um, because an LLM just picks up and churns up information and may make assumptions. One of the things that we had problems we had with Inlinks when we were testing the WAYC product is that we found that one of the, um, chatgpts or the Clauds or, or Perplexities said that Inlinks was a rank checking service and it's not. It's an entity SEO tool, but it's not specifically a rank checker. Um, and we were able to track that back to one, um, review. So somebody had done a positive, mostly positive review of Inlinks. Ah, but they'd said it had a rank checking feature and we reached out to them and it's the easiest fix I've ever had actually. You know, you remember how link building was a nightmare. You had to go out and ask for people for links and then they did and it's, you know, it was all a bit messy. Well, telling somebody they've got something factually wrong on the page, of course they wanted to fix that and uh, and, and they fixed it and within the, within the hour. Um, and the LLMs, as you know, you know, have never reported this fault since. Okay, um, maybe we got lucky with that but uh, but you know, easy fix. So fact checking is the, is the first thing. Using the, the fact checking function now in Waycay makes that very easy. But just going and asking the LLMs, what do you know about this product in the context of this um, brand and then just seeing the answers and then picking up the errors is, is really the first step I think for all of us. It's kind of like getting indexed properly versus getting to the top of the search engines. If you don't get indexed properly, you're not going to get to the top of the search engines. Uh, and then job number two is going to be um, well I've got to, I've got in the index and it understands me correctly now when I say what is, what are the best tempeg companies in the world? Why, why is mine not coming up? And then that's the next part of the way K story where what we're trying to do is different I think to others. We're not trying to just get your um, your URL linked cited in the, as a, as a, as a data source. We're trying to get, get you meaningfully put into the, the list if you like. Um, and for us we're thinking, okay, somebody's talking about tent pegs. So somebody's saying what's the best tent peg company? And it might start. The LLM might start with a number of parallel ideas. One, tent pegs are made of metal. What kind of metal do we need to go down and where do we get them? The next one might be tent pegs. What do they use? They use in the camping industry what sort of good camping brands and go down that thing, that route. Uh, another time it might say right, um, tempegs. Well, uh, you already told me in the last sentence that you're in um, in Florida. So we're going to have to kind of go down, you know, localize things in Florida so the LLMs will go in different ways in different times, but they will cover a whole load of different topics along the way. You know, the quality of the steel or the uh, or the length of the tent peg or how heavy it is if you're backpacking and stuff like that, uh, and come out with ideas. Now if you compare your brand with two other brands, um, and say, what do you know about 10 pegs in the context of these brands? You have a look at all. The LLM will now look at those three websites and go, and, well, those three. What it knows about those three websites and all of the strengths, uh, and the features that they're talking about and the, you know, the quality of the 10 pegs or the price of the 10 pegs, etc. And the different brands will talk about different things. Uh, and then we'll see. Be able to see what topics are, uh, talked about when the LLM are, ah, saying what's the best tent peg? Uh, and in that process we can then say, you know, what, uh, your website is talking about, you know, the quality of the steel. I'm being a little bit out on a limb here. The quality of the steel. But, but it's not talking about steel versus, um, aluminum 10 pegs, you know, um, and your competitor over here is so talking about the aluminum 10 pegs or square 10 pegs versus round ones. This is a really important gap in your information source because, because every time you can interrupt the LLM thought process with something that's relative to your brand, the more that the chance that you're going to end up in that final list of give me the top 10 PEG companies.

Speaker B: How would you go about. I mean, YK obviously shows those gaps that exist. Right. For me.

Speaker A: Yeah.

Speaker B: Like, I really like the visual that you put in lately where you just see, okay, what is, what is really, really going on or where the gaps are, which is definitely very intuitive.

Speaker A: Thanks. We got that out of Brighton. We saw because we had it kind of being as you built a thing, it kind of showed it. And then they said, we just want to be able to see that. And it disappeared. So we brought it back. Yeah, I'm glad you like it.

Speaker B: Good.

Speaker A: Thank you for that.

Speaker B: Um, uh, one thing. Uh, imagine now your tool comes up with a gap, right? There's not enough in a stick with, with. With my example, I'm, um, related to podcasts, but I'm not related. I don't know enough to SEO, for example. Right.

Speaker A: Yeah, yeah.

Speaker B: How would a brand go about filling such a gap if there is like a semantic Gap or what you want to call it between the brand and a particular topic. How can they go about it? Do they something, do they do something on their website? Do they go to other platforms? Where do they fill this gap?

Speaker A: Both of the above. So the, the way that we recommend it is, is in the, the action plan we tend to recommend. Make sure that your website covers this area and if you are covering it already then make sure it's more prominent because the LLM didn't pick it up, you know, um, because that's sometimes what happens. But you should be able to fix most of this stuff by improving by, by making uh, your content more clear. Um, in the same. So it's, it's really, that's where it really is the same as ah, old style SEO. Um what is different is that uh, you also have the opportunity to go out and put that content on other websites that um, the LLMs are taking and treating as gospel. So um, so there are more opportunities now to, to get there, just not just your own website. So I think, I think it is important to start looking at where your competitors are mentioned off of their own sites. Um, uh, and where their um, where the LLMs think those opinions matter. So you've got two opportunities. But yes, your own site I would say is the number one thing. You know, it's still about building the brand but uh, if it's very much about building the brand, but it's building the brand in the context of the products that you're selling. The problem with uh, with hennessy.com at the start is that I might be wrong and Jason is, is welcome to shoot me. He gave me permission to talk about this, this particular issue. So uh, uh, so, so uh, hennessy.com what he did way back in the day was he realized that there was a high performance tuning company called Hennessy and he said to them you know why don't you to different you know link to us and we'll link to you so that we can send people the other, you know. And that's why uh, we think um, that the LLM suddenly said this is a high performance dating company. That link may have created a connection that was not contextually um, appropriate. Um, but uh, on the whole you wouldn't do that. You know, you would happily get links on directory sites that if the LLMs are reading them and saying uh, these, these are uh, directory sites then that's important. Companies House in the UK is getting well and truly listed for UK brands. It keeps on um, looking at companies house um, which is a really logical knowledge source to uh, for an LLM to use. And I was always uh, surprised that search, um, engines didn't, didn't bring back Companies House a lot more um, in the UK because it lists every company in the uk. So you know it's a, it's a very logical thing to maybe go back and have a look at your, you know, how you've described your business if you're a UK company at Companies House. Because that's a, that's uh, a something that the LLMs are going to take and uh, digest and when you started your business you may be a very different beast to what you are now.

Speaker B: Absolutely, that makes sense. I mean I think one of the, one of the challenges we hear from clients these days, obviously they are getting fewer clicks to, to their website. Even a number one ranking is not what it used to be with AI overview on top and then we have a few Google components, etc.

Speaker A: Yeah.

Speaker B: What do you think is a strategy for a business these days to kind of still get in or still leverage search, which is just still getting so much traction compared to other AI channels these days? What would be a strategy where say hey, how much should they be still investing in content versus how much should they be maybe going to other platforms?

Speaker A: How much should they, they, they need to have that content written in a way that, that, that the machines are going to understand because the machine is going to do the thinking for the human now just, just continually, you know, AI overviews is just one example. But you know, a lot of us are now, you know, using our phone. I, I see Facebook has now got a, you know, on, on messenger you can have a chat with uh, their Llama product. Um, and uh, so that means that these LLMs are going to do all the thinking for us. And uh, that does so, so, so um, it doesn't mean to say that you don't want the content. The content is now being written for the machines as much as it is being written for the humans. Uh, and because the machines will interpret it, understand it and just give an answer to the human. So of course we're not going to get so much traffic. We don't need so much traffic. We're still going. The number of people buying pots and pans in the world is not going to change. You know, the number of people that buy these pots and pans is not going to change. The methodology in which they buy it may change. Uh, and so it doesn't. Uh, so I think that's the Important thing to remember here. We're all competing with the other people that are trying to buy the, that are trying to um, sell the same concept or product or service. And um, their traffic's going down too. And if their traffic is going down too, and you still know that the same number of people are buying the things that were buying them yesterday overall in the world, then it's not the traffic going down that's a problem. It's the thought process and the buying journey that the user is doing that's changing with LLM sometimes when they don't even know that there's an LLM involved. A lot of users so far they're still using Google, they're clicking on, they're doing stuff or they're seeing the answer on Google and then they're going down to the local store and buying the version. Um, it doesn't mean to say you didn't need to write the content. You still did need to write the content for them to be able to make that decision.

Speaker B: Right. Uh, I think we definitely see this, have been seeing this for a while. Uh, I've been telling my writing has been getting much more technical than they actually expected.

Speaker A: Yeah.

Speaker B: You need to pay attention to a lot more things and already had when the entities obviously came up where you were really instrumental educating the CEO industry as well. Thank um, you. But then definitely also see that the content still needs to sell. Right. And content still needs to sell. You need to somehow balance out what's the, the like technically what needs to be done. You can nlp, optimize an article and you don't. You can't read past the first paragraph because it sounds awful and the content still needs to sell. So we obviously try to, try to strike a little bit of a balance here, but it definitely reminds me of a couple of things. First of all, I saw a video of you from 10 years ago.

Speaker A: Sorry, a video of me of you.

Speaker B: Yeah. I think we both had less gray hair back then. That's interesting. Where you talked about, I think it was about Majestic and the topic coverage of topic analysis, which was really definitely a uh, pioneering, pioneering element back then.

Speaker A: Yeah.

Speaker B: But it was, it was interesting.

Speaker A: Right.

Speaker B: Where. Because I related it somehow to what sparkdoor is doing now, like pretty much looking for sources of influence and what you're doing with Wiki, obviously with the citations as well.

Speaker A: Yep.

Speaker B: Which ultimately is a very actionable, actionable way was like, hey, who is this source of influence? Um, based on LLMs, I have the citations, but when I research then the Audience with other tools as well. I see. Hey, the source of influence is Reddit. Or we do, for example, we have like audience researchers and say most of my audience that's interested in, I don't know, woodworking actually hang out on YouTube. So YouTube currently is a source of influence for this particular audience. How do you, how would you, how would you go about research that shows where actually my client base is. Research that your tool for example shows seems to be cited in LLMs.

Speaker A: Well, well the way I, the way I'm doing it is um, so I'm, I'm using Waycat on, in links, uh, right now. And, and because you start running it, you know, okay, I've got an unlimited budget, relatively unlimited, but I still got to pay the APIs. But so I'm still, so I'm running the reports every week, you know, uh, uh, instead of every month, which is the default we put in there. But it's starting to build up every time we run these reports which is kind of like rank checking reports I suppose, but the new, the modern rank checking reports, um, of course it churns out new sources, data sources. But it's the data sources as I say, that are not your own website and not your competitors websites that are interest interesting ones. Uh, for me. Um, and what I like Reddit for example. Um, so ah, what is interesting is that you and a lot of other SEOs are all saying Reddit is coming back is really important. Um, and I'm sure it is but I've not seen in the way k data as much Reddit um, stuff as I was expecting. Actually a fair amount of Wikipedia, fair amount of companies house. But then I'm in the UK and kind of just thinking about UK companies. Um, but uh, but not a lot, not a lot. I've read it less, less than um, I thought. But what I did see and Crystal, Crystal Carter from WIX also talked about this in a podcast that she, that she was doing. She found that there are a lot of listicles being mentioned, you know, so good old stuff that you and I wrote 10 years ago. Um, you know, um, and uh, and not necessarily on very well renowned websites, you know. So um, and uh, the reason that I think they are showing up so well is that they're comparing all the, all the tools in a vertical. So you know, um, so you know, the best SEO podcast, you know, someone you know, I'm sure there's 20 people that you know or 2,000 people have written the best SEO podcast post with a bunch of podcasts and that is really good information for an LLM to help them answer the question, you know, what are the best podcasts? If they can compare directly on a page. So I think one, um, thing that we can do is write a lot more about our competitor products than we have been before, uh, start to compare our features with their features. Um, be brave about that because then the LLMs are going to take the, take the data from, you know, take the information from our site as opposed to somebody else's site. Uh, and of course our opinion of what's good and the competitor's opinion of what's good may differ slightly, you know. So, uh, I think that's one thing that we can do.

Speaker B: Mhm. One thing that's interesting because you mentioned Reddit. What I find interesting is on the one hand we try to be as consistent as possible across different platforms when we want to bring information into LLMs. Getmatic is an SEO. There's probably hundreds of mentions out there on the web. We have been working on this to make sure I get my knowledge panel. But uh, at some point you see also things coming up that I just mentioned once. So I remember I once asked the LLM when this, I think it was early days of ChatGPT, maybe this doesn't happen anymore. But I asked, yeah, like top 10 running shoe brands and Nike wasn't there.

Speaker A: Right.

Speaker B: Um, and I was like, this is weird. Why isn't Nike there? And it came back and said Nike has manufacturing issues and was like, that's new. M. I researched this and the only thing I found was one single product review where one model a few years back had manufacturing issues. So this definitely brought up like an entire, an entire new reputation service we did here because I said, hey, this is going to be a thing.

Speaker A: Absolutely. Right. Yeah. I think right.

Speaker B: On the one hand, you try to be as consistent as possible to teach them about probability. And I don't know, there must be much more positive content out there about Nike and running shoes than just one single review. Can you imagine any, any reason why like a particular single layer page can actually impact an LLM generation?

Speaker A: I think, I think part of the problem, this is why I think you have to run, um, you have to, you have to see things over time why any one particular LLM output has a, has a, a reasonable chance of hallucinating, you know, um, but, but put through, ask it three times and unless it's playing a trick like how many, how many A's in, you know, in this word here, you know, some of These, some of the errors that the LLMs make, uh, it'll, it'll, it'll get it right most times. Um, I do think this is an interesting ah, thing. It's, that's exactly the same problem as the, as in links suddenly being seen as a rank checking tool. That was only one company that had mentioned that. Uh, the only thing I think is, you know, that's why the fact checking um, things in Waycare are a really quick way of getting through these problems because most of the time it gets it right. When it gets it wrong, what you want to be able to do is very quickly find that source and correct it. Because a source could be a bad review somewhere, um, which is hard to fix. It's more likely to be either something, an error on your own website, something that's changed in history. Like in Nike's case, they had one problem that's changed in history. Um, and uh, um, the way to deal with that is to um, maybe talk about the manufacturing process, um, at Nike. So have some content that is richer that explains the manufacturing process and the, you know, the degree of excellence and stuff and the you know, number of you know, the 99, 9 accuracy rate or whatever, whatever that happens in, in the, in the shoe industry, um, it's not necessarily going to fix the problem. The only way to fix the problem is to go back to that source and say that's not correct, please can you change that or take it down. Um, and I think that there is going to be in some brands, the ones with the money and the lawyers, they will lawyer up a little bit because um, it's not so much that the LLMs are lying. Ah, it's a page on, on the Internet which before wasn't significant but now is having an impact on what users are perceiving about the brand. Um, so, you know, yes, there's going to be A, a lot of negative AIO and B, uh, a lot of uh, a lot, a lot of, a lot of lawyers lawyering up over sometimes petty issues.

Speaker B: See those things opens up like a lot of, a lot of opportunities for negative work. Right. Or black.

Speaker A: Yeah, yeah.

Speaker B: Or like negative SEO was a thing, negative LLM thing or whatever put this out.

Speaker A: Um, and it's not as if it's, it's not as if it's not something that hasn't existed in, in the whole of history. You know, it's, it's basically propaganda and people will be using propaganda um, to educate the LLMs correctly or incorrectly and if they're a bad actor, then you know, they're going to be very hard to track down after the fact.

Speaker B: Awesome. I've got a final question before I let you go and appreciate the time you take for this conversation. Um, everybody should be definitely checking out YK as a tool.

Speaker A: Uh, well, thank you, Gert, and I'll send you the beer in the post.

Speaker B: Yeah, what can I just say? I'm not affiliate, but I do take the beer. Um, so, uh, one thing I find interesting or what's definitely for, I think new users is just a different process. So you have like, one report about the brand. So everything as I understand it, correct me if I'm wrong, everything the LLMs give back about the brand, which gives us an idea what kind of main entities are associated. But then you also have the possibility to essentially ask it for the relationship to different topics.

Speaker A: Yes, yes.

Speaker B: Um, how do you. How should somebody go about choosing which topics to ask Waycade to monitor? So I give you an example. We have SEO leverage here. One of the first things I typed in was SEO consulting, because I want to know how strong our relationship is with SEO consulting. But I might also want to know how strong our relationship is with Australia, because we have a lot of clients in Australia. Okay, come up for somebody searching for an SEO consultant in Australia. Or, uh.

Speaker A: Okay, how do you go about it? So, um, firstly, the difference between a brand report and a topic report for me is a brand report. If you've got Nike and Adidas and Reebok, they, at the brand level are trying to convince us all as consumers that they're different. Right? You know, they have. They want a different story. You know, one of them's technical, one of them's, you know, cheap, one of them's, you know, uh, uh, I don't know, got real athletes behind them, whatever it may be. They've got diff. They're trying to create a different narrative because they want to be differentiated until it comes to what shoe should I write wear at the New York Marathon? You know, um, and then they all want to be there. You know, they're all sponsoring the event. They're all, uh, uh, trying to get you to buy their shoe. That's, you know, now they don't want to be so different at all. Now they want you to buy the product. So, um, topic reports are much better for telling, uh, for being able to work out what you need to do differently, what content you need to provide to fill those topic gaps. Um, uh, brand reports, yes, you will have differences, but you might want those differences so at the topic, uh, the brand level you don't really. There's no point in having a topic gap plan at a brand level because you're actually trying to be different at a brand level, but at a product level you're trying to be the same. So uh, but when you get down to the uh, to the topic reports for you, uh, I mean SEO consulting makes a huge amount of sense. I think we were thinking that people will generally be thinking in terms of uh, the products as opposed to, in terms of Australia. It's interesting that you've got Australia as a, as a, as a concept. I don't know what's going to happen when you put in Australia. You probably already have, but I, my guess is that your competitors, it'll only work well if you also chose competitors who are in Australia. So what I would do in our tool, if I was really thinking about this, um, is I would have a SEO leverage tool for um, project for you and your main competitors in your main, you know, your usual competitors. But then for the Australian concept I would actually start with a different project. Doesn't cost to put more projects, it costs for, per report. But you can have report for Australia anyway. I put a project for SEO leverage and two Australian competitors. So now you don't need to particularly talk about Australia because you're using Australia as your, as your base. But you can now, uh, and it will start, but that will start to show you uh, what your competitors are talking about that you're not. And now you've got a relevant seed set. So I think it's about the two competitors that you choose. Create a different project and make sure there's Australian ones. Because it may be that frankly you're completely missing out on Perth, you know, and the main SEO market is Perth,

Speaker B: not Sydney where I was going after. Right. So obviously, uh, in your tool and any other similar tool that might be copying you in the future, they are going to limit the reports. So you need to kind of choose wisely which kind of topics you monitor requests. Um, costs through every tool through API costs, obviously.

Speaker A: Yeah.

Speaker B: They are going to be always limited.

Speaker A: Yeah.

Speaker B: So we need to somehow make a decision and. Okay, what are the like the main topics we want to m. Monitor and ideally maybe check every week, the most important ones or every month.

Speaker A: Yeah.

Speaker B: At some point we need to make this decision. So if I understand this correctly, you would go mostly with the main services or.

Speaker A: I would, yeah. And if you want to focus on, on Australia, then the thing is I, I, well I, I would not be hesitant of starting more than one project. Um, if. If I'm kind of. So you. We. We talked about the. The. The. The, uh. Site with uh. Completely different varied project.

Speaker B: I talked with your developer about it

Speaker A: and I should have. Yeah. What I should have said, but I wasn't thinking straight when you said it is. What they need to do is set up different projects for different.

Speaker B: Uh.

Speaker A: Different. Yeah, because there'll be different competitors. Uh, because there'll be no other company that's selling 10 bags and. And. And tires and, you know, tobacco, whatever it may be.

Speaker B: Yeah. Awesome. Now that makes it m. A little sense. I want to be careful with your time. Thank you so much. I very much appreciate you coming by again.

Speaker A: You're very welcome, Gert. I really appreciated it. And uh, uh, thanks for giving me a little bit of space before. Before we came on the call with my, uh. By technical challenges. Should we say.

Speaker B: No. No worries. No worries. Whatever you need. Uh, thank you so m. Much. If anybody wants to get in touch with you. What. What would be the best way right now?

Speaker A: Hey, well, uh, so waycay, in case you can't spell it guys, stands for what AI knows about you. It's.IO so waycay IO um, dixonjones.com and I've got, you know, the about page, contact page should give me options to see me or just type in Dixon Jones into Google. Let's see if I come up.

Speaker B: If you type into YouTube, you see a 10 year old video, by the way.

Speaker A: Oh, you do? Well, okay. Well, at least it's still there then.

Speaker B: It definitely shows the credibility. Thank you so much, Dixon, everybody in this episode and reading a summary, head over to SEO learage.com podcast Dixon Jones back to the show. Thank you so much. Thank you.