Should You Track AI Citations? The Best AI Visibility Tools for SEO Agencies with Nathan Gotch

James Dooley: Best AI search visibility tools. Today I'm joined with Nathan Gotch. I see a lot of his videos throughout YouTube. I see him on a lot of podcasts and he talks a lot about AI citations. So I want to go deep about AI search visibility because I've actually had one or two people... Before I get started on here, I just want to give a little bit of a... I've had one or two people saying, "Oh, I think the AI search visibility tools are flawed. If I go and do a search and I do another search the next day, it's different. Oh, if I do a search and my missus does a search, there's the history that comes with it and it's different." And I feel like data is key and I feel like you should be tracking it and in an overall share of voice kind of metric, you should be looking at it on a day-to-day basis. But that's my opinion, but I want to go straight into Nathan Gotch. First and foremost, do you think you should be tracking the LLM visibility and, if so, why?

Nathan Gotch: Yeah, well, let me start by saying I'm extremely biased. [Laughter] So, you know, I, I have a tracking tool, so obviously my answer's going to be yes, right? But it's, it's not just yes because I want to make money. It's yes because it's what I was already doing before I had the tool and in, in a lot more inefficient way, by the way. So, so that, you know, the value of any software is that it helps with scale. Like, you can, you can certainly run manual searches. I actually have a training where I show people how to actually run tracking without needing any tools at all. Uh, you can do it. It's just painful. It's painful. And if you're working on maybe one campaign or a tiny little campaign, you don't need a tool. You definitely don't. You can just set up manual runs and do that and collect data and be pretty solid, right? But as soon as you, you have a bigger campaign, you got a lot of targets, you got a lot of clients, a lot of camp... You need a tool, right? You just do. Um, um, so now as far as, like, some of the, uh, suspicions about AI tools or AI search tracking, uh, a lot of them are true, okay? Like, a lot of the things that people say that are, like, let's say, uh, speculation about the effectiveness of them or whatever it is. I'll, I'll say a couple of things that, um, I don't agree with and we'll kind of go from there, right? So, the first one is, is, uh, prompt volume, okay? So, any SEO has heard of search volume, but there's a new terminology called prompt volume, okay? That does not exist. That is not a real thing. There's no prompt volume metric that you can pull from ChatGPT. You can't go to ChatGPT and say, "What's the, what's the prompt volume of, uh, best SEO tools?" It doesn't do that, right? Um, and so that's not... That is not a thing. Now, what you'll see is a lot of, um... There, there are ways that these, some of the AI tools are trying to get this data. So, one of them is, uh, there's something called panels, where they actually will have... Someone will, uh, they'll pay someone to put a little injection into their Chrome browser, and then they track their behaviour, right? So, that... But even then, it's ex... That's such a small sample set that you can't get anything reliable out of that anyway. Um, and then the other mechanism they use is just basically pulling traditional search data and then saying that's what's happening in the AI platforms, right? And I think that's relatively fair. I think more than likely queries that happen in Google are likely queries that happen in AI too, just in a different format, right? That's, that's all it probably is. So, on that front, that's one thing I want to make sure is clear. The other thing is with query fan-out, right? We hear a lot about query fan-out and, like, "Oh, you got to track, like, all these sophisticated prompts and all that." You don't need to do any of that. Um, we have seen so much evidence that if I were to put a dissertation into ChatGPT of, like, all this context. Like, what's the best SEO tools for someone with blue eyes, is 6'2", and lives in Manchester, UK, right? You get, like, real specific, okay? All the AI's going to do is just take that huge context and just break it down into little queries and then go and search those queries. And it will expand on those queries, but they're usually really, really simple. Like, super simple queries. So, what we found, like, we tested this in the beginning when we first started getting into tracking, and we would do these long prompts and we would do all this stuff and we would collect the data. And then we would find out that it's like, well, we did all this, but then we run the short, the short queries. Might just be, like, "Best SEO tools for agencies." The same brands pop up over and over and over and over. It didn't even make a difference to run all these sophisticated queries. It did literally nothing as far as improving, like, our insights into that particular topic, right? So, it ends up being just a waste. And so, I, I've been trying to get people to waste less credits and stop tracking, like, dumb stuff and just, like, keep it... Keep it real simple. Like, you don't need these complex things. So, that's, like, that's the big one. Um, now, as far as how tracking should be done, this is why I personally think that you should split it between two different units. One tracking mechanism is purely the answer. Just the answer, nothing else. So, you see, is our brand mentioned in the answer? Yes or no. And it's a simple binary, right? And then from there, if we're mentioned, where do we appear, uh, in that response? So, are we, you know, at the top, in the middle, whatever, okay? And that's important because even, like, in traditional search, if you're number 10, it's not as good as being number one. So, you want to be the number one most recommended. And then you run that across every AI platform and say, "Okay, this is our coverage across every platform. This is our average position across every answer." And then you run that... You need to run it many, many times. Many times. To go back to what you said at the beginning of the conversation, um, yes, it is non-deterministic with these answers, right? So you... If I run a query right now in ChatGPT and then nine seconds later I run the same exact one, the answer's going to be slightly different. Slightly. But over time with enough runs, everything stabilises. You're going to... You could run best SEO tools five billion times and I guarantee Ahrefs and Semrush will pop up. It will not... Like, on average, they'll be in there 80, 90% of the time, right? But of course, if you run a prompt today and then run another one in 90 days, yeah, you're going to see some weird stuff because you're not... You've developed no base to work from. So we've just found that, like, consistent tracking and then tracking a lot of different variants that have unique intent is a good way to get some kind of, like, some situations where new competitors might pop into the mix. Um, and that gives you kind of a broader view. Like, an example, um, I ran on my channel, my YouTube channel, was, uh, best beef jerky. Okay, so that was, like, the seed, but then you did, like, best beef jerky for... For, um, or best beef jerky that has a lot of creatine, best beef jerky that has no salt, right? And you add these variants and it actually does change the makeup of who kind of shows up there, right? And, but at a broader view, when you're tracking 20, 30 unique prompts, you're going to get a good look. Like, "Okay, I have a good idea of who's competing here, right?" Um, and that's the way you do it. So yeah, we can kind of expand on that, but that's, that's where we are.

James Dooley: Before we move on to the actual tools, the best AI search visibility tools, um, you mentioned that about a listicle. Let's say the LLM is bringing back 10 brands and you're all number 10 in the brand, in, in the list. Is there anything that you're seeing that helps you climb up that list? And I want to throw a bit of a caveat onto this. Do you find that it's just more consensus of sources that you're getting in? Or do you think there's anything to do with the, the strength of the actual entity that's being put in there on the list?

Nathan Gotch: Uh, I think it's both. Yeah, I think it's both. Yeah. I think it's a portion of it is what's baked into the static corpus, right? So, like, actually baked into the model. And, like, you know, you think about, like, Semrush is a really good example because they started in 2008. So, you think about how much time they've had to just, you know, and, and ChatGPT trained on all of that historical data, right? So, they're just baked into the model. You don't even need to use web search and they'll show up, right? Um, so that's, like, being baked into the models really, really important. And unfortunately, there's no magical thing you can really do there other than just do good marketing and, you know, you do everything you can before the next training set, right? That's all you can really do. There's quite actually a little, little tip for you on the, with the parametric, like, the training data and the parametric one, is that a lot of the LLMs, you do know that they download Common Crawl database. Common Crawl database, um, in total, I think it's about 12 to 120 terabytes each month of what they crawl. I think it's something like 11 petabytes, which is just insane how big that is. Um, but the... If you start looking at harmonic centrality with regards instead of PageRank and how close you get to them, that shows how frequently you're being crawled in Common Crawl. And we've actually seen websites not being indexed in Google, but they end up being cited in AI Overviews. And it's mainly because of harmonic centrality and Common Crawl is what we believe, or wherever they're getting the training data from. That is just one source. This is a little tip for you on that. Have you done... Do you anything with the training data or the core, the actual trained corpus? Do you ever look into that?

James Dooley: Yeah, we... So what, for our page auditor, one of the mechanisms we have is probing the static corpus. So we'll try to... Yeah, so we'll try to see if that URL is showing up in the corpus and that's one variable that we look at. So yeah, I mean, Common Crawl is... It's very... Like, of, of the things that are basically confirmed, it's that, that they use Common Crawl in the training process. The question is, is how much do they use, right? And we don't really know at that point. We know it gets, like, here's that huge chunk, but how much of that gets synthesised into a very small little tiny thing that they train on. No one really... That's, that's the secret sauce that no one really knows, right?

Nathan Gotch: Well, well, but then it comes back down to more data is key and this is where tracking the visibility is, is a, is a vital part. But anyway, I want... I want to... I want to throw you in the deep end. So what I want to do is I want you to name, in your opinion, before we move on to Rankability and why we believe Rankability is the best AI visibility tool on the marketplace, who is your five biggest competitors, in your opinion, for not just competition, but who you believe are the best five AI visibility tools behind? So we're doing sixth, fifth, fourth, third, second. Rankability is the number one. But before we get on to Rankability, who's the other five AI visibility tools that you would recommend?

James Dooley: Yeah, I don't know if I would even recommend five, but... [Laughter] Yeah, yeah, let me... I'll tell you the ones I definitely like, the teams that I know are doing good work, for sure. And I'm, like... Like, I'm impressed with what they're doing. Uh, the one I would say is Peec. Yeah. Uh, I think, I think really, really good team, super smart guys. Uh, the reason I like them in particular is because they're... They have guys with SEO background, right? They're not just, like, VC-backed guys that just want to make a quick buck. Like, these are, like, they got some hardcore guys in there. So, I, I respect them a lot. Um, obviously, the, the some of the legacy players, they, they've made some good gains. I would say Ahrefs is, is doing some pretty good work. I, I actually like what Ahrefs has done recently with agentic-related things. They got that Agent A, and I think that's a good little thing that they built there. Um, their tracking is a different... We'll just say that much. I think it's not as capable because it's not, like, their thing, right? They're kind of broader tool. They're doing a lot of stuff. And I think Ahrefs is still a great tool. Semrush is, is pretty good too, right? You can... I mean, if you want, but it's, you know, a lot of these are pretty expensive. When you, when you start to do the math, it's like, uh, HubSpot, for example. You know, you get, like, to track ChatGPT for $50 a month. And it's like, "Whoa, okay. That's a lot." Um, so, uh, but I would say, like, the ones that are definitely are doing good work with, like, Peec. Profound is a good tool too. They've got some really smart guys there working as well. They've got a lot of funding. Tons of funding. Um, I like AirOps. AirOps is another good one. Uh, they've mainly... If you're looking more for, like, automation-based stuff, uh, and they've got some pretty solid tracking. Uh, and actually a, a sleeper that almost no one talks about is LLMrefs. I actually, I really like that tool. It's mainly because it's so simple. I like simple tools that do one thing really, really well. And they do that. They do that really, really well. Um, and there's also ZipTie.dev. That's another one, uh, run by a few really good SEOs. So, I, I gravitate towards the SEO guys, right? Because I, I think that SEOs are going to be the most successful in this new era. Just my opinion. So, yeah. So, so let's move on to Rankability, right?

Nathan Gotch: And I want you... You obviously own it, so the caveat's there, or it's a self-serving listicle and all that garbage. Yeah. Why do you genuinely believe... Because we spoke off air and you didn't... You could have said, "No, I like... I'm just doing it for the money and stuff." But you said, "I genuinely believe it's the best and I'm going to... If there's certain areas that need improving for what the customers want, I'll improve it in them areas. I want it to be number one." Why, why do you think it's the best in comparison to the other AI visibility tools?

James Dooley: Yeah, I mean, it's, it's obviously trying to be as objective as possible here about my own, my own thing. Um, I think the... I think the big, what's called value prop for us is that we focus on the full process. And that's the big thing because, like, I come from an SEO background and I don't just think about, hey, how do we get some cute little charts? Like, that's not... That's not the thing that I think about. I think about what are the mechanisms we need to drive results? The, the, the data and the analytics, that's all fun to look at and get benchmarks and create proposals. That's all cool. But when you, when you get down to it, it's just data and I want to influence that data. I want to change that data. So, for me, that's been our focus with building these tools. Like, "Okay, how do we, how do we tackle each part of that SEO process end to end?" And I feel like we're, we're pretty close to that. Obviously, there's always more you can do, right? There's always more. I have, as we talked about earlier today, uh, lots of ideas. I have tons of ideas that I want to go after, but it's more like I'm focusing on the big heavy-hitter things, right? So, from keyword research to auditing to building content to optimising existing pages, you know, all, everything in there. Even Google Business Profile optimisation, like the full pipeline. And that's, that's my focus. And, um, I always said that, like, once I could cancel certain tools, then I knew that we did the right thing. And, like, I won't say what tools I've cancelled, but I've cancelled those tools. And I... And this is... I'll say the biggest thing that I don't know if this is true for most SaaS products, but I actually eat our own dog food. Like, I use our tool to do SEO for our own business. [Laughter] So, and, like, you... I could show results, but, like, you know, we're a tiny little team of three. We got a team of three, got a lot of external help, but a team of three internally, and we have this... Like, we are competing and beating some companies that have $200 million of funding. We got $0. We got... Our funding comes from our own pocket. So, um, and so the way we're able to do that is obviously using the tool, right? That's the... That's, that's literally what we do. And, like, people want this, like, magical thing. Like, "How are you doing that?" I'm like, "I used Rankability." Like, they don't want to believe that I just used my own tool to do it. Like, but I do. So, um, so yeah. I, I'll tell you what I love about it. I love that anyone who's watching this now, go and check out the episode in the description where it talks about increasing AI search visibility. It talks about all the topical gaps. It talks about all the places where competitors are being mentioned and you're not being mentioned, and being able to reach out to them existing sources. And literally within a few hours, if you was to get your brand mentioned on that listicle, enforcing that URL, within a couple of hours could be the difference between you not being cited and you being cited. I love the fact that you've been a long OG in the SEO industry and you understand what it takes to win. And I think a lot of it comes back down to forget the what's baked into the LLMs, the live search, the RAG of it going pulling a live search and getting that information. There's not many people that know that better than you, that understand search engine optimisation better than you. So, the fact that you've created an LLM visibility tool that then is kind of not just tracking the rankings, but then telling you what you need to do to then improve upon them rankings. A lot of them aren't doing that. They are just literally giving you the data going, "Oh, by the way, you've gone up on mentions. Oh, you've gone down on citations." You're not... You're, you're basically turning around going, "This is how you reverse-engineer and get better than your competition." And that's what I do love about Rankability. And bear in mind, people are watching this, is going to think, "Oh my God, this is a paid promotion for Rankability." Um, and it's genuinely not. I just... There's a whole series about the AI visibility and agentic AI, and building a team, and what AI website builders should be using. I strongly recommend checking out all of the links in the description. I wanted to do this one last in the series of speaking about AI, so it interlinks everything together. But I wanted to talk about the AI visibility because off air, I knew that it was needed when I was speaking to him. I knew that tracking it is key because you've got to be making certain that what can't be... What can't be... What is it? What can't be managed or what can't be measured can't be managed and stuff like that. So, you've got to measure everything of what you're doing, and the data set needs to be high. And if you keep tracking on a day-to-day basis, you can start seeing slowly but surely the share of voice overall, AI citations of what you're doing is working. Anyway, anyone who's watching this, check out the... Put a comment in the comment section with you guys that... Has Nathan Gotch missed out an AI visibility tool that you believe is the top-rated, is the best in the world? I want to know which one, which one that is. Leave a comment in the comment section. I just want to say from myself, Nathan Gotch, it's been an absolute pleasure. And anyone, get over to Rankability. Is it Rankability.com? Is that the, the website?

Nathan Gotch: Yeah. Head on over to Rankability.com, sign up, go and start checking what LLM visibility you're getting across all the different platforms. And yeah, just want to say thank you very much, Nathan. It's been a pleasure.

James Dooley: Thank you so much for having me. Appreciate it.