The GEO Grift: What Actually Works in AI Search (SEO Myths Exposed)

David Mweeny, you have been involved in SEO since 1997, former head of content for Hrefs. You ran your own e-commerce business from 2006 to 2012, consulted for 15 years. Thank you for coming on the podcast. Thank you very much for having me. We're going to be talking about some of the biggest myths in SEO and geo today. And uh and somebody sent me this viral LinkedIn post and this this viral LinkedIn post was the catalyst for this episode. So somebody wrote somebody wrote, "I published a completely blank website, white page, zero visible content, but seven layers of structured data underneath. JSON, LLM.txt, TXT ed25519 signed entity claims. Within 36 hours, it became the number one cited source in Perplexity. Chat GPT cited it independently. No human ever saw content on that page. Completely blank website, white page, zero visible content, but it had JSON, it had lms.txt, it had these entity claims. And so, uh, to to start this podcast, David David Mweeny, do business owners and SEOs need to pay attention to Jasonldd, LLM.txt, and these signed entity claims to get more visibility in AI. Yeah, I mean, your local plumber has to get these signed entity claims implemented immediately, otherwise they're visible to AI. That's of course I am I am joking. um being sarcastic that particular study uh I didn't I didn't hate the study at the end of the day people doing things and experimenting is good um unfortunately the takeaways from it were completely flawed uh because the problem was the first thing I did was I looked at the source and there was that there was text was there that was what was getting Chat JPT and perplexity are very naive crude parsers. Um Google is more sophisticated. Um there's another issue there. We'll come on to that. But chat GPT and perplexity would have just got marked down which would have been the HTML content that was in the source. They don't care about the fact it was screen reader only. They would have seen the text. Sorry. They don't care about the fact that it was what? Well, we well like a screen reader only class on um so the text the text is that if you view it in a screen reader you can actually see the text but the point is it wasn't a blank page there is readable indexable parsible content on it. you could have stripped away all the other seven layers or nine layers or whatever it was of uh weird madeup jargon and you'd have got got the same results or or in in my opinion you'd have got the same results. Um it seemed to more though the guy's um original thing was he said within I think it was within 24 hours I was cited in Jack GPT perplexity. Now the other problem is cited for what? What was he cited for? H and it turned out in the comments he was cited for a very specific search for this very specific thing that he had invented which existed nowhere else on the internet. So if there's nothing else to index a site, then of course if you're the only option, then you're going to you're going to appear for that. This method of marketing is so effective, I had to make sure it wasn't against Google's rules before I kept using it. It's a form of SEO I call compact keywords. Whereas most SEO focuses on putting up articles to answer questions, how, what, when, compact keywords focuses on putting up dozens of pages that sell to searchers who are actually looking to buy. These pages rank on Google and convert so much better than normal that when I discovered this years ago, I couldn't believe this was allowed. It's less work, too. The average Compact Keywords landing page is only 415 words. Compact Keywords is a 13-hour deep course on getting sales with SEO. A customer said, "We spent nearly 18,000 in the last year and a half on marketing and SEO through different agencies locally, and that did nothing. We decided to take the leap on the compact keywords course. We're now getting about 6 to eight calls per day on a good day, which is just unheard of." Another customer said, "Give it to a junior employee. have them follow it exactly as Edwards laid out. You don't have to do anything and you're going to gain a six-figure SEO level employee just by having them go through this course. Compact Keywords is about setting up an SEO funnel that brings you sales for years and years and years. It works with AI. It's less work than traditional SEO and it makes way more money. You can get it now at compactkeywords.com. Back to the podcast. But in his post, he was talking about LLM's chat perplexity, but it seemed to pick up um and go into another area over the weekend, I think it was Mark Preston, who I know tangentially, um I'd shared it. And then Sean Anderson, another good Scott Oh, he's been on the show several times. Yeah. Yeah, I've seen I've seen one of his episodes. He has an excellent Scottish telephone voice. Yeah. Uh yeah so it seemed to go from what the guy had said originally I was cited in chat GPC perplexity to Google can index font content or pages with no content which was a completely different argument and a completely different thing. Um, so we had the guy originally talking about, as I say, the LLM side, which is explainable through the fact they would have just seen the the content as content. All the styling and everything would have just been stripped away. They would have seen markdown. They can read. Um, the Google side is maybe slightly more interesting. Should they be indexing blank pages? I mean, no. Uh however that where I quite like it is an interesting page. Have you tried the Konami code on it for example? I tried the what? The Konami code. No. Can you talk about that? Do you know the Konami code? I'll the Konami code. It's an old uh the Konami code. Yes. The Konami codes. No, I haven't tried I haven't tried that. Can you explain it? Well, I mean, looking at the source, if you go to his for this, I think it's phantom authority.ai or whatever it is, if you do up, up, down, down, left, right, left, right, ba, there's no star. You'll get a matrix overlay thing. Oh, yeah. You put in an Easter egg. Yeah. Yeah. Yeah. Which I quite like that kind of stuff. That's cool. I like Easter eggs. If you think about it, um, from Google's side, how they don't know what it is. They have no idea what it is. and looking at the the source there's lots of JavaScript like I said this Easter egg thing it could be an application it could be anything so I mean yeah I mean David Quaid who I know is on your podcast a lot I think he's on it just about every second day is great yeah no fantastic but um yeah he's very strong on the Google being cons content agnostic which to an extent I agree with I don't fully agree with it but in vain. I agree with it. But um we can do we can do a debate episode in the future. I'm much more closely aligned to them than uh yeah, we're we're closely aligned with most things obviously. Yeah. Yeah. But yeah, so this could be an example of the fact the very fact that Google's content agnostic. It doesn't know what it is. I mean there's a there's a JavaScript Easter. It could be a game, you know, it could be a game that you get there. So if you do this, if you do the Kami code, uh, you get a matrix sort of overlay thing that explains the the guy's concept. That's cool. That's a cool Easter egg. So to sum to summarize ba basically uh to summarize all of this, all of this text, human readable text was in the JSONLD and it like there was actually text for the LLMs to use. Yeah. But not the schema and the you know the I forget what you called it but the other the other things he talks about it was literally in HTML just standard divs like you could you can go and look at it as a class of SR Python only uh and the CSS hides it for non screen readers. Uh but yeah, I mean it it was only a couple of weeks after I'd been I'd been having a lot of arguments generally about people with the uh it was after the Cloud Flare, you know, on the Cloudflare doing the Mark 7 pages in Markdown, which I'd been uh or alternative version of Markdown, which I've been kind of vocal on. I then was running some tests on um chat GPT and um to I mean I knew the outcome to be honest but I wanted to sort of confirm it. So I was running some tests to show that they get pared markdown anyway from HTML uh and also that they can't see certain elements. They can see certain elements they cannot tell when an element's hidden or not. um fava openai slashgbt in their retrieval pipeline. But yeah, so I did a hidden element test and the hidden elements where chat GBT could see them and read the content. It doesn't care about styling or how things are rendered on the page. Very it's a crude naive parser as I say. Imagine, okay, so imagine that you have you had this plumber that you that you talked about as a client and it it's a very non-technical client that you have. You could have plumbers who are technical, but this one happens to be non-technical and this plumber sees this this viral LinkedIn post and then says to you like, "Do we need to be using JSON? Do we need to be using LLM's txt? Um I see that you wrote some article called the great geogrift. Oh yeah. Uh do we Yeah. And we'll talk about that. But the plumbers like do we like you're you're anti you're anti all of these things. Why um should we not be wasting time on that? Or why I'm I'm not anti- schema and JSON LD you know when it for the purposes of what what it's for. No no for but for geo but for geo. Yeah. Yeah. It I mean in my opinion and in an opinion and there's conflicting um there are conflicting opinions on this the structured data itself on a page other than you know shopping e-commerce obviously price and whatnot. I mean chat GBT uses Google's uh shopping feeds so it has direct effect there. um but putting your business and whatnot uh chat GPT currently they have they are possibly moving into having their own knowledge graph at the moment they're scraping Google's knowledge graph um I mean I mean it's worth putting these things in you know if you're a law firm it's worth putting in legal service and whatnot these are all worthwhile things but not it's not going to make you rank number one in chat GPT as everyone seems to say um and it's generally going to be strict in my again in my opinion it's going to be stripped out in pre-training it's going to be stripped out in retrieval in rag retrieval um lex is another story that's a complete waste of time you only need to go and look uh some prominent uh case studies some of the larger I'm going to going to do the this AEO tools Um, and they have LLM.ext. In fact, one of the larger AEO tools, a New York based uh, AEO tool, but probably quite near to yourself. Uh, they, one of their big studies, one of the main things they say they did was implement LLM.ext. Now, the LLM.ext, I went and viewed it the other day, is 2.3 megabytes of um, well, Mark 2.3 megabytes of markdown, which is a lot of markdown. It's 9 and a half, 9 and a half thousand lines. Um, that's crazy. And about 70% of the pages went to a 404. Um, they implement these things, they don't do anything, leave them to rot. Um, another one major US fintech company who seems to be client or a case study of every single AEO tool. I don't know what they all I don't know what they each of them do for them. But again, their own llms.ext, they have one. And uh certainly at least two, this was a smaller, this wasn't 2.3 megabytes of everything under the sun, but at least three or four of the links were broken on it, which again suggests it's just been abandoned because it does it does nothing. How much of an impact does uh does Jason does how much of an impact does schema have on getting more AI visibility? In my in my opinion, none other than two things. One, it may possibly from time to time in some cases. I'm being very hedgy here, but you know, because I'm not saying it does nothing for search, but if it helps your rankings and helps you rank higher for certain things, then you're more likely to show up in query far out or search and be retrieved. So that's the first instance. The second one is in my op I mean I I did a quite a bit of testing on chat GPT's retrieval although immediate um admittedly this was you know still test it all the time but I did a major deep dive about five months ago which was an which is now an age in the you know in internet AI times but at the time it and I believe it's still the case they They when they retrieve they get search results or search results they get set and if those steps have um like you know review stars or pricing information in them from the schema then they most likely get that in the initial scrape that they get either either a scrape or a direct API feed from from Bing if they're using Bing they use they seem to flip flop between a providers. Um so this the the additional information you know the review stars and whatnot within the search result within the initial scrape of the search results you see may make a particular result more attractive and in that case it might help that result to be retrieved. So rich. Yeah. But you're you're but you're you're saying that you're saying that if the SER uses schema and and then you have schema to to increase your click-through rate that's going to increase your rankings and by increasing your rankings then you will do better within AI. Well and two two things. one that is that side of it. But yeah, secondly is literally when Open AAI suppose I should hedge my language when if OpenAI happened to scrape Google through SER API which they were at one point in in that feed because you can go and use SER API yourself and can see what you get in the a Google result. They would see the review stars. They would see the price and whatnot. So, nothing to do with human clickth through behavior or ranking. You know, if it was position 10 and that was the one with the review stars, it may make it more attractive to the That makes sense. Initial uh but that but that also depends on the SER using schema. Yes, of course. Yeah, absolutely. Yeah. So yeah, I mean, you know, and for for schema, I'm not against schema, you should use the relevant structured data for your for your site. Um, I think I think the thing with structured data and schema, I'll be honest, like and a lot of these things with geo people and what structured data and schema sounds computer. It sounds uh LLM AI um you know other machines. So they want machine readable parsible structured data. So that makes you rank higher when in fact they've been trained on everything that's ever been written by a human. Um they can they they're large language models are not large schema models. They're designed to parse natural language. So, I I honestly think that part of the problem is that people just um people want there to be new things and they want there to be thing things they can talk about. So, a lot of the time people just say things that sound good and in effect they're not actually doing anything. and and yeah like there again going back to the original guy the guy with the nine stage or seven stage whatever I think it's good that people have hypothesis and test these hypotheses but it needs like um extraordinary claims require extraordinary evidence and if you're going to make claims that something has an effect on AI visibility um then you're going to have to come up with the the proof that's not just a lot of these studies are correlation causation studies. Um you know they'll say well 6 64% of cited pages had female. Well that's just a distribution of content that ranks on the internet. Um that was a madeup uh percentage there but you get you get the idea. Um and yeah, most most of these studies tend to be that and they tend to fall on the they go in the size of the data set. Uh they they say, you know, we examined, you know, 24 million prompts. Um it's an appeal to authority. Um big David Quaid would like that. He's he likes his fallacies, his logical fallacies. Uh yeah, it's it's an appeal to authority of we we've got a data set of 24 million um problems. You know, you can't argue with this here. Um but in reality, you could look at 10 prompts and if you looked if you tested a hypothesis and came up with something new and interesting, then that's much more I' I'd be more interested than that. I'd be more interested in someone testing one prompt if there was something other than add schema, use headings, um, you know, writing clear, logical, structured sentences. It also does it also makes sense, you know, um, let's if you even if you're using Yoast, which is probably the the top SEO plugin for WordPress and so many sites are made with WordPress and then are using Yoast. Yoast by default will put in very basic schema. And so it's like then you're going to have all of these websites that are ranking that are also using schema. Yeah. By the way, for listeners, I'm not against I'm not against schema. I just think a lot I think a lot of schema not I think just straight up a lot of schema isn't actually used or barely used and so it doesn't increase click-through rates and it and as a or like or it's just ignored in other in other factors and it's not getting you anywhere. It's a waste of time. It's something that that SEO agencies will charge clients for because it sounds fancy when it's that's not the thing that is going to get the clients results. and the clients are paying for things that aren't going to make the money and and and then and then there's like you said the correlation where it's not actually increasing AI visibility. You know what's increasing a AI visibility? It's relevance is having more authority. Exactly. Exactly. You wrote this art Yeah. Go on. Sorry. No, sorry. Cut. Oh yeah. I wanted to I wanted to ask about this article. So you wrote the the great geogrift and you call geo generative engine optimization a waste of time and money and you say that it is polluting the web. Yeah. I mean you know I don't think I'd had much crunch here that day. So it's a little bit agitated. I I don't know. I mean there's um there are things that are different about um search and retrieval but that's been the case you know as long as I've been doing search I mean I mean I I have always uh I mean I'm for my tool I'm a very white hat SEO guy and always have been and you know to my own detriment in a lot of cases Um people might call me a fool for that but it you know when it works it works very well and it's it it always it stands the test of time where other where other tactics don't but you know there are there are um there are changes in search and retrieval there's there's changes to the uh the way that the an the answer and you know the final part of the process how people consume the the results. Um but ultimately they're still searching for a the answer to a problem which they were doing in the days of you know 10 blue links as everyone says and if it's some nasty nasty way of browsing the web which it was not. Um so ultimately it's still search. Um whether you you know you have to consider certain other things or not it's still search retrieval. still surfacing information. It's still connecting uh people that to a solution to a problem they have be that anformational one, a transactional one or or otherwise. Um and that's the way I see you know the I don't think we need new names. I don't think answer engines makes any sense. Um and geo is a blooming it's it's a silly acronym because of the fact that you know it means earth and it's a prefix Um and you even see the funny thing about it now is you see you may have seen these on LinkedIn you see these um like AI generated nano banana infographics about the difference between go is AEO and SEO and a lot of the time now because in pre-training like the cut the training cut off for some of the models is before geo polluted the web so in a lot of the infographics they create like GEO equals uh you know location specific search. So the so the guy's like and it always uses a a picture of a a satellite dish as well. But I always find that quite funny. You know the guy's like he obviously doesn't really look at the infographic. He writes the thing about how geo is a thing then the infographic like contradicts what he's saying. But you know like at the end of the day all it is is a you know names can't hurt you and it's just a name. But h I what can hurt is the some of the griffing behind it which I see a lot of to be honest and that's why I did the geogriph um people trying to sell re repackage the old things as new for one but also going back to do this add LLM text add schema this selling services that have no no impact or or effect And the other problem with it all is um is the problem of how you actually measure success as well, which is where I have a big problem. Um I don't believe in prompt tracking. Um I very vocally don't believe in prompt tracking. I think it's silly with probabilistic answers. Some people think there's an argument for you know a directional uh you know indicator which is probably probably correct that we have anything better maybe not but because we don't have something better doesn't mean that we we say that this is accurate because it's not accurate um it is just not accurate and the other problem is citations. So the citations don't if you're cited it doesn't mean you influence the answer um had an argument with um another well-known person about this uh recently he he said you know I can prove that influence is the answer and he shared a screen recording and it literally did not it was cited he had a citation you know the bit in chat GPV you click that no one does to view the sources and scroll down the big list that no one does and it was there like position 16 or something but it wasn't a citation. I I'd see the actual citation is the one within the answer itself. That's even then it doesn't necessarily influence the answer because of the way it actually works. But that's another story. But if you're a cit you're a citation within the answer then there's an argument you've influenced the answer. the people are tracking this you know I was I was in my visibility was 3% this month well you know what what did you get from it what business outcome was there what did your revenue go up I mean it's all very immeasurable and um the size of it still I mean I'm seeing some strange studies coming out again from a few people trying to um claim that it's larger than it is. But I mean, I think we can all see the referral traffic from MLMs and you know, it's um Google is still the predominant source of transactional business. Uh for sure. Did you see um I wanted to ask, did you see Mark Williams cook uh duck t-shirts schema experiment? Yeah, I did see that. Yeah. What do you think of that? Um, yeah, it's so he uh I think it it was parsed as text from from memory. It was an interesting like experiment. Yeah. He made up he made up schema. He made up schema that didn't exist. And and wrote and wrote things about the business in that in that madeup schema. It's not real schema. Yes. and and then and then when asking AI about the business, the business repeated what was in the what was in this madeup schema even though the schema isn't real schema which shows that it was parsed as text like you said. Yeah. Yeah. and and schema schema wouldn't suffice you know like so it's not the LLM that tokenizes it's the tokenizer then the LLM and schema wouldn't and other people have pointed this out you know schema wouldn't survive tokenization anyway so um it it seems people are spend people are spending a lot of energy for this argument about schema and um you you as you said I agree, but you should implement schema on your website where where it makes sense, but not for the reason that you're going to get more citations in LLM and you'll get to see your line go up a little bit because you know the line might go up. So that's so that's one myth it. So we have schemas as a myth for increasing AI visibility is it's like a must. We have llm.txt is another myth for increasing AI visibility. Uh how about um people who say you got to have all of you got to have like every part of your site in markdown. Well, I mean we've already covered that. They already get markdown. Um in part so I I covered this in depth. Um there the problem with markdown um LLMs get delivered to markdown. It's parsed to get um OpenAI pulls down a page um parses the page, turns it into markdown, feeds probably not the full page, the the relevant snippets into the LLM. The LLM spits it back. It's not one thing doing everything. There's a parser and it needs to go through a parser. Even even if it's delivered markdown, it has to go through a parser for security reasons. like uh when you're pulling in third party content which is what um you know fetching fetching is you can't it has to operate in zero trust you can't trust external content and we all like for for one thing with LLMs you've got prompt injection so it has to be parsed to strip out any dangerous potentially dangerous pro even if it's marked down it still has to go through parsing um is does Does it help OpenAI and uh others to get markdown in the first case? Yes, it does help them because they don't have to convert HTML to markdown and then part, you know. So, but I mean I I see no reason why we should like they they still they're still going to have to parse it because they cannot trust it. If they do trust it, then they've got bigger issues, but you know, that's another story. Um, a lot a lot of this is about making their lives easier. I don't think we should be doing that. The web has existed perfectly fine for a long time with HTML. if they want to parse stuff in markdown for their LLMs, they can parse stuff in Markdown for their LLM. Parsing is like trivial. Um what what we should be doing and what we should be doing SEO people should be doing is is making properly formed semantic HTML you know with with heads main nav etc. That makes parsing much easier. Um we should be doing that and that's if they're asking for things they should be asking us to do that. Um but the main and those are SEO best practices. Yes. Well exactly exactly. Yes. Yeah. Um the other the big problem with the markdown thing and the cloud flare and turn your page into markdown is that very quickly what happens is people start going well you know what in the markdown version um because you can see that the request was for markdown. So so you can serve a a different version of the page in markdown if markdown's requested. So you could say, well in this we'll just fit this little extra sentence for the LLMs and I know that worked well. Let's just put in this little extra paragraph and then let's just change the page completely and serve something completely different that humans don't see. And then we have this like two versions of the infinite two sources of truth for one for humans, one for LLMs. And um you know we're just back to cloaking. We're just back to things that have like many things at the moment. We're going back like 20 years to things that were done in the past and we're just revisiting them all again. Um, so I mean this is a long answer, so apologies for that. I don't think people should be turning their websites into markdown. I don't think they're going to see any particular benefits for doing that. Um, I think the AI companies, it would benefit them because it's going to save them a bit of computer and paring, but I don't think we should be going about making their lives easier considering they How about how about somebody who gives a cloaked version of Markdown? Are they going to see a benefit now? Right now, today probably the chat JPT. Yeah, probably. Will it last? Um I wouldn't have thought so. Uh Google I mean well you're not going to be able to send the code on Google anyway. So we are talking about GPT. Yeah GPT like I said is naive. It's just it's a very crude like so I how is how is how is chat GPT getting the markdown files to begin with? Because it's doing a it's doing a web search and these markdown files aren't made to rank or are they made to rank. So yeah what's happening there? So do you mean that the when when it when the website serves markdown the the user agent can request markdown content content format is markdown and then the server the origin server can choose to serve markdown in response. Um, but that's probably like I don't know less than 1% of websites. In most cases, it's OpenAI gets HTML. They parse it into markdown and they then forward that on to the to the LLM because the LLM is just a small part of the J. Chat GPT is an application. Chat GPT is software and the LLM is just a part of that software. Um, you know, it's not it's not what a lot of people believe believe that it is. You know, chat GPT isn't some um sugar going out the internet by itself. You know, you've got basically scraping and parsing feed it in. You've got um semantic scoring to extract chunks of chunks of text. Um the LLM is just just a part of the chain that spits out the um the text that goes to the user at the end end of the process. So for people what what would you tell uh I guess what would you tell people who are like well okay he's saying that right now cloaking markdown is going to help me? Well, I I mean many things will help you in the short term and many things always have helped you in the short term. If your business model and if your plans for your business are short term, then go for it. Like it's the same with I mean I don't know if we'll touch on but how is uh how are like um is Google seeing these markdown files? Google won't be anywhere near this. No, because Google us Well, that's so so is Google going to hit you for cloaking markdown? Who's going to hit you? Is it going to be the the AI companies? Well, I mean, Avenge, although it has to be, and that's that's what I'm saying. They're very naive. They're you can see it. You can see the results. They're they're shockingly bad. Like, they're so open to as Google to be honest, but that's a different story. But Cat GPT is 10 times worse. It's we know what influences it. We know if you go and write, you know, 20 listicles and put yourself number one, then you're going to show up in CH. It It's so crude. Everything is like again going back to the roll going it back to the two you know early 2000s or whatnot like it's all stuff we've gone through in the past and it's like the they they have the reason the reason they have the reason I'm very interested in this is be the reason I'm very interested in this be is because from what I've been seeing doing markdown doing like these texton markdown versions isn't really moving the needle in terms of AI visibility. Um, and so I haven't really seen a lot of these like, yeah, you can cloak your markdown files and that's going to that's going to really juice things in terms of AI visibility. So that's why I'm asking more about this. I haven't actually seen a lot of these case studies. Yeah, but if you knew what you were doing, if you knew what you were doing with it and you served some because going back to chat GBT and Open AI's naivity, it does like lean very heavily on like cosign similarity scoring. So you you could you can serve like really robotic unhuman readable version of the page that are probably going to hit the scores they get they just get retrieved. So people are probably just if you just s serve like markdown it's not going to make a difference. But if you serve cloaked a cloaked version of the content written in such a way that it would drive a human to insanity but for a for a semantic scorer it's going to hit the right notes and then an LLM an weirdly an LLM then humanize it the the robot will humanize the robotic text when it feeds it back to the user. So, um I mean are people doing this? Bridgel probably are doing this. Um Oh, so you haven't seen case studies? I've I've not seen you know I've seen people talking about it. I haven't seen specific case studies on it. You know it's I mean it'd be worth it'd be worth something someone. It's it's interesting. Uh, yeah, I was going deeper on that cuz I'm like, okay, is this a is this a good use of time for businesses and SEOs? And it sounds like the answer is Yeah, the answer is it sounds like it's not a good use of time for businesses and SEOs. Um, you Okay, so something that's very popular in SEO that there are clear results and there's clear before and afters. Um, so you warn that some geo service providers are spamming Reddit and turnurning out listicles at scale. It's Reddit and it's their sites. Could you talk about what these practices look like and then what harm does this pose to businesses and to the broader internet? Well, hello. Hello. We got here. So going back to I talked about my philosophy is very white hat you know um some people are not like you know many people are just you do what works and you well see my my philosophy is I'm like you should do what works but a but a lot of the black hats are playing in the churn and burn space that's number one number two is the biggest wins you get come from compounding a brand and compounding authority that's because then you can rank for the really competitive keywords in your niche that so the biggest the biggest wins come there and if you're playing the turn and burn game you don't get that compounding effect. No. Uh and and and then the these what works right now these are clearly getting hit like the listicles are clear the self-promotional listicles are clearly at at scale self-promotional listicles at scale clearly are not working. Now Reddit Reddit you can do Reddit you can do more sophisticated but Reddit can also get hit. I'm seeing brands who are messing it up. And that's the other thing. If you're a beginner and you're messing this, if you're a beginner and you're trying to do these black hat tactics and you mess up, the consequences could be really bad. You mess up with a money site that you've been developing for years. Guess what? Now you've gotten a penalty that it's going to be very hard to recover from. You're trying to spam Reddit because you've never done it before. Guess what? Your entire domain has just been blocked from from your favorite subreddit or just from Reddit as a whole. and this happens and you see it happen. That's that's my take on all of this. I I would go further than that as well and say that and and Smarty who I've known for a long time like she's well and smart is great and like she's very good on the show and um like one of the problems I mean I'm I am a Redditor. I'm not a Reddit spammer. I've been on Reddit for 15 I think 2009 or whatever it was. I've been since J originally I've been on Reddit and um Redditors can smell promotional smell self promotion smell brands and never mind like getting banned from Reddit and things. If you're going in and spamming where your customers are, if you're going and spamming where your potential audience or customers are, then that's going to stick in their head and you've just lost their trust. You've lost their respect. And there's many many cases of brands just killing themselves on Reddit. Um because it's not it's it's weird. It's like everyone just discovered Reddit in I mean Google started stuff in Reddit a lot more in 2021ish 2022ish. Um I think there was the argument at the time that a lot of people were appending Reddit, you know, site reddit.com to Google searches. Um, and then obviously with the rise of LLMs and Reddit being surfaced a lot and cited and it became more of a everyone was talking about it, but like I do wonder like Reddit's Reddit has long been one of the most popular websites on the internet. I I don't know. I don't have the current figure to hand, but it's it's probably been your top 10 for a long time. There was always the Reddit effect. You know, if you got if your content hit the front page of Reddit, previous to that was the dig effect and then it was the Reddit effect. You would get the Reddit hug of death and it would crash your server because you get so much traffic. Um, and it was like there's many there's a lot of cases like this, but it's like some people just discovered this hugely popular internet like four or five years ago. And you get all these uh LinkedIn posts saying, you know, uh, here's the quiet shift that people aren't talking about right now. If you're not on Reddit, your brand is invisible. Most people don't know about Reddit. Most people like it's one of the most popular websites on the internet. Um, and and while I say I was a I'm a Redditor and I'm not a Reddit spammer. I'm not a Reddit spammer, but I did like from time to time on my my personal Reddit accounts over the years share things that I thought would be popular of of my own or clients on subreddits. And this is going back 10 years like, you know, 2015, 2016. And in many cases, they brought in a lot of traffic and they brought in a lot of links. Um so like Reddit marketing and if you go back in reads um much of it is uh like you know because content gets refreshed much of the thing many of the things I wrote were a while ago and but they're all still available in archive.org or and there a lot of my articles um I talk about Reddit and you know marketing on Reddit and and you know get going into communities and subreddits and getting involved and being an actual active participant in the community. And as I say this is like this goes back to 2012 2013 I was writing this stuff because like community h going where your your target audience is is not a new thing for LLMs. What's new for LLMs is going to where your target audience is. Well, it's not new either because people did this for links on forums and whatnot, people seem to think they can just go in and drop a link or not a link, drop a bloom, uh my uh prompt tracker is the best prompt tracker that ever did track prompts and then go away. Um, and like you know it just gets but but yeah like going where your customers are is is just the way market this goes back to the geo thing and everything being repurposed. It's just marketing. Never mind SEO. It's just like you go where your audience is. you have to go to your audience and it's not a new thing like and yeah like so many people seem to have discovered it and so many people talk about it being the quiet shift that no one is talking about right now. I want to um I I had I had this thought which is that I know we are going to have in the comments some some people who are like well actually schema does make a difference for getting cited in LLMs lms.txt TXT does make a difference. These things do make a difference. And since just feeling that this is an inevitability, what would you say upfront to these people to get ahead of it? Well, I would I would say show me your like should show me your evidence on it. proved that it was the the I say going back to the correlation causation proved that it was this the schema that h and even citations are not even right but anyway let's we'll just say citations for the sake of it but you know in reality it should be actual business a business outcome the the other problem with this and and this is a long-winded answer that no one ever shares the prompts no one ever shares what the searches It's like we looked at another one of these tools that does these a lot. We looked at 100,000 prompts and articles. Articles, everything's an article appeared 14% of the time. How-to pages 13% of the time, but they don't say what percentage of the prompts were how-to queries. you know, if it maybe 13% of the prompts were how-to queries. They say product pages appeared 46% of the time, but they don't say the percentage of the prompts that were buy X or buy, you know, Y. Like everything's woolly and like they just data dump and expect you to accept that data. And, you know, I personally require like a reasonable amount of evidence for something. um there's no logical reason why it should like if it if it is as simple as you can go to your website and add schema then that's like that's a a joke like that's ridiculous where's you know like who wins how do how how does anyone win everyone just goes and adds schema and says it's listical but there's it has benefits there are you I use schema Everyone uses schema but not to make yourself visible more visible in LLM or you know I might be wrong like that's fine if someone wants to comment and tell me I'm wrong comment and tell me I'm wrong someone someone messaged me on LinkedIn two days ago won't give away their name but they said they messaged me two days ago he said yeah I won't give away the name was going to call you publicly on the geo stuff but that's whack of me lol so I took it down. But dude, at least debate me via DM. Why do you hold such a view? This was Friday morning at 6:59 a.m. uh Scotland time. Now, I uh I was, you know, happened to be sleeping at that particular moment. 9:19 a.m. on the Friday, I replied and said, uh which view about geo as a term something specific about AI search? and I've heard nothing. All right. So, here's I love a debate. Like, I love a debate. And if I'm wrong, I'm wrong. That's fine. Like, that's how we like, you know, that's how we move on. I mean, like I said, if if I'm wrong, all you need to all you need to do is add schema. Then we've got bigger problems. But, you know, that's a different that's a different story. But, yeah, this guy this guy prominent reasonably prominent guy. Oh, really? Yeah. Like, you would know his name. He's a geo guy said challenged me to like debate me and then kind of I'm so curious who I'm so curious who it was. Yeah. Like immediately said great. Let's do it. Share share the name. We'll get him on the show and then we can do the debate publicly. Can't share the name. Can't share. I know. I know. I know. Well, I mean it might go public anyway, so we'll see. We'll see. Yeah. Uh what are some what are some things that business owners and marketers should look for when hiring people in companies who say that they can enhance AI visibility? So how do people avoid the as you wrote you wrote the geo snake oil the geo snake oil how do people avoid that when when hiring people in companies who can enhance AI visibility? I think if they're selling it, I think the first thing would be if they're selling it as we can enhance AI visibility in isolation and then I think that you should run away from them. But you kind of but you can but you can you can you can do good SEO and that influences well. Yeah. Yeah. But that's what I mean. If they're selling it, if they're not being uh upfront and honest about the fact that most of is, you know, if they're saying we'll do this work that that is SEO will help to increase your AI visibility at the same time. But if they're just saying the there are many of these companies, we'll rank you number one. No one searches in Google anymore. Your customers don't go to Google anymore. You're you're invisible to AI. We can help you rank number one in chat. if that's their pitch and there are these agencies now coming up um then you should run away from that like you know if there there are there are I mean and I would also be running away from anyone that's like just saying scale your content write listicles unless you have very very shortterm business uh goals you know if you want to if you want rank and rank and tank rank and bank. You know, if you if you want to just rank for 6 months and then go away and into the sunset, then that's fine. But otherwise, you know, if you're a an established business and someone is coming along and suggesting you add, you know, well well a smaller like a, you know, a smaller site is let's say I don't know a thousand pages. If someone's coming along and suggesting you add 5,000 pages of skilled content to your website, then you should be running away. You'll see you'll see the line go up because indexing is not a problem. And this is another this is another thing that um a lot of people are there's a lot of these scaled um content case studies on a Twitter X can't really get used to calling it X um and people are celebrating getting their their pages indexed. You know they're saying I added 10,000 pages and they show the search console chart showing that 3,000 are now indexed. Well, indexing was never the goal. I mean, indexing the the goal wasn't even rankings. The goal was traffic and then the goal wasn't even traffic. The goal is conversions. Exactly. Exactly. But but indexing was way way down the the mix because Google I mean generally Google's going to sort of I mean if it's an established site that's been around for a while Google's going to index this stuff and then it's going to drop it out like you know it's a weird flex that a lot of people are doing on uh on X at the moment and you know they're all many people many of these people are like new to the industry a lot of people have come from. Yeah, that's true. There's a lot of sales people in the industry now. Um there's a lot of uh Yeah, they try to come on my show and I and I ignore them. Yeah. So, so I speak um I've always because you know I've been writing about stuff for a long time. I've been talking about this stuff for a long time. I speak very directly in like in language that um the layman can understand or or try to you know I may have failed at it time to time there's a lot of corporate marketing speak I watched a um podcast or a skimmed podcast uh or a webinar from one of again one of the leading EEO companies um and it had the guest on the show was one of the names. He's in a lot of listicles actually of his own making. So if you go and ask who are the top AI search experts right now, he'll be in that list and you'll know his name. And I watched I skimmed like 10 minutes of it, right? and his sector there was like can't be too specific on it but there was like a certain amount of tips for um the top the top tactic or the most advanced tactic for uh increasing your AI search visibility and they were they were not advanced they were basic tactics but he dressed it up in such absolute they're talking to CFOs everyone's talking to CFOs in enterprise Uh they they know how to talk boardroom stuff, but they don't know how to talk directly to CFOs or CMOs. Uh C well either, you know, financial market. Well, I mean, there you go. I don't care is the answer to that. You're right. Probably the CMO. I don't care. I don't use I don't use these acronyms, you know, other than when I'm evaluating some stocks, maybe to lose some money on some options, but that's a different story. But you know that's that's not marketing language. That's like that's boardroom. Like there's a lot of these people around now. They're not they're not marketers. They're not technical people. They're not um they're not the nerds like like us, you know? They're They're the boardroom. Everyone's talking. And it's the problem with content now as well. Everyone's collapsed into this dull gray monotonous same uh claudisms everywhere ship lotisms landscapes. Uh all right. So so David David if someone listening runs a business what are three specific things they should do this week to increase their chances of being cited in AI answers? Three specific things they should do this week. I mean that's a tough question. It's a tough question. Yeah, it's a tough question because I I could answer it in five seconds. You answer right. You answer in five seconds then I'll All right. Okay. All right. Have uh use purchase intent. Find purchase intent keywords. Put those in your page titles, URL slugs, H1s, and beginning of the first sentence. That's number one. Okay. Right. What? Number one. They've probably already done that, so carry on. Number two is is build links because then you're going to rank better. You're going to get discovered more and your relevance. That's just that's just SEO. Carry on. No, I I know. But these are three these are specific things that basic stuff like so I don't That's what I'm saying. That's what I'm saying. I'm going to I'm going to I'm going to have an out a little bit. Right. Yeah. All right. Let's do it. Because I could have done I could have said the basic stuff, right? But that's what I thought you were going to say. You were trying. That's what I thought you were going to say. But I don't want to say that stuff. What's the point? Everyone knows that. Everyone knows that. No, that's not true. That's not true. That's why if that was if that was the case, you wouldn't have these CMOs falling for the geogrift. Well, I mean, that's that's that's a different story. I I've written this stuff years ago, right? So, I don't repeat myself. I hate repeating myself. You know, this commodity content thing, right? Commodity content. Oh, yeah. Yeah. Right. That's how I That's why I stopped writing about search for a while. I came started writing again because I had a lot of things to say again because things got interesting. I said all this stuff like 10 15 years ago and like but but David you're a marketer. David you're you're a marketer. You know that people need to hear the same thing multiple times before they take action. But but they've heard it 100 thousand times. If they haven't if it hasn't sunk 100,000 in one well it's not going to sink in if I say it for theund one,000 time. Like honestly, head of content at Hrefs. Head of content and HFS here. So I'm not saying like I disagree. Like they're good. It's good advice. I agree with you. But like it's basic stuff like Yeah. Yeah. Yeah. Oh yeah. For sure. For sure. For sure. So, so I like I mean I basic stuff but it's the but that's the stuff that will get the most results literally that's going to get the like I guess I guess uh the other thing might be okay find all the listicles that are being cited the most and then offer to pay the authors to get your brand in a high position. That's the other I'll give you this one but I'll tell you what the way it should be done and the way I did it for years before LLMs please. The first thing I did when I took on an e-commerce client and I've always done was go and find and and it's all spammed to death and destroyed now, but I would go and find affiliate sites that had these top 10 listicles back this is like 10 years ago or you know before LLMs and I would because most e-commerce sites have an affiliate program and I would go out and just make sure you were included on these these listicles because and you know it wasn't to do with getting cited LLM. It was just this this you want to make money, you want to give sales. Exactly. Yeah. So I would I would I mean the problem is now now it has been so spammed and like I'm almost like if if I told someone if an e-commerce client came to me and I said to them, "Oh, what we want to do is we want to go and find the the lists of the uh top 10 X for Y and want to make sure you included on that." They think I was a spammer, but I wasn't. You know, I'm not crazy. You want to be you want to be in these list. This is normal PR. This is normal public rel relations. Ex. Exactly. But the point was they were they were real listicles or they were real real again because obviously I mean I I had many affiliate sites over the years and you know were they real maybe not but you know they were certainly realer than um the brand writing these things themselves. but like it's like it's like people just discovered one stack in the past, but people I actually saw a webinar where someone was like one of the biggest things about LLMs and AI search visibility is affiliates. you want to be this is this is and they were talking like it was brand new as if affiliate marketing was a new thing like you know like now you want to be reaching out to affiliates and um trying to make sure your your your brand is included on the lists but like I I just look at this stuff and like why were you not doing that for the last decade why were you not doing that for the last 15 years what were you doing honestly sorry I'm ranting on a bit here but like a lot of the time I look at this stuff and I think what were you Well, what did you do? What did you actually do? Because I was doing all this stuff for years and I was talking about this stuff for years. And in my opinion was doing this, but it turns out probably people weren't. Um I think I think I I think what's happened over the last few years has actually exposed a lot of the a lot of our industry to be honest for um for the laziness and the Yeah. appliances commodities basically you know get them in out you know now now it's outsourced to AI but for a long time it was outsourced to you know wherever uh take their cut maybe the chart was up sometimes charts go up just because of time you know because of nothing you did sometimes charts just go up if you don't do anything stupid take the money move on with your life you know like and people talking about PR PR and like I've always been a PR type link builder. That's that's been my approach. I've always gone for I've never been a huge one on I build links to pages and whatnot to an extent, but I would always go for branded PR campaign. Those are those are those are my favorite types too because th links that that drive qualified referral traffic. Referral traffic and qualified referral traffic. Those are the best links. Exactly. Yep. Yep. And they matter for LLMs. Yes. No, abs. Absolutely. Absolutely. And that's like it seems to be people have not people have very much not been doing this and like some major people with prominent voices. Um because again they treat it like it's a new thing. It's a new thing for GE. By the way, I I I want to ask you, there's this um you said there's this person who is doing really well if you search if you if you prompt uh who are the top like AI search experts and then there's this person who will come up all the time, but this person is actually a grifter. So my question is I wouldn't say he's a grifter is too strong for this. Okay, cuz I I wanted to ask I wanted to ask if this person is is so full of if this person is like a BSER, how come this person is getting named the number one AI search person? because they're a spammer. Because they're spamming. You can spam. If you spam, then spam works sometime. Yeah, you could go and write some of them to do that. The question is So, okay. So, then how long do you think the listical strategy will work for? And will there come a point where like people like this are they're they're going to fall out because these listicles aren't going to rank anymore or do you think like it's a actually there's there could be some longevity to the strategy? I I think there's I mean again it goes back to if there's longevity to the strategy then we have problems. The web is destroyed you know. Um I I've looked at a lot of these websites like I've got a couple of tools I could my own that I use to look at crawl a website and then look at the percentage of listicles and whatnot self grandizing listicles in particular and this this guy I'm talking about um 20% of his website is listicles ranking himself at number one. I'm gonna do this right now on on ChachiPT. I want to Who are the top AI search experts? I'm not Well, I'm not going to say the name. I want to just see if I want to see if uh I'm like, "Oh, yeah. Okay, I recognize this." No, no, it's it's saying people like Sam Alman and Sundar. right. Well, they're probably Yeah. I'm not getting um I'm not getting the SEO people. I literally my prompt was who are the top AI search experts, right? I mean, you know, it's a it's probably a reasonable answer if you actually think about it. They they actually I don't think I got any. Although Sam Alton's probably not a good example because I don't think he knows what day of the week it is, never mind how it's actually operating. Top AI search experts for getting cited in CHA GPT or LLM. I think that's a better it's a better prompt to use. Let's see if we get like SEO people now. Okay. Uh okay. Oh my god. This one person who is called the founder of geo as a discipline. The founder of gio. Yeah. The founder. Yeah. CEO of I think the Greeks are the founder of geo. Oh, is it somebody? Oh, wow. Some of these people have been on the Yeah, a bunch of these Lily Ray is listed here. That's cool. Um there I mean there's lots of there's lots of legit people here. To be honest, I I I don't I'm wondering who the the spammy person is cuz there's so many legit names who are here. Lily is here. Lea's here. Oh. Oh, okay. Wait, I think I know who you're talking about. Yeah, I think I know who you're talking about. All right. I'm not going to because every You're right. A lot of people a lot of people do know this person. Okay. Yeah. But I mean I mean the point is it's not like the whole point of search anyway. I always said like for a long time with SEO I always said you know the main thing is if you want to rank number one be number one be the best be the best product be the best service be the best uh yeah but that's but that's yo but that's the flaw but that's the the flaw of product people is that like if you're making something great you still have to do marketing for it. You still have to get people to find out about it. Even if you are the best you still it's it's uh you got you got to tell people about it. Yeah. But but but the starting point for that all of that becomes easier if you just you have the best product. Everything's harder if you're not you know if you're just going out because the problem is everything collapses into the you can go and write 10 listicles saying you're the best AI search expert. I can go and write 10 listicles saying I'm the best AI search expert. Joe Blogs from down the street can go and write 10 articles saying he's the best LLM. So let me let me let me ask you let me ask you this. If if a if a client came to you and they're like, "Okay, I want to put up uh I want to put up 10 listicles or 20 30 listicles saying that I am the the we are the best so and so." Mhm. Um what would you tell them and and what would your reasoning be? I would tell them probably not to do it. I would and why I would I would tell them to do it depends on the size of their website as well and it depends what their website is you know if they're a like if they're a big hu you know if they're a huge site with tens of thousands of pages then they can probably add 50 listicles and nothing's going to happen but if they're a smaller business with 500 pages and they add the same 50 listicles then they're going to have a problem. You think a large site can can still get away with you think a large site Oh, go ahead. I hate the term list to be honest like listicles to me or BuzzFeed stuff from the you know the like 2010 or whatever. I I had many like I ran for for a while like I was doing like affiliate marketing on affiliate sites and I would you know 10 best this 10 10 bit best that like it's just content. It's just an article. just a format of an article. You know, I would do them in a in a way that think is is a little bit better. Now, everyone just gets clawed to do a paragraph and then five bullet points. Number two, paragraph, five bullet points. Number three, paragraph, five bullet points. Add like, you know, the stuff at the top, some FAQs at the bottom. They're all the same. Every single one of them is exactly the same. Um I I used to do so I would do like if it was an e-commerce business like you're selling like red shoes then you probably do want like the 10 breast red shoe retailers or something. It's just an article like it's you maybe do want that and you want to put forward a strong strong argument about why you are the best. Um do you think um that's true a strong argument. Do you think a large site they can really put up 50 self-promotion listicles in a short amount of time and be fine? It depends on the size of the site. Like it really does depend on I mean there are there are large sites that have done it. I've got there. So one large I mean it's not even a huge amount of pages. Well, they do it and then like a lot of them get mount AI as well. Well, there's one there's one that's got about 4,000 index pages and it's got 220 self-promotional listicles and they've been doing well for about two years. Oh, wow. So, there we go. Like it it's the everyone seems to work for them. They seem to be the case study for everyone. They have two they've literally 220 self. So, 5% of the site is self promotional. Why do you think why do you think uh it's working for them even though everybody because of domain authority? the main authority master everything. It's the same as you can get away with things. Yeah, but but how about but how about Shopify? Shopify put up put up t tons of these self-promotional listicles as well and they it sometimes catches up on you but it it can mass things for a long time. Um people are following the strategies of enterprise businesses and um that's true. Uh like and enterprise businesses have high authority and they can get away with anything for Yeah, exactly. It does not work for 99% of websites, right? Um, yeah, Shopify. Like there there are some high-profile sites that got hit. There are many clients of some of these tools that have that have been hit. Um, but this particular site seems to be still churning along quite nicely. But I don't mean to say that in a couple of weeks time, a month's time, six months time, a year's time, all of a sudden it just doesn't. Two years is a good two years is a good stretch. Two years is a is a is a pretty solid stretch. It it is for most businesses, but it's probably not for this businesses because you got VCs with a lot of money that want to return on that money and they're putting a lot of money into this business and they ain't going to see that return 5 to 10 years. So, if that business is not around in three years because of what they've been doing, then that's not a long stretch. That's true. I'm surprised that that um normally when when businesses that are doing spam techniques go super they get a lot of attention, Google steps in. I I mean some of us have been talking about this as well. Like I think Google's been sleeping the wheel for I mean I had a big Google and I've had people come on the show and say that too. 2021 like I've had a big rat about Google. Like Google's Google's Google. I mean at the end of the day they have their own priorities and they have their own priorities are making revenue and their ads you know which they seem to be doing quite well on judging by the last quarter although some of that is uh a little bit disguised but that's a different story but yeah like Google's priority ultimately is not to satisfy the user it is not it's to make ad revenue for their you know and they which is why they seem to take a long time I think now longer and longer the m when Pat stopped being the sort of public face of Google spam team and whatnot. Um, it was a big mess. It was a big sort of loss for the industry, I think. Um, you think Google doesn't care as doesn't care a ton about it was a big loss. Do you think Google doesn't care about satisfying the user? I would make the argument that they care a lot, especially now there's so much competition. I don't think I don't think they satisfied us though. I don't know. You can you can look at the search results and see that they don't they have disrespect for the user. Well, yeah, but but like Reddit I think Reddit is a good example. It's like what we talked about because like how many people have you met in real life who are like, "Yeah, I I basically append Reddit to all of my searches because I'm dissatisfied with like what we're seeing with SEO and I want real I want like real answers, user generated content." And so what has Google done? they've served a lot more user generated content, not just Reddit, but all types of UGC. And I would make the argument it's like Google is looking at the trends of what searchers want. I would make the argument they could go once access to Reddit data to train their LRM. They just look at the search results. They get worse and worse. It's all designed to track the user. A lot of people do agree with you and and I see that in the comments of this podcast and I see that I see that on comments of search engine roundt posts that I share. A lot of people also agree with you. I find it I would find it hard to like AI overviews for example. AI overviews you know the pool takes span and it takes you out the search results and you can't just like that that is disrespectful to the user. Why do they do that? Why is that done? That's done to like change behavior away from Yeah. Like that's just it's not it's it's li it's being lying strong. But it's maybe maybe or maybe they maybe Yeah. But maybe they have the data to back up that actual that actually searchers. They don't because you can see it all the time. Like it's not like I mean or or it's just an experiment and if it doesn't work then they'll stop doing it. But it but it it is manipulative and it's um misinforming. So it's you know it's a you you've we all use accordians on websites. We all use expand to see more. Use the expected behavior of expand to see more is it expands in line. It doesn't take you away from where you are and block there is no reason for that to block you away from the organic results other than to make to take those organic results away and make you ask more questions and stay on for longer and then ultimately get served more more ads. Um, I mean like Google I mean Google's the lesser of is it lesser? I don't know these days, but like I I'm a big Google champion in terms of Google is where the traffic is. Google is still what matters. Google will win the AI race because they have the comput, they have the the brain power, they have the access to people, they have the data, they have everything. and they have the capital, they have the TPUs, they have, you know, there are so many reasons they they invented the transformer, you know, like there there's so many reasons, but yeah, um like so I'm a big champion in terms of like when people say, "Oh, you don't need to worry about Google, you need to worry about chat GT now." Um, but I'm no good old fanboy these days and um, you know, I I think what they're doing is very very very very disrespectful to the user and you cannot find you cannot find in search now what you used you were looking for and the search has got worse like I used to you used to be able to I'm saying I that's an anecdote but I think generally you know people you used to be able to find what you're looking for now Google shows you what it wants you to find. I don't know. Oh, see I will just I I would disagree actually. I think I used to have a much harder time finding what I was looking for. Like I would I would do a search and then I would get these like super long articles with a ton of fluff at the beginning and then when I found then when I found the section of the article that talked about what I was looking for, it wouldn't even give a direct answer. It wouldn't give a great answer and I would leave dissatisfied having wasted time searching through this article trying to find what I'm looking for. And uh and that was my that's been my experience with with search for a very long time. And I actually feel like it has gotten better now. Instead, I get a 30-se secondond YouTube short. There's a much better chance that that YouTube short is going to tell me what I want. And even if it doesn't, it's it's more engaging and it's a lot it's a lot faster. And my but and my experience has been these like long skyscraper fluff pieces that have wasted my time and so many others. Yeah. I mean, you could be, you know, you could be right. I'm not this your opinion. It's my opinion. You know, we David, David, I will admit, I will admit that I am the unpopular voice on this because lots of people, lots of listeners of this podcast, they do not like they do not share or like my opinion on that. Like, yeah, I think search results like Google has made a genuine effort to try to make search results better. A lot of a lot of people, a lot of listeners, SEOs, the community strongly disagrees with me on this one, and I'm well aware of that. the you YouTube shorts like at the end of the day YouTube is Google Google's property and Google owns YouTube obviously it's in its interest to show YouTube um talk sometimes sometimes they recommend Tik Tok and it sucks I'm not like I like I enjoy like maybe I mean I enjoy reading depth guides and you know and whatot some people don't some people like digestible bite-sized content which the YouTube shorts are um it depends what you're searching ultimately as well if anything. You used to be able to find everything and anything on Google. It used to be the the Yeah. the modern library of Alexand Alexandria, the the gateway to the world's information. It's not that now. It's a gateway to a slice of the web, slice of the internet. Yeah. And I mean, you know, it's my opinion, but it's opinion shared by many. But I I look at search results every day, obviously. Well, and I'll say I'll say that it's become I think it's become more censored. There's there's like some things that some people want that you just can't find anymore. And you used to be able to find that 15 15 years ago very easily and that can be very frustrating. You can't go on like a rabbit hole anymore. You know, you used to be able to go on a like a you know a particular topic or something you wanted to dive deep into and you go to page 165 of the results. and just you get 14 pages of result and then nothing. So, okay, I want to I want to ask you because you said um you said that prompt tracking is like you think it's kind of a waste of time. So, how should businesses actually measure success in AI search? So, I don't have the answer to that question. Um and I have cost myself like prompt tracking is the easiest thing like prompt tracking you could do you could go and set up a prompt tracking thing you know it's literally just scraping you can use bright data to scrape chat BPT uh use an LLM to extract the prompts uh sorry the brands from the answer track it over um you know it it's so basic like anyone can and that's why we have like 150 million prompt tracking businesses popping up um so many of them. So I like I have a like platform for things that no other tool seems to do or think about. And I specifically don't Yeah. Yeah. And I specifically don't do prompt tracking. And people ask me, do you do you have a prompt tracker? I say no because prompt tracking's like for the birds. Prompt tracking wouldn't work. Uh and like I could go and add a prompt tracker like I had a prompt tracker 3 months ago like and it would have given a pretty chart, right? It was given a pretty chart, but like it doesn't it doesn't there's no action. A good quote I heard recently was if you stopped tracking your AI visibility tomorrow, nothing would change about your business. Yeah, people are track people are just looking at these lines in these dashboards. They just don't tell you anything. I agree with that. So, how do you measure it? Uh I generally sort of I do look at I do look at the answers to prompts. Obviously, I do look at them. Um, but like the core main topics and services and whatnot, not not for a business. I don't obsessively track them because they're probabilistic. Um, I I have a I know how to get into them. I have a process that I use that um gets you into AI overviews and AI mode like pretty much guaranteed that I use. I've spoke about it a lot. I mean, people don't seem to be following it for some reason, but I um I I sort of two bigger bigger articles that were like um you know, I wrote the Chat GT one which I think um you know, it went quite it was quite widely shared. Then I wrote a follow-up in February or something which is more or even January which is more focused on AI mode and it sort of slipped under the radar. Um, if you have a service business and if you're ranking top 10 to top 20 and you follow this process, Google's going to show you and it's AI mode and AI overview. What's the process? What's the process? What's the process? We got to know I've written about it. It's there's 5,000 words you can read about it or you can watch a 19minute video. If you It's 5,000 words. What's it What's the TLDDR for the listeners? I'm not telling you. Go and watch the video. Oh my gosh. I I've written about it so many times. like it's there. H I've got there's a video where I walk through the process of doing it. Um and and there's a 5,000word blog post which explains why it works. And it works because of the maths. But you don't need to worry about the maths. You just It works because of what? Because of the maths or the math. The maths or American American press the math. Yeah. Like I've written it I've told I'm not going to like talk like I've talked about it many times and like people can choose to ignore it and they they have but I'm using it to rank service businesses like straight into the what's the name of the what's the name of the article? Uh the busting the geojargon busting the geo jargon why you're probably over complicating AI search. Yeah, that's so that there right at this moment in time if you follow that process and like you know if you want SEO has always been about reverse engineering. Oh man, you're you are the path of least resistance. You're short circuit the fan out. You short circuit the fan out. Okay, interesting. So, so honestly this was like my chat GPT article got very widely shared like a lot of citations and whatnot and I mean citations by humans not by not by chat GPT and it still gets a lot of traffic to the website like it still brings in traffic like organically socially but that article there is is actually if you want to if you want to like do something going back to what you said what should you do today to show up in more answers go and do this because it worked. um it might not work tomorrow but at the moment it works because of the way that AI overviews work and the way query fan out works and the way the maths works behind the scenes but yeah I've I've talked about it a lot like I've I've done it in many I say a lot I don't like I'm not I'm the I'm I like to think I'm a good marketer but I'm a terrible salesman and I hate like I don't I I feel if I share something once or twice I don't want to share a third time because I feel like I'm being spammy, which I'm not ultimately. But I've talked about this like a lot and it's out there and if you know people can choose to ignore it or test it and and like I'm also if someone thinks I'm wrong, they can tell me I'm wrong. But like I've done this quite a lot of times now and it and it works. You have to you have to rank. You have to rank top 10, top 20. But and yeah, and it also helps if you know what you're doing with SEO, you know, if you know how to optimize a page generally to boost its rankings and then you do these specific things, then because of the maths that's going on behind because AI mode and AI overview, the way it works, it generates an answer, an expected answer or parts of answers. Then that's where the query fan outs go. they pull in the uh bits of results and then basically it's after the fact citation. So it's basically just matching the bits that support. It's like the LLM knows the answer. It doesn't know anything but we'll just say it knows the answer. And um it's like yeah that that backs up what I say. So that's the citation. That's what I'll use in the answer. Um and you can change but you can change this is where going back to this influence in the answer. you can influence that answer if you know what the things are that it's looking for and you do that by probing rather than looking at the if you look at best x for y you don't look at best x for y you look at the you look you're looking at the things that it's looking for the things that it learned in its pre-training when when the llm's basically learn concepts words and semantics you know things that are associated with specific types of service um or information and that's where it's knowledge is formed. So you want to try and get to know what that part of semantic spaces that you're trying to target um or vector space and then you use these words without thinking about the math and then the maths aligns and you become the answer. Does it work 100% of the time? doesn't work 100% of the time, but it's it works a lot of the time and it works a lot better a lot of the time than adding llm text or schema. And it is not spam. It's not spam. It's just surfacing the things that you It's being honest about your business and surfacing those honest things in a way that's extractable by LLMs. Um, and I don't want I don't want to wrap that in. It's not LLMs can parse human text perfectly fine. It's just making them prominent and making that the bit that's retrieved that then matches the answer. That's but it's just I mean I see that as SEO. It's not like um it's just a a new thing in SEO. It's not some new geo thing. And weirdly, it's more effective than any geo tactic that exists other than spam. So, I mean, this is what you can do without having to write 50 million listicles. Um, and…

Transcript truncated. Watch the full video for the complete content.