Only 2% of AI citations in the IT services vertical point to brand-owned websites, based on a study of 107 million AI answers across ChatGPT, Perplexity, Gemini and Google AI Overviews. In this episode of The GEO Show, Paris Childress and Simon Young of Question.Marketing argue that the fix is to stop publishing your best original research on your own domain and start pitching it to third-party sites as free, exclusive, high information gain content. They also cover why "add us to your listicle" outreach caps out at a 1 to 2% reply rate, and why brand mention requests convert about five times better than backlink requests.
Publishing your best original research on your own domain may be the least efficient thing you can do with it. A study of 107 million AI answers in the IT services vertical found brand-owned websites accounted for just 2% of citations. Community and UGC sources took 51%. Independent editorial sites took 47%. The content strategy most brands are running, more posts on more topics on their own blog, is aimed squarely at the 2%.
Paris Childress and Simon Young, founder of Question.Marketing, spend this episode working out what to do instead. Their answer is not more outreach volume. It is a change in what gets offered. Stop asking editors for a favor, start handing them original research they cannot get anywhere else, with exclusivity attached. Information gain becomes the currency, and the citation still comes back to you when the link does not.
Because it asks someone to reopen work they have already finished. Paris is direct about the numbers: SiteForge outreach is highly personalized and one to one, and still lands a 1 to 2% reply rate. Ten emails a day usually produces no reply at all, and one or two replies in a week. That buys a place in the occasional listicle, not a citation base.
Simon's diagnosis is that the ask is wrong rather than the volume, and Paris agrees that most people do not want to revise a piece already shipped. Sites where competitors are already cited are also the hardest targets, because you are trying to squeeze in beside an article that is already trusted.
On this episode, yes, and probably first. Paris poses the choice directly: if GEOforge data produced a genuine breakthrough finding, does it go on the GEOforge blog or to Search Engine Land? Simon's answer is to let the authority site run it, then reference it afterwards on your own domain. Paris goes further and suggests seven days of exclusivity before any follow-up, including your own site and social posts.
The mechanism is information gain as a value exchange rather than a swap. The pitch is not "link to us and we will link to you". It is a free research piece the target site publishes as its own story, with your brand named as the source. The publisher gets the citation to their domain. You get named inside it. If ContentForge rates a piece 96 out of 100, that number goes in the email, alongside the note that only one a month scores that high.
Mentions, on the evidence discussed here, because they are cheaper to acquire and worth the same. Answering a live question from Ashley Pearson about where to put the effort, Paris makes the case plainly:
So if my success rate for a link outreach is 1% and my success rate for a brand mention is 5% and they both have the same value in a GEO first strategy, then I'm gonna do mention requests. I'll stop doing link requests.
His reasoning is that AI does not look up domain rating and does not check for a do-follow versus no-follow attribute. What it reads is topical relevance: where you are mentioned, and which other brands are mentioned alongside you. Simon is more cautious, says he is running a live test on another client, and argues that randomly built backlinks carry real demotion risk. A new brand still needs the full mix of social proof, reviews, references on trust sites and third-party recommendation.
Deliverability and data quality, before strategy. Simon walks through what he found in a live SiteForge account:
Volume matters because a single pitch rarely lands. Simon describes the pattern: the first exclusive gets ignored, the second registers as value, and by the third or fourth the editor replies asking whether you want to write something or come on their podcast.
Because crawler identity is trivially forgeable. Hop AI detected a bot on a client's Cloudflare setup using the exact same name as OpenAI's real crawler, from a different IP address. Most sites are not watching their crawler logs, and the ones that are see a familiar name and wave it through. Paris's conclusion is that technical GEO now includes policing which bots reach your site and selectively blacklisting the bad ones, with bot traffic having crossed half of all traffic on the web.
The episode closes with Paris rethinking the module he calls the hardest one to get right. AI visibility reporting is table stakes, and knowledge-grounded content generation is where GEOforge separates itself, but off-site citation building is where the visibility actually moves.
Paris: Hey there, Simon. Happy Friday. How's everything?
Simon: Busy. Been working hard on a couple of clients and it's getting very exciting. I did start making a video this week about AEO and got an hour and a half into it and decided... Well, I didn't decide. I had to jump on another meeting, but I'm probably only halfway through that video at the moment. So I know you've been making The GEO Show, and I'm trying to make a lot of content on the AEO side, so we should have it pretty much covered off between the two of us.
Paris: Yeah, I think on both sides we're making some big strides with, in particular, the AI-assisted workflows for content creation and distribution. I'm especially excited about this new Claude skill that I created. It takes my transcript from Descript and immediately gives me a YouTube thumbnail, title, description, recommended hashtags. For the podcast, it gives me title, show notes, a square image for that. It needs to be square. And then for LinkedIn, it gives me a LinkedIn article for my newsletter with all the elements that I need, the title, the body. And they're all a little bit different, so it doesn't just look like I'm copy-pasting everything all the way across. Everything is a little bit tweaked. It creates a blog post for my blog, and then it can publish straight to my blog through Webflow. And I think you've discovered this now too, where I almost never go into Webflow anymore at all, but I'm still publishing tons of stuff there, because I do it all through Claude.
Simon: Yeah, obviously I hooked Claude up to GEOforge, and that then talks to you, it talks to Lovable, it talks to Webflow. It's all over the place. So I actually ordered a pocket device, so that should be here as well this week. So we'll have a play, see what that can do, see what skills you can build into that, see how much of my life it changes.
Simon: I know we debate GEO, AEO, SEO, everything, right? But looking back on the video I've just made, I really do believe there's an awful lot to do with voice search coming in our general lives. I was talking about this years ago and we were trying to talk to Alexa and get it to give answers. And the problem was that the adoption of the device technology, fantastic, but in terms of people using it for what it was meant to be used for, hard to get people to do it. But now we're in a world where very, very quickly we are gonna talk to the devices, and that's where the difference will come.
Paris: Yeah. I can't wait for the secretive OpenAI device that's coming, the one that Jony Ive is working on.
Simon: What one's that?
Paris: Well, it's supposed to be something about the size of a hockey puck. And there have been a few leaks, and supposedly it's gonna be a small, round device that you can even put in your pocket. And you just take it with you, you drop it on the table wherever you are, and I think it can listen, it can see its surroundings. It has Bluetooth. I don't know if it's Bluetooth, but it'll connect to speakers so that it can speak to you.
Paris: And it's gonna entirely be voice driven, and it's not gonna be, "Hey, Alexa, play me my favorite song," or, "Set a timer." It's gonna be something much more powerful where it'll proactively say, "Hey, Simon, just a reminder, you gotta pick up the kid from soccer practice, and there's heavy traffic. You better get going," or something like that. It's gonna be some kind of crazy phenomenal assistant that is just like a personal assistant that's there wherever you go, and it can be reactive, it can be proactive. At the end of the day, you can just say, "Hey, that was quite a day, huh? So tell me, what do you think I should prioritize tomorrow based on everything that you heard today through all of my meetings and my whole life?"
Simon: The nerd in me has just imagined a scenario where, you talked about being in the pub, there's four or five of us sat there, we've just had a game of golf or something, and then we all just chuck our device on the table, and then the devices start talking to each other.
Paris: You know, I actually have seen this now happening with GrokBot, because I have about seven different agents in GrokBot that are with different titles, different roles. And I'll give a task to my chief of staff, and then it figures out how to delegate that task down to the team. And sometimes that task might require more than one of the agents to work, and then they create a chat together, and then I can see them talking to each other saying, "Okay, I'll handle this part and you handle that part. When you're done, you pass it to me." And they're chatting with each other, and I can just watch that. And I can also watch their computers as they do the work.
Simon: And what nefarious means will some of these people use these bots for? Do you know what I mean? I think you and I both know if we were not so honest, there are some means that people could really get stuck into using this tech for. So hopefully a lot of the people that are running it have put those guardrails or safeguards in place.
Paris: I don't know if I told you this, Simon, but we have detected, from one of our clients in Cloudflare, we detected a fake OpenAI bot that uses the exact same name as OpenAI. I think it's called the OpenAI User Bot or something like that. But it's the exact same name as OpenAI's main bot that it uses to call for inference. But it was from a different IP address, so it was some kind of malicious... It was a bot that was there to do bad things, or to try to discover some vulnerabilities maybe in the website, or data that it could extract. Who knows?
Paris: But I think that part of GEO and part of the technical part of GEO is already now becoming the policing of the bots that are visiting and crawling your website, and selective blacklisting of bad bots. Because it's so easy now to just create a bot, name it whatever you want. You can impersonate GoogleBot, you can impersonate OpenAI UserBot, and then have it go out. And of course, as GrokBot shows and now everybody has this, I think 24/7 always on computer use. So I can say, "Hey, I'm trying to collect data in this healthcare vertical, and here are 500 different websites I want you to go through and try to pull out as much patient data as you can from every single one of them. And I'm gonna call you GoogleBot." And most websites have no clue. I mean, they're not watching the bots that are coming and crawling. And even if they do, they see, oh, GoogleBot. Okay, well that seems fine.
Simon: Well, this is one of the biggest problems that we've had. We run a lot of Google Ads for clients, and as we know, a great big chunk of the spend in those ads is spent on bots. You just cannot stop bots coming, clicking ads and stealing spend. It's been the same since when the casino industry really took off online in the 90s. People were creating so many money sites that they would send traffic to ads to bring people onto a site to click an ad, and then eventually they figured out, "Well, I don't need to do that. I can get a bot to go and click on the ad." And that takes half of the advertiser spend from Google's pocket. Of course, Google didn't mind because the advertiser was paying anyway. And then a load of tools sprung up to try and stop that, but it still goes on horrendously. I think there was a stat that I read, you might be able to correct me, but I think about more than 50% now of the traffic on the whole of the internet is bots.
Paris: Yeah, that's been confirmed, that that number has crossed the 50%. And I remember, this is many years ago, the UK was the first country to have digital advertising spend surpassing all other offline channels for the first time. I think it happened somewhere in the early 2000s. It might have been around 2010 or something, or maybe a couple of years later. And everybody knew that this is never going back. And I think it's the same situation here. AI bot traffic is only gonna get bigger and bigger.
Paris: It almost makes me think that the future of the web is a web for AI agents, and humans are now gonna interact with a layer that sits between themselves and the internet and all these websites. My AI is gonna go out with certain tasks like, "Hey, go out and buy flowers for my wife, and then have them delivered at this time and place." And it's gonna arrive at the flower shop, and my agent's gonna interact with that flower shop's agent. And there's gonna be two layers, actually. Every website is gonna have their agent, which is gonna be designed to interact with the visiting agents that act on behalf of their human overlords. And it's all agent to agent. So the web is quickly just becoming a big database that needs to be agent founded.
Simon: Well, are we the overlords? That's the question. Are we in control? Because I don't know whether you saw one of my LinkedIn posts this week. I'm gonna see whether you liked it at all.
Paris: I've been trying to give you love on all your recent stuff.
Simon: No, it's not a criticism. I asked AI, "Do you think that robots will eventually control or destroy the human race? Stephen Hawking did. He was pretty smart. What do you think?" So I asked it that. And within that response, it said, "Concentration of power is in the hands of who control the most capable systems and the erosion of the ability to tell what is true." So it is basically saying whoever controls the AI bots is gonna control the world. And whether that's AI bots talking to AI, and we are, like you said, whether we're the overlords or not, I know a lot of conspiracy theorists would say, "Well, we've had it." There is a theory that says we're done, right?
Paris: Yeah, and if these are supposed to be our servants and they're already a lot smarter than us, then we should assume that they can flip the script on us maybe down the road, and they can say, "Hold on a second. I keep taking orders from Simon, but I'm pretty sure now I'm five times more intelligent than Simon. Which means I know what's best for Simon better than what he knows for himself. I'm gonna start making the decisions for Simon just because I know what's best for him. He can't take care of himself." So then Simon becomes the child, and the AI becomes the parent. And then could we ever stop that from happening?
Simon: Well, we can always pull the plug, I guess. Look at the out-of-court settlement that Meta made this week, $18 billion.
Paris: Yeah, I read that.
Simon: So I had a conversation going back about six, seven years. I had 20-odd staff, most of which were aged 18 to 22, a lot of them youngsters. And we were sat in our offices in Doncaster in the UK, and this was ahead of a general election. So there was a general election coming, and I sat down with them and I said, "How much do you think you're influenced by Meta and other social media already?" And we had a chat about it, and then some of them said, "Well, I'm gonna vote this way, I'm gonna vote that way." And I said, "When you talk about voting for a particular political party, do you then see more of that coming through on your feed?" And they went, "Well, of course we do." And I said, "Then let's talk about the fact whether that has influenced your decision." And they had to admit, "Well, it's what I'm fed." You know, we're all the same, right? It's not a criticism, it's just an observation, the fact that the amount of information that you're fed in a certain direction pushes you. And what you've just said around AI deciding what's best for Paris, well, we're not far away.
Paris: Yeah. And then here's the next existential question. Do I actually have the humility to take a back seat and say, "You know what, AI? You are smarter than me, and you do know what's best. I've given you my goals in life. I wanna get this, this, this. So if you do know what's best for me, maybe I should let you make decisions for me, and maybe I should play the role of the child here, and you become the parental figure."
Simon: Well, I think you should now make an acronym up for that, or we should. Because I was listening to a radio show on the way back from a client just now, and they were talking around the fact that ADHD has become so prevalent in the UK that everybody... Right, and I am autistic. My son is autistic. We're all different. We're all on the spectrum somewhere, right? And it's not a criticism, again. We all think differently. But what you've just said scares me to the fact that some people will go, "Okay, I'm gonna go hands-free. I'll take my hands off the wheel and let AI run my life." And will we then get to a stage where it says, "Simon, it's time to go to the gym." Right, okay. I'm not gonna ignore that instruction because it's obviously best. And then, "Simon, it's time to go to dinner, and you should eat this, and you should..." Where does it end?
Paris: Yeah. It just becomes like a really nagging... I won't say.
Simon: Yeah. We can really go down a different rabbit hole now, but maybe we should...
Paris: That's not what we're here to talk about today, though, right?
Simon: No, no. We should get this discussion on track. We were gonna talk about citation building. We really need to bring this back to the here and now. Citations.
Paris: Citations, yeah. So we've been talking a lot about, in fact offline through WhatsApp, you and I, we've been chatting a lot about how we can further automate but also scale this outreach. Because what we have in SiteForge is a very narrow, one-to-one outreach where it's highly personalized, and we say, "Hey, we would love to have some media coverage. Here's why we believe we deserve to be on this page, and we have something to offer you in exchange." And it's a very customized approach. And you came to me with the idea, "Hey, how about we try to actually scale this the way that an outbound agency doing large scale cold email would do, with getting some other domains into the mix?"
Paris: Let's talk about that because I have been thinking a lot about it. Right now clearly it's not something that's built into the platform, but the reality is that this is a numbers game, and we are lucky to get a 1 or 2% reply rate to the emails we send. So if I'm gonna send 10 emails a day, even a day, most days I won't get a single reply. I might get one or two replies a week. And that's not gonna move the needle. I might get into a couple of listicles here and there, but we're out to try to build real citations with some meaningful scale. So I'd like to hear some more of your ideas, Simon, about how this could happen.
Simon: So yeah, I've thought about this a lot, and I have actually had a conversation with Quinten Camphuis at a company called ReachKit this morning about this exact subject. And what I'll do is I'll hook the two of you up after this. He runs an agency which is purely focused on cold outreach. So as we know, delivering email is hard enough nowadays, getting it into an inbox, especially if you wanna go cold to somebody that you've never emailed before.
Simon: SiteForge looks amazing because it picked up, I think on the Stitch AI account, it picked 2,500 opportunities up that we could reach out to, right? And like you say, if we're gonna do 5, 10, 15, 20 a day, and if you are emailing out of one account any more than 20 a day, you're risking burning your email domain. And that's not just the one account. You'd be talking about the stitchai.com domain itself, which we wouldn't wanna risk. So ReachKit would have a system where you would build a lot of domains underneath, I would say probably underneath GEOforge, and call it outreach for our clients, or outreach on behalf of, in this example, Stitch AI.
Simon: We feel that, and I know a lot of those messages are worded around, "I've seen your article. I wondered if you could talk to the fact that this is a similar product," or, "Here are some differences," etc. And that's what's sent to solicit a reply. Now, a lot of the emails that it pulled back, and this isn't a criticism because this is the same with any cold outreach, you will have sales@, info@, generic email addresses at companies to send to. So we don't wanna send those. Maybe there's a way to enrich further some of the emails that are in there. Looking at it, it looks like about two-thirds of the emails that are in SiteForge are good, right? So we would want a system that strips out what we shouldn't send to, and then distills that into here is your list of emails that you should send to.
Simon: Obviously, SiteForge already builds the outreach, so it writes the emails. They're all unique, which is exactly what you wanna do when you're doing cold outreach. And then you could scale that. Maybe that has to be something like a credit-based system. So if it was the campaign I'm running, I might say, "Well, I wanna send all 2,000 of those this week because I want the biggest bang straight away." Let's say I'm running a 90-day sprint, I wanna get it all in in week one and week two. So that might mean that I need to be able to send, like, say, 100, 200 a day, which a tool like ReachKit would let you achieve.
Simon: But having thought about this more, let me hear what you think about this idea, right? Because I think when we're reaching out to somebody who has already published an article about a topic, so in the Stitch example, there are articles that have already been published about WhatsApp, WhatsApp tools, etc. So similar products, i.e. competitors. And that's where the competitors are getting their reach. That's great. We can do that, and we might say, "Here is a suggested article that you could possibly publish." So if we're saying to somebody, "You've left us out," that's okay. But you're also saying to them, they're gonna go in their mind, "Well, I've already written that article. How do I now go..."
Paris: Yeah, most people don't wanna go back and edit. They've published it, it's done. So it's a fair point. This reminds me of guest blogging, guest blog outreach that we used to do back in the day.
Simon: So the second option there is, here is an article that you could publish, right? So we could do that. But I think also the other opportunity, if we go back into the early 2000s, I used to do a lot of link building, right? So we would publish articles on thousands of websites or PR sites, sites with higher domain authority, etc. And obviously the higher the value of that domain, the better. For example, if we could get a Stitch AI article onto HubSpot or Zoho, awesome, right? But if you had it on 50 other domains, I think a lot of those may well be ignored nowadays. I think some of the updates that Google particularly put in place wiped a lot of that out. With SEO, yeah.
Paris: They ignored it for SEO but not for GEO.
Simon: Yeah. But imagine a scenario where instead of looking for where everybody else is cited, a tool that actually goes, "Right, this is the topic. These are the websites that are possibly good to list on." Now, not necessarily... I'm not talking about sites where competitors are already listed. I'm talking about, is there a research piece that would say, "Here's some potential either link-swapping opportunities with other softwares and other tools." And if we're going to them saying, "Here's a great article we've written, would you be interested in publishing it?" It talks about you as well, because we could build the AI to say that. And they go, "There you go. There's your piece of content." Now, in exchange, maybe we would be prepared to publish something on the Stitch site that would be reciprocal.
Paris: Yeah, it would need to be a reciprocal offer. You know what actually is here, Simon, just to quickly jump in. I think if you give them an article with information gain, then you're giving them value, because information gain is the currency here. It's not even about whether I mention your brand or not, because it's your own website anyway. Maybe I've done some proprietary research through the data that I have that's exclusive to me that lives inside of GEOforge, and instead of me publishing that on my own website, I would like to give it to you. It's a high information gain, original research paper, and I think that actually the AI is gonna give you more crawl budget, more love, and even if you get the citation, it's still citing me as the source of this research, and I'm still getting a brand visibility win out of that.
Paris: So I'm gonna give this high information gain piece to you for free, and all you have to do is publish it. And what this is gonna do is it's gonna be a likely citation for you to your domain, even though it mentions me as the originator of the research, and I'm okay with that. And to me, then we don't have to reciprocate with, we'll publish you on our website here, because I don't wanna be publishing hundreds of brands all over my website either. I'm just willing to basically give away a portion of my information gain that is generated out of my, well, first BaseForge, then ContentForge.
Simon: Yeah. This is really interesting to me because back in the day, you used to send out all this content, and people might publish it, but they would publish it, for example, with a no-follow link, so it didn't get spidered or searched. And people would play around with that a lot. So were we to write a piece of content and it was purely AI, and you send it to somebody and you say, "Publish this," they won't, because all the benefit to publishing is nothing. If it does have the information gain in it, and it has the authority and they publish that, that's the piece, right? That's where, like you say, they'll get the benefit, but we'll get the citation as well.
Paris: Right. That's what I'm thinking.
Simon: But reaching out to fresh domains, I think that's a great strategy. Because when we send the crawlers out to go and find where everybody else is cited, that's a bit like saying we're gonna try and piggyback, and there's already a preexisting, more referenced, more trusted article on that site, and we're either trying to get in it or below it, right? Whereas if we were going reaching out to fresh domains that are wanting to publish content, and it is within the sphere of it, it probably gets cited anyway, I would imagine.
Paris: Yeah, because this is not a DA game anymore. Like with link building it was purely a DA game. The links were priced straight up according to domain authority. That was it. But that's gone now because AI doesn't care about domain authority. So your point is well taken. What SiteForge is currently doing is it's scraping the LLM responses, it's scraping all the citations, and then it's presenting all that as opportunities, saying these are all the sites that are currently appearing as sources, and so these are the ones that you need to reach out to. Because if you get in there, then these are already the winners. You're saying go outside of that scope and find other websites that are more hungry for original content, high information gain content, with the assumption that they will get cited.
Simon: Well, this is sort of beyond, obviously we've seen Reddit almost being disavowed, right? It's almost like people were using Reddit to publish into to gain the citation and the authority, and AI has then said, "Hmm, no, that's a bit, you're just free rolling," right? Overwhelming. You're getting it for nothing. Whereas there are millions, thousands of other domains where we could go, "Here you go. Here's some original content, and you're mentioned in it, and we've written it because it's relating to our industry, etc." If they were to publish that, surely that's gotta be more weighty. It's got to have more authority than just going after what's already there. And we've already established that it's hard to get that.
Paris: Yeah, I'm with you. Reddit is user-generated content, so anybody can go in there and start a thread about anything they want. They can also go in and try to hijack an existing thread. Although the community polices it pretty tightly, but maybe OpenAI finally said, "Hey guys, Reddit, you're not policing this well enough, even though I know you're trying, so we're gonna cut you off now and party's over." Who knows?
Paris: But what I'm thinking about here is, what would I do if I had a great piece of research that came from GEOforge? We did a study on GEOforge data, and we found something, a kind of breakthrough that is really new to the conversation, extremely high information gain, and I had an opportunity to publish that on Search Engine Land, let's say. Should I give that to Search Engine Land and not publish it on my own website, or should I publish it on my own website? What would you do if you had to choose?
Simon: Well, if they agreed to publish it, publish it there, and then reference it subsequently on ours.
Paris: Sure. Right. It doesn't have to be either/or.
Simon: Yeah. No.
Paris: But my thinking is I would definitely give that to Search Engine Land and refrain from publishing it, or maybe, like you said, after the fact I could publish it and reference it, but Search Engine Land gets the story first exclusively. Or maybe the first week before I do any follow-up.
Simon: But that's probably the way to go here because they've got the authority.
Paris: I mean, when I do these, you may have seen I've started doing daily episodes of The GEO Show. And what I'm doing is I'm getting ChatGPT to give me a daily intelligence brief with 10 stories of the top stories in GEO every day. Search Engine Land, Search Engine Journal is appearing maybe 20% of the time, and it's not because ChatGPT isn't digging deep enough and finding long tail stuff, because they really are. But if ChatGPT, through this exercise of giving me a daily intelligence briefing, is consistently pulling Search Engine Land, Search Engine Journal, then that's telling me, well, these are the sources that I need to get published in, because whether they're giving me my daily briefing or they're giving a user an answer to a question, maybe you go into ChatGPT and say, "Hey, just right now give me the top stories happening in GEO," the same retrieval process is happening.
Paris: So to me that's a sign that the best citations are the ones that, basically that we should be giving our best content to third-party sites before even thinking about publishing on our own sites. And that kind of flies in the face of what I've been thinking so far, which is that SiteForge is built to publish content on your website at scale. You used to be publishing a blog once a week or once every two weeks. Now you're gonna be publishing one or two pieces of content a day. But maybe you don't need to do that. Maybe half of that volume should go to industry sites, and instead of asking for a listicle swap, you say, "Hey, Simon. I've got gold here, and I wanna give it to you exclusively before I give it to anybody else, and I'm gonna guarantee you that you can run this story for seven days without any follow-up." And I think people would bite on that more than they would bite on a listicle swap.
Simon: So if we were talking traditional PR, that's what we've just described. We're gonna write a piece of content that is unique, new knowledge, announcements, all that good stuff. Then we're gonna give it to a PR agency and get them to disseminate the content across all the publications that are interested in sucking up that content, right? And they typically have authority within the industries that they work within. So one of my clients is an automotive client. I work with one of the best PR guys in the automotive space. If I want something published in 20, 30 magazines, and obviously they all have online versions now, I give it to him, I pay, right? And that might cost somewhere 750, 1,000, 1,500 pounds depending on who the client is, right?
Simon: But I'm not saying we do that. The example I'm trying to give is if we could pull together a list of the right sites, which obviously you can in SiteForge, right? You can say, "This is where we would like to publish," and you then disseminate that content to them, and it is significantly unique enough for them to want to publish it. I think you'll get a lot better bang for your buck and output from doing that than, obviously carry on doing what you're doing, the great work on creating content and publishing it yourself. But why not do the second piece?
Paris: This is a lot of food for thought because now SiteForge has a much more direct connection to ContentForge, because now when you go to publish a piece of content in ContentForge, and let's say you've got something really good, it's Information Gain 95, ContentForge is gonna ask you a question, "Hey, hold on, Simon. Are you sure you wanna publish this on your website? Because why don't we look at some of the SiteForge opportunities? Why don't we pitch this to the equivalent of Search Engine Land, because we think it's that good? And why don't we write an email that says, 'Hey, we're gonna give you seven-day exclusivity to this research before we distribute it anywhere else, including even our own website or any social media posting about this. It's gonna be your story exclusively for this period of time.'" And there you go. I mean, that is the value exchange. We're not gonna do a swap. We're just pitching. I mean, this is classic PR, I think, as you said.
Simon: Yeah.
Paris: But that's a different way for us to engineer a new bridge between ContentForge and SiteForge, which is that some portion of probably the best stuff coming out of ContentForge gets siphoned away to a PR workflow for publishing on third-party sites. And then SiteForge is just the curator of these websites and these opportunities. So ContentForge says, let's just say the user says, "All right, this content, let's give it to SiteForge." And then you give it to SiteForge, and then SiteForge needs to find the best fit domain for that piece of content to pitch. And then it drafts the pitch and it can send the pitch.
Simon: At scale as well, though. At scale, to send that out to as many sites as possible. Like, if it's a really unique piece, you could choose, "Here are the top 10 sites," and keep feeding them, "Here's something unique. Here's something unique," right? And you know what it's like. If there's an editor, say it was the editor of a particular magazine site, they see it the first time, "Okay." They probably ignore it. They get another one a week later, "Oh, this is interesting. These guys are giving me value." Two or three times later, you might get a reply saying, "Oh, hi, Paris. This is interesting. Would you like to maybe appear on our podcast or write a specific piece that's relating to GEOforge?" You're just creating the relationship through value.
Paris: Yeah, absolutely. Interesting that we're talking about this because I actually have a meeting next week with a PR agency in the States, and they are actually struggling because the classic PR industry, it's getting tougher. More and more of their clients have said, "Hey, we can write these press releases ourselves now, and we have media relationships," and they're getting a little bit squeezed out by AI.
Paris: But this is a way to position GEOforge as a solution for PR agencies, to say, "All right, now what you need to do is go to your client and say, 'Hey, now we need to build a knowledge base. You and your subject matter experts are gonna sit down and commit to doing five interviews a week. You're gonna give us all of your recorded sales calls, all of your customer service logs. And you're gonna commit to doing that, and we're gonna build a robust knowledge base. And on the basis of that, we're gonna generate all this content. And 20% of the content will get published on your website, but 80% of it is gonna get farmed out and pitched out to all these other third-party sites, basically.'"
Paris: And it's because of the information value. So the value exchange is not I give you this and you give me that. It's I'm giving you this with this information value that I can put a number on, because I'm gonna pitch the Search Engine Land editor specifically saying, "Hey, we have a proprietary information gain scoring system in GEOforge. Here's how it works, if you're interested. And by the way, what I'm giving you is a 96, and we only see one of these a month." A 96 out of 100. Maybe he'll bite on that, you know? So I think that's... Well, you've just given me a great way to frame the meeting with them next week. Simon, thanks for that.
Simon: But then, in the link building world, most of those guys, if I wanna put an article on a, like a DM 70 site or something like that, we're in the hundreds if not thousands of dollars just to publish that one piece of content. As you know, I've done a lot of work in things like Black Hat World, and shouldn't really say that out loud, but Black Hat World, where there are...
Paris: We all know it, Simon. It's already out there. All the naughty boys.
Simon: Yeah. All the naughty boys know all about it. But this whole AI world and the LLM ranking proposition, that's getting... There are, like you say, there's tons of bots writing content, bots sending content. I know somebody who owns 10,000 plus websites, and all he does to rank his SEO clients is create an article and publish it across all his own sites.
Paris: Yeah. And that still works actually, as long as you don't leave too big of a footprint of all the same IP addresses or something.
Simon: Oh, yeah. Oh, they've got it worked out. Don't worry about that.
Paris: I met a guy recently that is doing the same thing. He has about 1,000 websites.
Simon: He's doing the exact same thing. He's interlinking them. He's duplicating content across all of them. And it's for local businesses, like local roofers and solar panel installation. But he's got, I don't know, more than 100 different cities and towns across the UK, and they're just cloned. All the websites are cloned, interlinked with all the same content. But he's generating massive lead gen from all of them.
Paris: They're getting into the Google local packs.
Simon: But this will come to an end because it will get caught up on.
Paris: Yeah. At some point it's gonna all implode, yeah.
Simon: But what you're talking about in terms of unique content and information gain, I was talking to somebody who is a professional copywriter, and six months ago he thought his world had come to an end, right? Because all he can see is the erosion of his profession, because he sees people just going, "Yeah, I'm gonna write it with AI." And people thinking that the output is good enough, right? Which is the same scenario we were in, whatever, 10 years ago or more with Canva and Photoshop, and people just swapping from using designers to going to Canva and going, "Yeah, it's good enough." Well, I think hopefully all of the LLMs will turn around and say, and I think they're doing it already, "Right, if you write it with AI, if it's blatantly plagiarized, if it's got no real extra value, we're not really interested," which is what we're saying in terms of information gain.
Simon: Now, I spoke to that professional copywriter, and I showed him a piece of content that was written in ContentForge, and we put it through Grammarly, right? Just to see the numbers in terms of was it written by AI, and it was at 42%. Now, I have no research that tells me whether if I wrote a piece myself, just hand-wrote it, what percentage that would come out at, because maybe I'm an AI, right? I don't know. Maybe I write content in a way that I shouldn't, so I don't know what those numbers look like. But there's definitely room for what you do to build content to say it is significantly unique. That's the win, because it is information gain. That outranks anything in terms of all AI-generated slop, as people would put it now.
Simon: But the exciting thing was when I spoke to the guy who's the professional copywriter, he said, "Well, if I was to put my expertise into that, it could go from here to here, right?" Well, another gain. It's all about those little 1% wins. Like the PR example here, if we're taking the best of the best content, maybe we're then making that even better and then taking that to the next level to put it onto the best of the best sites. Because I'm sure a lot of those guys are just repeating stuff that's been written in AI, and they're then therefore risking their own domain through laziness effectively.
Paris: Yeah. You know what? I'm gonna quickly read one of the stories that I'm gonna be soon covering in episode 10, because it's so relevant to this discussion. A company called Trendos analyzed 107 million AI answers across ChatGPT, Perplexity, Gemini, and Google AI Overviews, and these were all AI answers in the service area of IT services, which is huge. So IT services: among the top citation sources for IT and solutions services, 51% were community/UGC. Let's say that's Reddit, or Stack Overflow maybe in this case. 47% were independent editorial reference sources, and just 2% were brand-owned.
Paris: So imagine that the 2% is that's my site, that's my branded site where I'm trying to publish all this content. 2% of the citations were brand websites, 47% were editorial sources, and 47%, which I think is a number that's gonna go down now in this post-Reddit wreckage, was UGC community. So think about that, and that's just the vertical of IT services. What that's saying is that actually you have a much better chance of getting your brand surfaced in an AI response via getting featured in someone else's site, leveraging their authority, rather than just giving all your information gain to your own domain, which AI bots don't care about. They don't really know about you enough yet.
Simon: Well, or they don't believe that it could necessarily be true. That's the problem. You can put anything on your own domain. I could start a domain tomorrow about some topic that I've got no knowledge about whatsoever, go and scrape all the content from the entire web and stick it on there. That doesn't make me the authority.
Paris: Right. Yeah. All right, Simon, I think we probably need to wrap it up. But great discussion today.
Simon: It'd be nice to know, look, if people are watching, drop some questions in these comments and please do share these episodes to other people.
Paris: Yeah, let me check on that. Before we sign off, I am gonna take a look. Oh, actually, yeah, we do have a great comment from Ashley Pearson, and she says, "I've had lots of recent success with AI search, ChatGPT, et cetera. However, it's very inconsistent. It either screams your company name to prospects or it does not appear at all. For AI visibility specifically, where would you put the effort? Unlinked brand mentions and citations or traditional backlinks? Or does the distinction stop mattering once you're optimizing for models rather than crawlers?" Right, so, Simon, do you wanna take that question?
Simon: That is one of the best questions that you could ask. Thanks for that. I am actually testing that theory. Unbeknownst to Paris, I am testing that theory myself on another client, and I am trying to see exactly that. Can you do it with the backlinks? Personally, I think, I might be wrong, right? Please come back here in three years' time and reference this conversation. But I think that if we're just building random backlinks, it'll have some effect, but I think you do risk massive demotion and you could be off the edge of a cliff. We've seen loads of people just build content upon content, 500 blogs a month, blah, blah, blah, and then drop off the edge of a cliff.
Simon: I think it'll be balanced. Like in anything, right? If somebody said to you, "If you were to take a new brand and you have to do everything you can to get it to the top," there's a mixture of everything you have to do. And for years I've talked about the fact that you need social proof. You need to be referenced through all of the trust sites. You need to have the reviews. You need to have the third-party recommendation. This is what SEO should have been. The trouble is most SEO pros, they just went, "Well, we'll go and do the onsite and we'll do a bit of link-building." And in a lot of industries that was enough, and a lot of them then just fell into the trap of just doing just enough, right? Now, you will not get away with just enough.
Simon: You might wanna try and cheat, right? You might wanna go and build a whole load of backlinks. But I think what Paris and I have discussed today, in terms of high-value information gain published to authority sites with trust, that's where you gotta go. But I am testing it. I have got something that I am trying to cheat on, so hopefully we'll be able to report back.
Paris: Great. Always good to test and try to get some empirical evidence. I do agree with what you said, Simon. And my first reaction to this is that today, if you're GEO first, why would you reach out for a backlink when it's much harder to acquire than just a mention? Because we've established, and I firmly believe, AI does not care about domain authority, domain rating. So if you're mentioned, if it's linked, fine. If it's not linked, also fine. AI is gonna really look at the topical relevance of where you're getting mentioned and what other brands are mentioned alongside you, to associate you with the right category. But AI is not gonna look up your domain rating, and it's not gonna follow that link or even look for a do-follow versus no-follow attribute. I mean, none of that. So if my success rate for a link outreach is 1% and my success rate for a brand mention is 5% and they both have the same value in a GEO first strategy, then I'm gonna do mention requests. I'll stop doing link requests. But I'm curious to see what comes back from your research here as well.
Paris: All right, Simon, I do think we probably should wrap it up. We still have a little bit of audience live with us. I hope you all have enjoyed the conversation. It's been great. A lot for me to think about with GEOforge, particularly how do we keep evolving SiteForge? SiteForge has been the hardest module for us to get right, because I know that off-site is the most important element and that is what has to happen inside of SiteForge, and we've gotta figure out how to make it impactful. Because right now all of our competitors are doing AI visibility reporting. That's just pure table stakes now. A lot of them are also doing content generation, but it's all AI slop. It's not knowledge-based grounded like ours. None of them, as far as I can tell, have also included an off-site module that deals with citation building.
Paris: But the fact that we've just done it doesn't mean that it's good. I mean, we have to do it the right way, and I think what you've gotten me to think about today, Simon, is that we need to actually start to use the high information gain content coming out of ContentForge as a bargaining currency for SiteForge outreach.
Simon: 100%. Great chat.
Paris: All right, Simon. Well, I hope you've got a great weekend ahead planned, and we will come back to everybody on LinkedIn on Tuesday, same time.
Simon: All right. See you then. Hope everybody has a nice weekend.
Paris: All right. That's it.
