Latest / Semantic SEO Podcast / External Topical Maps for Semantic SEO: Building Third-Party Consensus and AI Visibility
Transcript
- James Dooley: External topical maps. Getting those third-party sources to reference you is unbelievable for AI visibility, and I've wanted to have this episode with you for a long time. Thank you. So for anyone who's listening to this, what is an external topical map? Koray Tu: The first thing is actually during our discussion, you called it in that way. Uh, in, in my case actually, again I called it actually semantic content network, but we just create this network in external sources. It was something that actually we were doing earlier too, to change the definition of the concepts in the Google's eyes, or to, let's say, affect or impact the co-occurrences, or change autocompletes or, uh, suggested query terms on Google SERPs as well. But technically, I always thought the semantic content network was for your website, not for, not including external sources. It's the main part that we focused, like, 90% of the time. But sometimes, uh, the SEO projects are not performed for ranking a site. It is also performed for, let's say, changing an autocomplete or changing related search terms or changing SERP features as well. Or when we ask basically, "Who is the best?" or, "What is this?" we, if we want to change a result beyond our site, then we need to go beyond our site to publish extra things, basically. And technically this external topical map is about bringing multiple different domains that, that are able to link or mention each other while also trying to impact the attributes of the same entity. Recently actually in my socials, I shared a project that I am helping to an agency as well, and the project name is, maybe I can also slightly tell, Slip Start, and it is ranking way better in the local results thanks to these external mentions and the publications as well. Because as I explained in the visual semantics too, uh, we try to attribute or assign certain attributes to the entity, like the biggest law firm in Houston, or most successful in Houston, or the, the law brand that actually got the highest level of compensations, etc. You, if you are able to know which attributes are the most important for you, you connect these attributes to the keywords and these keywords basically becomes the macro context of your topical map that you'll be publishing externally. Then you create again the content briefs, you write the articles, you submit these articles to these external sites, and they publish it with certain frequency by mentioning your brand in certain ways. It doesn't about... It's not about also the links too because you can get three things from these publications. Link, mention, claim. We don't care about the link part that much all the time. We usually care about the mention and especially co-occurrence around that mention. And the last part is the claim. The claim again is like the biggest law firm or the most successful law firm or the oldest or the most experienced. You can find these attributes even from Google's AI Mode as well. Or if you are talking about QR code generators, let's say, and your QR code generator is for enterprises mainly, but you also want to be able for B2C, the moment that you focus on B2C keywords, it's about free usually. You have to rank for the free, like free VPN or free QR code generator, meaning that your external topical map should be mentioning you as a free source or free to use for two weeks, or there is a premium version, to convince the search engine to do this. And you can't use every domain for every topic. And you can't also, uh, let's say, uh, publish every type of, let's say, content brief again for every domain as well. So first filtering the domains, then choosing anchors, annotational text around the anchor, then creating the brief and ordering these will be important there too. The hard part is usually adjusting the domains to witness your greatness on the search, basically. James Dooley: See, I want to go through a couple of things on these external topical maps, right? So, first and foremost, you seem to use the term wasteful domains. Yeah. I hate that saying, by the way. Right. I think you should change because it's not wasteful. I think it's... I think it's helping you. I think it's supplementary domains or secondary domains, as opposed to wasteful sounds like it's just, don't throw it in the bin, it's a waste. So, that's the first bit of the... That's just a naming thing. But in my opinion, that, that needs changing. So, moving forward, don't call it wasteful domains anymore, please. But something that we started to do, right, and we've had these, not, not debates, but we've had these conversations, and I talk about a lot about query fan-out, right? And I know that you say it's just query augmentation, right? And I, and I get it, right? But for me, what started to happen was when I started to do certain searches in AI, like ChatGPT, they go and do, like, a lot of the time it's more like a paragraph kind of like, or a full sentence search, instead of it just being two or three words like it used to be with, with Google. And the LLMs would extrapolate synthetic queries, which is what query fan-out is, and then augment it, and then it come back with certain sayings like FatRank versus Checkatrade. Now, Checkatrade is a lead generation agency and FatRank is a lead generation agency, and the amount of people that are in the semantic community that then would cover that in the semantic content network on their own website. But I hate that myself because that's a, that's a first-party source. FatRank is saying FatRank is better than Checkatrade. And what I tried to do was go out to all my clients that was using Checkatrade that now use us and now believe that we are better than Checkatrade. And I do the article on their website and I word it in a way that's still relevant to their site and their growth, but it's on my third-party site that says, and it's an opinionated type article, "We was using Checkatrade. We was spending this amount of money. We was getting this return. And then when we moved to FatRank, we now get this much higher return and much better." And the third-party corroborative source, which I kind of got from Jason Barnard, was the proof part. So the claim and the frame on my own website was, "I believe we're the best." But everyone says they're the best on their own website. And the proof element of the external topical map, now we're like, hang on a minute, externally we don't cover this part of the reputation. We don't have enough awards, not just on our own site, but on the third-party sites. And I feel like it's almost like a, the external topical map, as well as maybe covering topics that you might not want to cover on your own website, the reputation element is huge. Koray Tu: That's why actually in, in the, in the first part I called it a little bit waste because you actually create lots of content there that you can't control the quality of it that much because you scale too, too quickly. Maybe you will be publishing like 2,000 articles around the web. There is no way of making all of them quality. But still, that repetition actually creates the consensus. The first, technically, what you create, that, that consensus called actually corroboration pages. You basically put your proof to the pages that are corroborating with each other to prove the point to the search engine. And all these actually come from a research paper in 2009 from Mindi Mind Vu. And he, she actually directly explained that even the Honda website, for instance, is giving different values for the Honda Civic mileage, for instance. And in that case actually the search engine is confused, like what is the mileage of Honda Civic? Even the Honda site doesn't give the right answer as it should be. In that case, search engines usually have something that they call basically answer scoring. And he, she also explains what type of methods they can use, even part where the answer is located on the page or how well it is covered on the page or whether it is actually the only answer on the page. There are different type of methods that they use. And along with that actually, if they're able to see the numeric values, especially for the answers, if it is in a certain range, they didn't give a name to it, but I called it truth range. Meaning that actually if your answer and the values in the answers is beyond reasonable numbers, they usually don't accept the answer at all. But if you are in a certain range, they start to accept it and they start to calculate it to actually see the end result on layer and too. But technically when we create an external topical map, it is about staying in that truth range across the different web sources in, in the most convincing way as much as possible to convince Google that there, there is an entity, there is an attribute, and the value of that attribute is this. And when it comes to the who is best or which one is best, you can also try to use chain of reasoning. In my opinion, again, you can use agents from ChatGPT, Claude, or other platforms and try to see how they are using chain of reasoning because if you compare two different brands to each other, you will need to see what attributes the search engine will compare first. You can directly ask it to Google, by the way. You can go to the Google AI Mode, say, let's say, X versus Y, and what are the most important attributes to compare these three, four to each other? And then you can actually start manipulating the values of these attributes first. Maybe the last part there, uh, might be, uh, once you get these attributes also, uh, you can try to create a scraper, like let's say, "The best lead generation platforms according to..." You can say XYZ. You can put three attributes, every three time change these attributes, scrape the results, and see according to what attribute how the answers of the AI is changing. Then you can start manipulating from there actually in an easier way. Actually with Alias, we were trying to create a script for this as well. James Dooley: Yeah. I mean, like, so I'll give you one or two examples for what we was doing. There were certain websites, we created single-page websites and we created a new brand, and all we did was an external topical map, and all we did was create this, where you call it semantic content network, but on all these third-party websites that then built up everything that was about this single page. So there's nothing on the website. It was a single-page brand, and everywhere else it repeated who we are, what we do, and why we're freaking brilliant. And everything was defining. So who, who we are was defining the entity. What we do is all the services that we offer. Why we're freaking brilliant was the reputation, the awards, the testimonials, the reviews, the case studies. Yeah. And we now, with that single-page website, now get cited in ChatGPT as being the preferred supplier or preferred service, and in Claude and in Gemini and in AI Mode, without it even being the claim and frame on our website because the proof on these external sources was able to... Koray Tu: That's slightly different in my terminology. I call them extension domain. They are extension domain. Yeah. There are two different domains in my system. One of them is the main domain, your main, main money site. Extension domain is what you just mentioned. It directly using some exact or partial match words that I especially try to manipulate the prompt results for them, basically. And for, let's say, the eSIM industry, we try to rank for the best eSIM Africa, then I'm getting basically a site like eSIM Africa dot travel. Then I explain there why the, let's say, XYZ brands are best in that part. This is the extension. The wasteful domain, the third part, which I will change the name... But I... That part is basically where I am just scaling without that much control, I can say. But these are all on websites that you own? Is that what you're saying? I... Yes, we own and control them. When it comes to the external topical map, that's the fourth layer, which we directly... We don't own any of these sites. We just pay them or we just convince them to publish these type of articles on their sources in the way that actually can support my network as well. James Dooley: Yeah. So prime example, like SEO Mastery Summit. Biad's a very close friend of both of ours, good mentor for both of... When I attended his conference, one of my brands won best lead generation agency, and that went on his award event kind of website and the award was positioned on there. And now when someone was to type in, "Who is the best lead generation agency?" part of the reason of them saying... When I ask back, "Why did you say that FatRank is the best lead generation agency?" it says it's an award-winning agency that won this at S1, and it cites their source on... And these are on sites that I don't own. And they're the, the ones that are kind of like the wasteful domains or the extension domains are almost like PBNs that you own that asset. I love it on the external topical map on sites that I don't own, but they can say why I'm freaking brilliant and they can get lots of different referring domains, then that kind of... Koray Tu: That's true. Just in the health niche, doing this is not that easy because they are regulated heavily. But in every industry, like especially in hair transplant clinics too, it's very common, for instance, publishing actually fake reports or fake actually statistics. And let's say you publish fake hair transplant statistics, a journal from UK is citing it actually. They even don't check whether the research is true or not. They just believe in it. And I believe in industry too many people do that as well. They just say that, "I basically checked six million ChatGPT prompts and their results, and here are the stats." Whenever you use actually this, like, "I checked XYZ, six million, five million, this," and actually these AI systems, overall all the LLMs actually are looking for these type of stats or these type of unique surveys, which is also coming back to the algorithmic authorship because they are, these LLMs are trying to find these sentence groups and these paragraph groups as well. James Dooley: See, I want you to create a module in your course because you are the semantic SEO goddess. Thank you, little princess. Thank you. Um, and the reason why I want you to create it, right, is because I want to talk a little bit about consensus, right? Because it's fine creating an article, let's say, about the award that you might have won, but ideally, you want multiple awards that you won. So you're a multi-award-winning agency. Or you want multiple testimonials. You want multiple reasons or multiple different corroborative sources saying why you are the best. And I think, I think somebody like yourself would be able to create this external, whether you call it an external topical map or an external semantic content network, of pages strategically written. And you might be able to understand that more weight should be carried in certain areas. That should be, okay, you might need... Because this isn't just on your own website where you're going to open up two pages for similar topics because you're going to have cannibalisation on your own website. You almost want the cannibalisation on these third-party sources because you want... That's just building consensus, having multiple people saying it. And I feel I'd love somebody like yourself to put the time and effort into... Because I feel like a lot of people do backlinks, and they do it for the backlink, but I don't think many people are doing it for now the AI visibility and the mentions and the semantic triples that are going on these other sites that's defining the entity. It's also, which I've also done a lot of research on, massively strengthens your KGMID score within the Knowledge Graph from Google's API. So Google stopped doing PageRank years and years ago. However, they still update the confidence and clarity score within Google's Knowledge Graph API. This is on Google's website. It's Google Knowledge Graph API, and they're still updating that. Why would they still be updating it if they're not using it? And it... I know it's related to... It's not PageRank, but it's confidence and clarity of the brand, which mainly comes from third-party corroborative sources, which then means your knowledge panel gets strengthened, which then means you could form part of the training data that's being done. And I feel all this element needs to form under your course, under semantic... Koray Tu: After the visual semantics. Uh, actually it's also in the plan too. And actually I want to merge the visual semantics model announcement together with our mastermind, like in September. And I want actually the attendees to see it first, check it. If they approve, I will launch it. If they don't approve, I will improve it. And I will again present it to the beta testers. If it is approved, then I will add it to basically the course. And the external topical map actually can be the next one, definitely. James Dooley: I love you. Anyone who's watching this, we've just done another episode. He mentions there visual semantics. Make sure you check out the link in the description. I'm going to keep pushing him for this third-party topical map, external topical map, whatever you want to call it. Koray, you can call it as long... It's not wasteful domains. You can call it what you want, but I want this external topical map done before Christmas this year. Koray, I love you. And anyone that's not yet actually got his course, I strongly recommend you go and do it for the on-page semantic content networks, the topical maps. He's a good friend, Tuğgy. And thank you very much for everything that you do for the community. Koray Tu: Thank you. It's always a glitch.