AI. CTR. Chaos. What the Smartest SEOs Are Doing Differently in 2025

We just anecdotally we just saw our first lead come in for one of our clients from a chat GBT referral and I'm just itching to try to engineer back what was that citation where did it come from and you don't have to have a backlink though what you need is your brand to be mentioned it could be mentioned in a video it could be mentioned in a in the right context hopefully in the right flavor so I think um a backlink is a subcategory of a citation and I think it's citations that are important kind of like having like a very confident friend I mean I think people need to think about these AI answer machines as like friends. Like you can have you ask your friend, you know, I'm having heartburn. One person go like, "Yeah, you're fine. Just eat more, I don't know, hamburgers." The other person may say, "You're having a heart attack. Go to the emergency room right now." And they're very confident in their answers. I think that's what AIO reviews are kind of like. 96.5% of content never gets any traffic and just we have some crazy numbers. Search is is dwindling. People are in other places and AI overviews are everywhere. And also AI search is everywhere. [Music]

Well, welcome to SE Rankings SEO Foresight 2025 online conference. conference and we're talking about navigating AI, Google updates, and search habits. I'm Aaron Sparks, host of Edge of the Web and owner of Site Strategi and Edge Media Studios. I'm very honored to be able to be asked to moderate this conversation today. We're going to be speaking to some incredible talent in the search optimization space today and I'm excited to actually watch their presentations. So, after each presentation, we'll have a chance to be able to ask the presenter questions. We've got three presentations and a panel at the end of our conference today. With that, that's brief housekeeping. Let's get this conference rolling. Let's talk about who's on the panel, who's on the agenda today. So, we're going to be talking to Peter me. He's going to be talking about the future of link building sustainable strategies for 2025. We're also going to be talking to Sarah Tahar, uh, the new SEO, navigating the AIO and LLM revolution in search, and talking to Dixon Jones. He's going to be talking about mastering topical authority, how to become the go-to source in your niche

for 2025. and an expert panel talking about the state of AI overviews in 2024 research insights and future forecasts with Barry Schwarz Blau and Anastasia Kot Kotabinskia the future of link building sustainable strategies for 2025 by Peter me he's the head of SEO at bring performance Peter thanks for joining the conference today thank you all that said hey you know what I'm going to te it up to you go ahead and take it away so my talk today is the future of link building sustainable strategies for 2025. Link building is back in style, baby. I guess the question is that did it ever really go out of style. Well, yes, it did according to some. Maybe some people never gave up on it, but some people did get scared off. Uh I know a lot of clients did, but we do know that there is uh with link building there are certain risks that get carried with it. I mean, pretty fair to say that if I'm receiving link offers here in my inbox

for $5 for a DA92 link, I'm not so sure whether I'm going to take that offer. But, uh, yes, we do need links. Um, but we want these links sustainably. And so, the sustainable strategy for 2025 is integrated digital PR. So just diving a little bit more about me. I've been a longtime advocate and supporter of the SEO community getting involved in lots of um community uh events, meetups here in Melbourne, the Melbourne SEO meetup groups in general. I'm just an old school SEO guy. I'm passionate about ideas and I just want to share my experience. And so what would the process? Well, I'm breaking this down into sort of four main areas. the research and strategy and development, content creation, optimization, outreach and amplification, analysis and iteration. So let's go to the first bit of this which is the uh research and strategy development. So here's where

we're defining the goals, align them with the audience's needs and behaviors. So, I think there's there's a fair bit here, but I've been thinking about this stuff, personas, all the sort of content marketing side of things, which is typically outside of the realms of the SEO approach. But it's definitely definitely worth it to go through that process of understanding who are the personas, who are the who are you, who is your audience, who are your potential customers. Pick pick existing customer and work through that. And I mean there's some tools here. There's a link here you can go through to the HubSpot. There's plenty of ways of doing this sort of thing, but the more research, the better results you're going to see. The next stage here is the uh developing content that's optimized for search, highly sharable, and aligns with the PR goals. So, this is the content development side of things. There's plenty of ways of doing this. SE Ranking has a great content marketing section in their tools. uh surfer has a you know

also has the NLP stuff. So the NLP options I mean you know these tools are always helpful. They're great. At the end of the day though you know some of our uh some of our people are just using word docs. They're just writing articles cuz they're not don't want to get too hung up on the SEO side of the article. outreach is, you know, this can this can go wrong, but we want to leverage the outreach to earn the backlinks. Media coverage and social engagement. It's not just about getting the links. It's the media coverage, the social engagement, the amplification. There's plenty of ways. There's plenty of tools for doing this. Help a reporter out. Of course, they're going away, getting shut down. I think Sision owns them, but Sision's a bit more of an expensive tool. Uh there's plenty of other tools you can use. uh even things like LinkedIn, Sales Navigator, Twitter, just get on to Twitter and search for people. Often their contact details are there the and also old school go to Google, get a Google sheet and just start collecting

details. Start small rather than just sort of rather than plowing in. So the fourth step here is to measure the performance and refine the strategies based on the data and the insights. We want to improve. We want things. We want to look at how well did we do, what worked and you know that this going back to this client again and uh looking into the results over the last 7 days for our client, we can see we had 73 new referring domains in the last 7 days. This is so uh there's a backlink checker in SE ranking which is really good. But you need to decide what the metrics are that are important to you for your, you know, some people go by DR. Some people think DR is a load of not important anchext follow no follow. I think relevance to me relevance is the biggest most important metric. Uh, also I want to see some referral traffic from some of these links. Um, I just think if you're going to get a link just for the sake of the DR, it's on its own it's not

enough. It doesn't make that much sense. Yeah, there's all kinds of metrics that you need to figure out what it is for you that you consider it to be quality. And the other really important thing here is to is the team. This is so this kind of project I'm talking about, obviously there's a fair few people in the team here, but when we're talking about, you know, you might take this on uh as a smaller business perhaps um you might only have a few people in your team. So you might split up some of these roles like some of the SEO tasks there might be one person doing them. Uh you might have you know content creator and a a writer might be the same person. This kind of a thing. Whatever you decide to do the takeaways here is I think it's important to remember that Google is in the business of dishing out penalties. You know, if you've sort of seen anything that's been going on lately in the SEO world, you can see that Google goes after the biggest, you

know, biggest and the best. It's not shy. Just think of is what I think is important. But spend time on the research and the strategy. Create the highest quality content you can because that will help you to attract links. The links will come much easier. Avoid the outreach spam. You want to make sure that your domain doesn't end up on some kind of a MX blacklist or you know your your emails are all going in the spam folder. Iterate and analyze and iterate on the kinds of things that work and uh build your team. So it takes a lot of effort but above all enjoy the process. It's a lot of fun. I'm happy to keep the conversation going. Let there be content and links. Thank you. Thank you very much, Peter. They got a number of questions already teeing up for you. Uh Andrew said, "Peter, as Google's increasingly emphasizing user content and entity based search, uh how do you see link building evolving to prioritize meaningful connections over traditional

metrics like domain authority, which you were already talking about, but that meaningful focus is the relevancy. What What would you say to Andrew there?" Yeah, thanks Andrew for that question. I think definitely uh Andrew would best approach would be double down on all of the of the as I said I had the sort of trifecta there where it's the it's the it's the the digital PR and promotion it's the content development but it's also the technical SEO and I think for me what I'm really saying to people is work on all aspects of that. Absolutely. Absolutely. Ethan Wallace has a question. Uh Ethan uh actually said, "Thanks. Hey, a with AI generated content flooding the web, how should SEOs adapt their outreach tactics and ensure they're actually building quality links?" Yeah, great Ethan question. Yeah, I mean it absolutely some of those numbers like 96.5% of content never gets any traffic. And that's just Yeah, some crazy

numbers. Um you're absolutely right. AI generated content is flooding the web. Now, of course, we know that we know that this can work and there's plenty of people who have made the, you know, just using all kinds of programmatic SEO approaches and and winning and doing well and hats off to them. It's it's great. But if you're what I would say really would be the underlying approach here would be to work on the building PR. We want to be promoting the brand. So that at the end of the day, we want, you know, Google loves brands. Google's kind of got a special place in their heart for brands. And so we want to win the heart of Google by by promoting the brand. So I'd be thinking, you know, the long game would be to to sort of use that content marketing, use that digital PR and work on getting that brand exposure in front of people alongside the link, you know, and getting branded links and that kind of thing. Excellent. And uh you spoke about the knowledge graph last year there was kind of an

explosion like a 33% growth of the knowledge graph data vault where it also opened up brands to a great deal as well as other types of contributors. So that co-citation type of references the the different brand references are being valued even more in the in the knowledge graph space. Would you agree? Oh yeah. Yeah. Absolutely. um even to the point of um you know help it along a little bit with your entity and your EIT and do some schema and help it along and you can get some good results. Absolutely. Another question from Natalie Ford to to finish this off. How do you envision local SEO link building changing in 2025 as search engine results become more personalized and location focused? Maybe one one little takeaway would be, you know, uh if you're a local business in your local neighborhood, uh maybe get involved in a local charity and, you know, sponsor them and and do something and get a link that way. This stuff still works, right? But pick if you're

going to go and I mean directory, uh things like your your citations are still still really good thing to do, but get on to the quality directories. Vet them. Don't get onto any spammy looking direct. I mean, the the last thing you should do is be doing bulk directory submissions. I'd be just picking out, you know, a good maybe dozen of good quality directories to be submitting your NAP and your name, address, phone number, and really, but work on your obviously we're talking about Google business profile here. Of course, there's the Bing uh places, but what when we talk about the Google business profile, work on optimizing that as the center of that. you know, the descriptions, the services, the photos, update it with posts, keep it current. Um, make sure you're interacting with people who are leaving reviews and uh so yeah, there's a there's a whole strategy right there, but certainly local link building is not going away. Um, but just again, don't be spammy. Well, Peter, it's been a pleasure listening to you

and thank you so much for sharing what you sh at 2 am in the morning. We certainly you kicking off this presentation, this conference today. So, thanks so much and everybody follow Peter on social and uh your deck is going to be shared after the uh program as well. So, thank you so much for your time today. Thanks Aaron. Sarah, it's a pleasure to connect with you again. Thanks for coming on and for this presentation. Thank you so much Aaron for having me today. You're more than welcome. More than welcome. Well, tell the listeners and the uh the watchers of the video a little bit about yourself. I've been doing SEO for 10 years now. I'm I've been speaking a lot. I'm very active on LinkedIn. Uh if you follow me on LinkedIn, you've probably came across one of my SEO riddles. That's about it. Well, you've been certainly participating over the last 10 years and you're show certainly showing uh giving a lot of great information in LinkedIn. So, make sure you follow Sarah on LinkedIn. But at this time, I'm just going to hand it over to you. Go ahead and present to us what we're dealing with in the LLM arena. Thank you. Hey everyone, thank you for being here today. Basically, we keep hearing a lot that SEO is changing. That's actually

not true. SEO has changed already. This is our traditional way of doing SEO, right? We are focused on Google and the 10 blue links. We have heavily checklisted workflows uh for everything. And optimizing basically means we're checking as much items off our checklists following all best practices. And then we also heavily rely on SEO tools uh for our data insights, analysis and so on. Our focus is on three things or three areas. the technical, the on-page and off- page that's the classical focus and our the tools and the data we use as I said traditional SEO tools and the traditional data we have like G4 Google search console offtheshelf crawlers tools that are ready for us to do uh specific task obviously Google sheets here's the thing there is a tornado and things have changed a lot we have the API leaks and uh department of justice documents uh they have revealed so much to us LMS and RAG. Rag is like um retrieval augmented

generation which is basically retrieving the documents and then generating based off that document which makes process more grounded in facts. So u an AI overviews is not we're not sure 100% but it's probably using rag. Uh so we also have the AI overviews and search now and we have also the user that their behavior changed. So users behavior changed in the way they search. they they don't necessarily start their search at Google anymore. Maybe they discover a product on Tik Tok or on Facebook or uh somewhere else, right? And then the type of content they also consume, it's no longer blogs or this is not what they expect or only want when they go to Google or any other place. So there's a lot going on. Uh a lot of that is because of AI, right? A lot of things are changing and have changed and we need to adapt. What does that mean? How does the new SEO look like? So, we're no longer just focused on Google. We're also focused on LLMs, uh, YouTube, Tik Tok, Reddit, Kora, anything that's

taking a space in the search results. Any platform should be part of your SEO strategy. It's no longer about your website and competitors websites. We're talking about other platforms that also you also need to incorporate into your SEO strategy. Yes, we do still need uh checklists, but we also need to make sure that our processes are agile. What do I mean by that? We need to be able to try new things, adapt quickly uh in a balanced way, which I'm going to discuss at the end. Uh but there has to be some waggle room. Don't have th those solid strategies or solid checklists where this is the only way you operate across all clients, across the search results, whatever there is. optimizing. Yes, it means following best practices, but it also means testing things out and see what will work for you. Again, I'm going to touch on this later on with uh the uh custom click uh through rate curve. And then just something to remember that a best practice or any tactic, the more it is used, the less impactful it becomes,

right? And that's why you need to test things out and get your own insights for your own business, right? We still we're going to use the SEO tools obviously for keyword research and so on but we also need to make our own tools and this doesn't have to be very complicated or anything. Uh again I'm going to discuss this briefly uh later. The new buckets we still have the three main focus areas technical on page and off page but we also now have CRO user experience. We have discovered from department of justice documents and from the API leaks that user behavior is important. It's it's probably not probably it is the magic behind Google search and then data analysis. It's no longer just what the tool is telling you and those highlevel insights. Now you need to do your own data analysis and sometimes come up with even a new new matrix and then automation obviously if you're not automating you're missing out. Testing as I said it's essential part of what you do. Last but not least the tools and data. We're still going to use the traditional SEO tools, but then we need, as I said, custom DIY tools for data

analysis and automation. LLMs. You'll be surprised how many people do not use LLM in their day-to-day. I'm not sure when is like more people should be doing that. Uh should be using LLM in their day-to-day. However, we should also know very well when to use LLMs and what for and not just use them like for for everything. Of course, uh I'm going to touch on that a bit later. And then some of the tools we should introduce to our traditional tech stack. Some of the tools that should be normal for us to use stuff like Microsoft Clarity, Hot Jar or Lucky Orange. Uh Microsoft Clarity is free. Uh which is surprising probably collecting data. Uh anyway, so um these tools will help you uh analyze user behavior on a page and therefore make optimizations that would help you rank better. Let's look at that a bit more in details. So our old playbook is like the Titanic. It's sunk. Okay, here's the code for Michael King. I

really like it. Search as we know it has been irrevocably changed by generative AI. It's changed. Our old ways are not effective anymore or are not enough anymore. We need to up update our processes, update our mindset. So let's talk about user experience. Okay. So, user experience definitely changed the way we look at SEO. So, Department of Justice documents and the API leaks both pointed to many things that show that user behavior on search in in the SERs and on the page itself matter and would impact your ranking. Right? Some of these things have been actually said long time ago by people like Ran Fishkin, but a lot of people would push back and they didn't believe this. Now, we know these as fact. So Google doesn't understand the documents. Google imitates the behavior of understanding. So Google's challenge is basically trying to tell you like you're you're ordering a cake and Google is trying to serve you the best cake without actually

tasting the cake. So how it's going to guess which cake is better? Maybe okay, the color, does it have fruits on top? like the size, the weight, and then it's going to serve it to you and take your feedback, store that and learn the next time it serves to the other user and so on and collect all of this aggregate data and make a decision decision on okay, this cake is more popular than that cake, therefore this cake probably tastes better, right? Without tasting the cake, which is it's a challenge for Google, of course. So, user behavior is essential and some of the things were that were uncovered things like nav boost and glue. So now boost is basically how users behave. They specifically click on results in search, right? And things like short clicks and long clicks, which is basically if a user clicks on the result, goes to your page and then bounces back quickly or do they stay spend more time? Yes, we have heard over and over that bounce rate is not a ranking factor. Maybe the the measurement here is not called bounce rate. Maybe it's called something else. Maybe the way they measure it is a bit different than the way bounce rate is

measured. However, the concept is right. The amount of time a user spends on a page is important. We've learned that Google uses Chrome data and quickstream data to determine overall site quality by looking at user behavior. So all of these things show us that user behavior are very very important. Probably the most important factor when it comes to ranking. Think of ranking as three things. What the document itself, the content, what it says about itself, the backlink profile or the the authority of that document, what other people are saying, and then how the user interacts with that document. So this is the most important element of ranking generally speaking. So okay the next question would be hey Sarah what should we do do about this? Okay I have this example from Erica and I love it. So what did Erica do? Erica basically used Microsoft clarity to um analyze the behavior of users on a page and she noticed that uh they scroll to a like when users land on a page they usually scroll to a specific section of a page or use the table of content to jump to a specific section. So, so she all she did literally was

restructured the content and pushed the content that users look for to be more at the start of the content versus having it at the bottom. And within 2 weeks, she have seen 50% increase in traffic. This is big. She had not done any other changes. Just by looking at the behavior of users on the page, she was able to improve their the the traffic to the page. So, why are we not doing this? Why is this not part of our SEO audits and SEO work? We need to have access to those tools and we need to analyze the behavior of the user on the pages as part of our if you want to call it on page or ideally it's a separate bucket of of and on itself the user experience. Another thing is CTR manipulation. CTR manipulation basically at a high level is when you get bots to imitate user behavior on a search result and specifically click on a result with the intention of manipulating its rankings. So as we said user behavior in the search result is important result they click on is going to impact uh its

ranking. So there is two case studies on CTR manipulation. It's something spammy. I do not recommend of course doing this. We're just going through those case studies to learn from them the value of CTR. So basically one website with CTR manipulation actually moved from 34th to 10th position in 22 days without doing anything else other than CTR manipulation. It's also like shared by some SEOs that one of the ways to recover from helpful content update is CTR manipulation. Again, I'm not saying or suggesting you should do this. I'm just telling you the power of having a lot of clicks and search and and this actually aligns with the Department of Justice documents in the Google API leaks. Remember the cake example? Let's say people are picking a specific cake every time. Google will realize this is the most popular cake. If someone comes and asks for a cake, they're probably going to serve this first, right? The thing is we thought if we improve rankings we will improve our CTR right that's the entire thing we want to go up in search because the higher we are the higher CTR we are but what if it's the

other way around actually the better our CTR the higher we go up in search so this small graph you see here is from it's based off the data from AWR and it shows the estimated CTR for each location in search started position one to position 10 however I searched for what is SEO. This is the numbers for the top 10 results. They're all long form content and they all are directly talking about what is SEO. But if you look at the traffic each page is getting and this is the traffic for all the keywords related to the topic, not just what is SEO, you'll see that surprisingly the like the first result is not getting the most or the highest number of clicks. The third result is actually getting the highest clicks, right? Then the fix is getting the third highest number of clicks. So what does that tell us? It it tell us two things. Um if you go to the first place, you may still not get the highest number of uh clicks and it also says that CTR will take you up in rankings, but there are other factors and this is why even

though a page is getting the highest number of clicks, it's still not at the position one, right? Okay. What to do about CTR? Test different pages uh page titles. Uh run those tests. I've left a lot of links throughout this uh deck to get into the specifics of things. Something that we need to not discuss anymore is whether or not user behavior matter. It matters. The question is how do we deal with this? Okay. Second thing is SEO isn't just about Google anymore. If any platform is ranking in search, then it should be part of your strategy. Okay? So, there's a lot of search engines and a lot of other platforms. So if you look at this chart for example, Google is the most traffic referral on the web. However, duck.go is as as well this the second. So why not also look at what how we're doing in duck.go uh and monitor our performance there because you know uh maybe we have a higher chance on on ranking for a specific term

on that uh search engine versus Google. And another thing is in the era of zero clicks which is basically people seeing the results in search or maybe they go to a platform see the result and leave right and a platform I mean not your website Twitter right Facebook Reddit LinkedIn all of these things we need to run a tight ship we need to make sure we're optimizing our SERs our CTR and SERs and we're optimizing our conversion rate to make sure that every click that comes to our website, it counts. And then we also need to le work with the social media people to leverage all those other platforms in a white hat non-spammy way. Just to be clear, I'm not promoting spam. Again, I'll I'll repeat this again. If any platform is appearing in SERs, it's now part of your strategy. We will leave no stone unturned. So something else in addition to this, um Google likes video because users like video, right? So image is the second largest search engine. And then

YouTube uh or videos generally speaking are very popular and YouTube is the third largest search engine. And in a recent study by Rand Fishkin, YouTube keywords, branded keywords are the most searched brand keywords on Google. So if you're not there, you're definitely missing out. Uh it has not been any easier to create videos with AI right now. And the same way we recommend blogs, we should try to promote for video production. And I understand that's a blocker for some uh businesses. They don't create videos. But another thing you can look into and I usually do if I'm working with a client is if they have existing video assets on YouTube, let's optimize those and let's pull all the videos of their competitors and and analyze those and see what we can learn from there. Remember why video is important. It ranks on Google SERs and then it ranks in the video filter on Google and then you will also have a chance to rank on YouTube. So it's three areas, right? versus your blog is mainly for Google. Um, so the question is, are

you auditing your client's YouTube channel and their competitors as part of your strategy? You should because there's a lot of optimizations opportunities there. Um, if you ask, okay, how should I optimize video for search and for ranking and to get more traffic? That's what you need to work on and create your own checklist, right? Okay. Something very interesting I came across. Perplexity is pulling answers from inside a video. So I searched for no makeup makeup and then I found this like it's it's exactly pulling the the this answer this segment of the answer is being pulled from a YouTube video which is very interesting that shows you how much this is important if you're not on YouTube if you don't have video content you're missing out big time so tell me again why video is not part of your uh content recommendations right and I as I said I understand that there may be push back from businesses because it's not easy to create videos it is easier today than it used to be in the past. That's one thing. And then the second thing, you can go for more creative ideas like white uh Louis whiteboard animation videos or whatever. It's an area you should explore. I know

it's not easy to get buying for that, but as I said, our don't leave a stone unturned and our role as SEOs is to give recommendations. In the same time, in a perfect world, if we have everything we want, this is one thing we would ask for. Okay, so FYI, in AI overviews, the most frequently linked URL was YouTube again. So again, if if we're having trouble with all this AI tornado coming our way, YouTube again is one of the most frequently linked URL. Let that sink in. And then another platform we should look into or platforms, Chad, GPT, and LLMs, of course. So in search to rank in in Google, we're talking about keywords. In LLMs, the currency is different. Again, Renfishkin has a lot of cool stuff there. The currency of LLM is mentions across training data. This is why I like this made me pause like why people want to block LLMs from crawling their websites. Like if you block LLMs from crawling your website,

you're not going to be cited there. It's not about oh, I don't want them to take my information. They're going to get it from somewhere else and site someone else, right? So, this is something you need to think about. one thing uh you know how we talk about internal linking now we have internal brand mentions are you you I I've seen a blogs for example where they discuss a topic and they don't mention the brand within the the blog heavily relying on the branding of the like you have the logo you have the navigation you have like you know it's obvious what brand this is but is your brand mentioned in the content where it should be so that's um something to think of again don't do this in in a spammy way. So, we were talking about keyword stuffing. Now, we're not we don't want to switch that to brand stuffing. Don't do that. But just make sure that I mean it's generally speaking a good practice to link to your products. For example, in blogs where it's relevant, right? Uh and it's also part of CRO and increasing your engagement on a page. So, everything is connected together. And then brand

mentions generally and in in training data are the new back links, right? So let's take Reddit for example. Uh we don't like Reddit because it's taking so much uh real estate on Google, but it's also making up 4% of the training data for GPT3 for example, but it has 22% of weight or influence. Do you know how often your brand is mentioned on Reddit? Uh what type of mentions is it there? Are people complaining about your brand? Are people asking questions? uh you need to have some uh legitimate presence on Reddit as a brand, not in a spammy way. We're going to touch on this uh a bit more. So something to think of, right? You need to audit like site operator Reddit and search for your brand, right? You need to see this is training data. You need to be there and you need to be there in the right way. Uh another thing is pull pa data from search for your main keyword and put them into Chad GPT. analyze the answers and see okay uh let's say best agencies in Toronto and then you plug this in uh Chad GPT and you get the results look at the brand

mentions of those brands the brands that Chad GPT came back uh with and where they're mentioned where they getting links and see if you can get the same for yourself and this is where something like automation would come in handy because you don't want to do this manually right you want to plug in a list of paas pu push them to chat GPT pull it back uh with the answers and analyze this uh automatically. So this is something that can actually be automated and that's why I'm saying we need to have our own uh tools DIY tools because if we rely on existing tools we are heavily uh limiting ourselves. So this example from Chris Long it's really really good. So Chris said they've noticed that at Goofish Digital they're starting to get uh business from referrals from uh Chad GPT. they went and did search for the digital PR agency there and notice that CHP is listing um some agencies and their agency is there but other agencies have more information which is basically notable clients that they don't have that was like okay uh

obviously they want to look good in those results as well so they went back to their website found an article that they know that GP is citing uh regularly and added a section about multiple clients there and What happened is that within a week their the chat GPT answer for for their agency included the the notable client. So yes, you can influence shad GPT answers and you should actually make sure that if you're appearing there, you're improving your potential to get referrals. If you're not appearing there, fix that. So this is something part of the new era of SEO. It's, as I said, it's no longer just about Google or just about blogs. You have video content, you have Ched GBT and so on and other platforms and AI overviews. Everyone's complaining uh zero clicks or it's not we're not gonna you know it's going to take away our clicks and so on. I see it as an opportunity versus a threat if you understand how it works. So we think it's like keyword AI overview views pulls the direct match document of those

this keyword and then generates the AI overview which is not the case. So AI overview is a system that doesn't select search results based on one specific query. It can extend to related keywords, recent keywords, imply keywords. So it has a lot of other ways of selecting documents, right? It it does not necessarily select the top 10 results as we all know. So this is an opportunity because you can be ranking on second or third page, but you're still appearing in AI overviews. And this also tells us that the pages ranking in the top 10 are not necessarily the best pages in terms of content. If if AI overviews is typing pages on URLs on page three versus a result in page one, that means th this one actually has better content. And the reason those pages are ranking on page one even though they're not cited in AI overviews are other factors than just the helpfulness of the content. Maybe authority or CTR people like that

result. That means also skyscraper where you pull the top 10 results and rewrite them is not really a good idea to appear in AI overviews and it's not a great idea anymore. You want to have something unique. You want to have something different. So if the top results lack diversity, it can extend its search beyond that. And uh something that has been shown as well that there's a direct relation or there's a relation and it's debatable. This is debatable. However, there's a relation between the uh the appearance of featured snippets and AI overviews and there is some overlap. So if you try to go and feature snippets are uses an AI system by the way. So if you try to appear in feature snippets, maybe that would help you get an AI AI overviews as well. Something to think of. This is something that needs also to be part of your strategy. How are you going after AI overviews? How you understand why a URL that's ranking in the first page not linked to in the AI overviews? That's because simply it's not about it's not the the ranking on the first page is not just about the

content contrary to popular belief and then uh hacks may work and we will come al across a lot of hacks and the more they are used the less effective they become. So here Cyrus came up with a mini case study and I find it very interesting. uh he had a page that's ranking on the first page and he wanted to get it cited in AI overviews. So he took the answer to that question or that keyword and from AI overviews and written something that's similar to it on that page, right? And he was in the top 10. So he got cited in the AI overviews. However, he did lose the rich snippet. And yeah, that's that's a hack. Of course, uh you can try it. However, as I said, the more we try these things, the less effective they are. This is not a sustainable strategy. It may work for now but not on the long run. Okay. User generated content. It's a bit of a challenge and we complain a lot about Reddit, Kora. Recently I've seen post random posts actually from Facebook groups and sometimes just a Facebook group. So um being linked to in search

and it's time we don't look at these as oh social media platforms it's out of our scope. They are in our scope now because they are they appear in search. So we need to think of these as micro sites. You need to have some presence there. Cooperate with the social media team. You need to have some presence there. If there's a a group that's ranking, maybe you need to to have some presence there with legitimate brand content, good quality, and exist there um on Reddit as well. Maybe you want to pull all the kaas from search. So, oh yes, one when we're optimizing content, maybe there's 20 or 50 PAAAS in search about your topic and you don't want to include all of these in your blog. Okay, why not take some of those and create uh some posts on Reddit about them and answer them as a brand, right? So, think of uh you user generated content or social media platforms as micro sites. Again, this needs to be a collaboration

between social media teams, but it's if you look at the screenshot, this is the search results for Tesla Cybert truck. The last thing I would expect to see there is just a Facebook group link for a Cybert truck. Uh like I'm not sure like this is this the most helpful, but apparently this is what people want. So, this is where you should be, right? If you want to con if you're selling, if you're a reseller or whatever, maybe you want to be there providing helpful content. uh not necessarily huge. You know, social media posts don't have to be long or anything and you can mention your brand as well. All of these things are definitely helpful. No one knows all the answers right now. However, it's worth looking into and testing. I stumbled upon this accidentally and Rick Steve Europe. This website is in my opinion a perfect brand. It's well established. If you search for who is Rick Steve on any LLM, they'll tell you that he's like a travel American travel entrepreneur. He's very well well established. Why? He has all types of content, a shop as well there. He has blogs for tips. He has a forum which is

really interesting. I love that. And I I in a perfect world, a forum is a community, right? And in a perfect world, you want to build a community around your brand. Obviously, that's not possible. But some brands can do that, right? Uh and then they have content to watch. They have videos and they have podcasts. So, in a perfect world, you want to be like Rick Steves. I understand there are like uh blockers like resources or uh getting buy in but also remember that if you come with a recommendation maybe you can give the the the stakeholders some uh ideas on how to create these things. So for example you can use notebook LM today to create a podcast on a blog that you have right like it's never been easier to create podcasts or videos uh based off your blog and the concept of create once distribute forever. This is this is definitely worth exploring. So I highly recommend you have a look at this website. It's very basic in terms of design. However, it's very well established because it has a very strong brand and uh brand is also very important to us. Okay, given all the

stuff that we have just mentioned, the new tools and data. So reminder, we are all collectively using the same tools and data as everyone else. There's no competitive edge here. If you're doing keyword research uh for best running shoes, your competitor are also doing keyword research for best running shoes and they're using you know there are few tools or specific tools out there. They're probably using one of them. Same tools, same data that's there's no competitive edge there. And as I said, the more we use something, the less impactful it is. If we're all building our strategies on the top of the same data using the same tools, there's no advantage there. How how can you win if you're doing the same thing? This means we need new data or we need new matrix and we need our own custom tools. Okay, to have a unique edge. What new matrix? Okay, so first of all, there are existing metrics that we didn't use when we're reporting or analyzing SEO. Things like in J4, for example, engagement rate, bounce rate, conversion rate. All

of these things, time on page or dwell time. All of these things we need to look at and start analyzing because they matter when we're talking about ranking and improving getting more traffic in this era, right? CTR. I think it's time we focus or shift the conversation from talking about ranking to CTR. And in one minute, I'm going to show you an example why CTR focusing on improving CTR is going to make you think differently than just improving rankings. As I said, you can improve rankings and still not get that much traffic. And sometimes the effort to move from the fifth position to the first position is not worth the amount of traffic you're going to get. It's like really small and the effort and cost is going to be so high. So you want to work on improving your CTR. And then uh we also need to start monitoring AI overviews visibility. There is uh this you can use this tool in SE ranking brand visibility. I don't know why, but I've seen so many marketers, so many don't divide traffic by branded versus non-branded when they're reporting. I

can't explain how important this is. Uh, and just remember that branded traffic can be a result of SEO or it can be result of other channels. So, there's a lot of moving parts. You need to make this distinction to be able to have a clear view of what is working and your actual performance. One more thing I don't want to skip, how PPC and SEO work together. uh something for you to keep in mind. Uh there's al this myth circulating that for example PPC shouldn't bid on brand words because SEO capture it. Actually studies shown that if you're running a PPC ad you have and you're ranking in the search your overall CTR is higher than if you have one or the other individually. Anyways uh back to uh brand visibility. You need to to monitor organic referrals from social. you need to uh the traffic from LLMs uh share voice and then and com like comparing your brand your visibility with the competitors impressions are also important because the more people see you in search the more brand aware they become so impressions is an important metric if we're talking about awareness if you're

running an awareness campaign in SEO if you're a new brand and you want to establish yourself impressions is a good trick to look at okay and then new metrics that you make yourself based on what you want to measure so for example one of the things I measure is content push that's how I call it I know other people may have different names for it which is the average number of clicks per page right the purpose of this is you want to see okay on average if I create a new blog today how many without doing any optimization following the exact process we have been following so far how many clicks would we get on average and then you would be surprised sometimes 50 clicks per month per page which is not a good number so this is you this is where you can can get buy in hey we need to optimize this we we need to uh improve our content push, right? And then percentage of dead weight pages. You need to have an understanding of how many pages you're just that are dead weight. You're just a burden. You're just moving around with a with a burden on your shoulders. Uh so this is an example for a small client I did in the past total number of URLs that get 100 or less impressions. This can be

your definition. You can say no, I'm going to go for thousand impressions. I'm going to go for a,000 impressions in the in the last 6 months, not 12. I went for 12 months because I wanted like a full cycle of all the seasonality and everything. So this is the total number of URLs that got 100 or less impressions. Then you want to go after clicks, right? Total number of URLs considered and then you calculate the dead weight pages on that website. And this is one of the most powerful way to get buying for your content strategy because you're basically going to tell them this is the content you're creating and 55% of this did not see the light at least because if if you change the like the threshold of 100 this number will increase. So you're telling them 55% or more of your content has not seen the light of day in the last 12 months. Okay. These three matrix that the ones I'm about to discuss they're not mine. They're by a guy named Jason Bulman. I have uh seen it not this year, last year in Toronto IRL and they stuck with me. They're very interesting. Traffic velocity, organic traffic over website

age and then keyword velocity, number of keywords over website age, backlink velocity, number of backlinks over website age and website age starts by when the time your pages start to get indexed, right? uh all of these things uh and may you may come up with your own matrix depending on what you're trying to measure but relying just on keyword difficulty search volume that's not enough anymore on this I I really like this example here uh so we need also to track AIdriven traffic and use this data as directional it's not 100% accurate we understand that uh I linked to the tweet and the tutorial here take this as directional so what happened here is in this example he looked at the landing pages perplexity is sending via to to the website via G4. Okay, this page is getting traffic from Plexity. What type of questions are on this page? And they go back to Perplexity and search for those questions and they found their articles ranking there and this is give you this can give you an insight how you're ranking, how you can improve that, why is something else being cited

uh before you and you're not right. So this is this is really good because and also what structure or type of content or paragraph format that perplexity is pulling from. So getting this granular and looking at those things and doing those measurements it's not part of our usual process but they need to be new tools. Okay. Uh before I go on rambling here if you look at this graph this graph is for my website. I analyzed my CTR across the results and I am getting the highest number of clicks when I'm ranking between five and six position. Right? So contrary to popular belief, if you analyze your own CTR and I've did this using Python and this is why I'm saying you need to build your own tools. This is one thing I built. If you analyze your own numbers, the average or the uh established data may not be it's different than your average or your own data, right? And and again, that's one reason why you should focus shift your focus from rankings to CTR new tools. Uh

as I said, DIY tools to automate your workflows. For example, you can create metad descriptions in bulk or if you're doing a programmatic SEO, uh that would be very helpful if you write good prompts that are uh customized. Uh definitely going to help you. And then data analysis like this custom click curve. But then we also want to use existing tools that are not traditionally part of our SEO stack like Microsoft Clarity. Make sure we're using CHP and LLMs the right way. Uh just for you to know there are LLN specialized. So for coding for example, yes you can use Chad GPT and Claude and Gemini, but you can also uh look for coding LLMs specifically designed to help you code for example. So whatever you're working on, try to find if there's a specialized LLM for it or you can just use strateg uh as I said a lot of people don't use them enough yet. Uh you're as good as your prompt. If you are trying to generate uh whatever you're trying to generate, you're as good as your prompt. Avoid using like learn from other people's prompts but don't necessarily

use the same prompt. and then when it's good to use. So there's this recent experiment or uh thing that has been run by hrefs the SM O and uh BB where they were judging Orion. They had two people Ryan and another person. Or Ryan has 14 years experience in marketing and the other person is a newbie and they were using Chad GPT for challenges and um uh Semo and BB were judging. I have a link for the video here which output is better and what you will learn from this is chat GPT is good when there's a lot of ambiguity for example PPC copy but not for a blog post or social media post they the it's not going to look good but PPC copy it's really small it needs to be very precise definitely go for it and when there's a lot of ambiguity in the input like you don't have enough data to come up with a concept maybe you can go to chat GPT and see what it has to say just make sure that it doesn't limit your thinking. Right? So to wrap this up, what we know for sure so far, no one

has all the answers, the only right answer is to be agile. Uh be open to new ideas, test fast, be okay with failing because you will um incor incorporate other strategies in your uh uh other platforms in your strategy. Brands connect everything. User experience has to be part of your strategy. Always be analyzing, always be testing. and more. But remember, the more a tactic is used, the less effective it becomes. Yes. The last but not least, don't take radical measures. Don't say, "Hey, I'm gonna block LM from accessing my website. I'm going to drop using schema on." Don't just take those um radical measurements yet, right? Without like really thinking things through. And don't be casual with learn with learning. Things are changing. This means you need to dedicate time and have a structured way of learning what's happening and to think even what you should be doing about all this. And that's it. I've been rambling for a while. So maybe I should take some questions now. Thank you. Thank you very much Sarah. That was a wealth of information that presentation

and we're certainly going to have that deck shared as well. Um we have a couple questions and we do have a bit of a time wedge here. Though I I do want to bring up uh one recent question from uh Steven Mr. SOS Solid. He actually asks pretty much impossible to focus on all the points when working with smaller brands and smaller resources. Where should what would your advice be to focus on for the best results when when we're talking about L&Ms and and visibility in in these new spaces? So small brands actually have an advantage of being able to move fast, right? Like don't think that just because someone is a bigger brand and they may have higher much much more resources which by the way not necessarily the case they will always out win you they they it's not the case because you are able to more be more agile and experiment with things faster so you have an edge there when I'm where to focus there are things you can focus on that won't cost you a lot like u as I said Microsoft clarity is free you're going to compensate with more effort if

you're if you don't have the the resources so the less resources you have We're going to do more effort to make up for that. All right. Very good. Very good. Thank you. One more question and then then we'll finish up. Um, what are some tools in the market that can help optimize SEO for LLMs? Baron Kelman says, um, okay, I wouldn't say there's a specific tool yet, but I think this is the tool that would help you to do that. The there's a lot of manual work you'll have to do yourself. You need to dig into search l use lens and search and see what's happening there and analyze on your own. I don't think there is something that's doing that yet. Yeah. Yeah. And uh you give an example of researching uh GA4 to be able to see citation references and deferrals, but that's almost reverse engineering. You really can't dig into chat GBT to be able to see all the citations that you have. So, you really have to reverse engineer from the questions that are answered on the pages to then be able to see what's really populating in in in the LLMs right now. Is that correct?

Yep. Of course. Excellent. Well, thank you so much, Sarah. Like I said, a wealth of information and and I'm certainly going to dig into there and see a lot of the links and sources that you put there. So, thank you very much for your time today, Sarah. Thank you so much, Erin, for having me today. and we're talking about mastering topical authority. How to become the go-to source in your niche for 2025. Hey, this is Dixon Jones. He's the CEO of End Links. Welcome, Dixon. Hello, Erin. Hello, everybody. You're the man, the math. It's great to be able to connect with you again, Dixon. Thanks so much for joining us. Thanks for Thanks for having me, Erin. And thanks for having me. Search Engine Ranking as well. Uh, I really appreciate it. Absolutely. Dixon, can you tell us our listeners, our watchers, a little bit about your history? Yeah, sure. I mean, I'm going to go to into it a little little bit about my beak my murky past in the presentation itself, but uh most people know me uh in my past as being marketing director of Majestic. Um I before that ran founded an agency called Receptional, which is still going today.

I'm not involved in it sadly. I was at an awards ceremony last night and they're still winning an awards. So that was that was good. And then uh the CEO of Inlinks, which is uh an entity based SEO technology and uh it's kind of like the new kid on the block really um in terms of uh approaching SEO. Absolutely. We're huge fans of Inlinks over here at Edge and Site Strategics as you well know. Thank you so much for what you've been doing on that front. Hey, I'm going to give you the mic. Go ahead and give us a presentation with the hat. I'm going to be talking about what makes you a master of a topic. Clearly, PowerPoint is not my mastery, but I'm going to be talking about how you determine when you've covered a topic enough in your internet presence versus when you're beating a topic to death. And I think it's a very difficult thing to to get right. One of the things about me is that I'm old. I just turned 60 last literally last week. And one of the things about being old is that you have the perspective of time when you're trying to uh look at your marketing. you've got the advantage of a lot of

years behind you. So I've been 24 years in in search engine optimization, 10 years before that writing content. So 35 years really. And the the the content that I used to write was in 1989 to 1999 was writing murder mystery evenings for a living. So that's kind of what I did and that was my niche before the internet came along. And it's an interesting uh story uh about how you connect ideas when you write murder mysteries. And it plays to what I want to talk about today because the content as I I I wrote as murder mysteries really connects thoughts and connects ideas and they have to in order for people to solve plots. Uh and so I find it quite an interesting transition for me and is going to play into the talk that we did today. But one of the things that I've learned from this perspective of time is that on the whole content planning tools are broken. Keyword planning tools particularly are broken. And it doesn't really matter whether um we're looking at keyword planning tools

that are machineled or whether we're looking at keyword planning tools that are humanled. They don't really work. If we go to content planning tools or keyword planning tools, traditionally what they do is they look at um keyword volumes, they look at keyword variations and they really don't look at context. So the keyword volume information doesn't really help if you're um looking up the word Mustang. You know, there's no concept of whether that's for a horse or whether for that's for a car um or whether it's for an airplane. If you're looking at keyword, you know, it comes up with all these keyword variations that are really the same thing just with an s or with ing on the end of it and and variations of the same same word that doesn't really help us. Or they come up with um suggestions of words that you just would never use in your business, synonyms that are irrelevant possibly, or more importantly, to get to Steven's question in the end, they suggest things for stuff that your business just doesn't do. And that's

because they're all out of context. And I think that this is a real problem. So keyword planning tools um particularly uh are not working very very well and it doesn't really matter if you put away all the machines and start to go to humans and you start brainstorming when you with with content planning because when you do that you tend to have human bias coming in. So you don't you think of things from within your organization not from the point of view of possibly your your audience your wider audience. So you come up with creative ideas to get around that. So you might use something like Edward Deono's lateral thinking puzzle where you sort of throw an orange into the loop and say right let's try and think up keywords and ideas that that involve the word orange but that sends people away from uh what search is designed to do which is to narrow a user's focus to the thing that they want to get. If you start creating things out of the blue then you you are actually working against that idea of of getting the user to the content that that they want to find. Or another tool

that you might use as a human is mind maps which you think are very very um dynamic in nature. But actually when you uh put a a mind map together it creates actually quite a linear system. So you end up with websites like this where you kind of talking about you know your navigational structure really. So if you got a DIY site you talk about parts and then you got screws, nails and washers and then power tools and you got them bent and hand tools etc. and they're all very linear and it's kind of typical sort of design as a site but doesn't really work to the way in which uh we want to develop uh our keywords and our content and this has some implications for the web and for our businesses. This um inability to create relevant keywords, the ones that you need to write about in the right context and at the right time has a huge number of problems at a macro level. Google is seeing loads and loads of websites and we've all done it um if we've been in SEO long enough that are

creating duplicated, uninspiring, repetitive content. A good example is you and you do plumbing calls in Daytona and Ohio and California and New York and then just have a page on every single one of those towns and really it's the same sort of content exactly the same uninspiring repetitive and that creates a massive amount of volume that Google has to crawl and I think that that is one of Google's biggest problems after lawsuits but go leaving Google aside for a second it's also got huge m implications for us micro implications for us as businesses because it doesn't really tell us when we need to create a new web page. Uh or when does a topic warrant enough need for our business to create multiple pages for us to be able to be seen as an authority around that topic? Or when do we delete a page that doesn't have value to our business specifically? Uh it may have value to somebody in the world but maybe not to

our business. So when do you delete? when you when do you augment uh and when do you uh and when do you need to start creating things from from scratch? It keyword keyword planning tools don't really do that at the moment. Um and that's what I want to want to talk about a bit today because I think when I wrote murder mysteries uh I tied ideas together in threads in stories. So, I would uh tie a tie a murder mystery together by by uh connecting connecting um dots. And I want you to think of your content in terms of connected scenarios uh with the magic being in how you connect those ideas into topic clusters because that is what's going to take you out of just writing around keywords and starting to connect those topics and ideas together. And there are a huge amount of uh things out there that have looked at um modeling topics and modeling ideas. So there were um a lot of uh algorithms out

there uh around um connecting um topics uh through words. Um so things like LDA uh LSI and engrams particularly engrams for Google are all methods for crawling large amounts of data. i.e. the internet um and connecting ideas but they were all um keyword focused and not topic focused um so so uh later on people came out with something like T TF and topical page rank which is something that Google doesn't talk about very much but something they had um and these are these are topic based but there's been a lot a lot of work around trying to model topics and see what work what topics are semantically close to each other. What's interesting is that nobody agrees on what a topic is. Uh and so uh there was this study in 2019 which looked at I think it was 64 different um market

research paper or research papers all about topic modeling in marketing. So you might think that topic modeling is a a new phrase to marketing, but there are already 2,00 64 research papers about topic modeling in marketing in re back in 2019. And this meta study kind of looked at all of those and tried to find uh commonality. And what it came out with was all 64 papers were really using pretty much different definitions of what a topic was. And this is a challenge for us in the search industry, in the content industry, because unless we can all agree on a common definition of a topic, then every time anybody comes out with some research, they end up with different underlying data and it doesn't really work. And this is where entities come in. This is why entities are so important um from a from my perspective. So I want to spend a couple of seconds on explaining the difference between entities and topics and keywords. Keywords are things like, you know, any way to describe any kind of

topic, it's words really. Um, but a topic is quite a specific thing. So, a topic, let's take the Eiffel Tower. When you see the Eiffel Tower, you know the Eiffel Tower. You know that it's that big metal thing in Paris. It doesn't matter if you say Eiffel Tower Lator fell or show a picture of it. It's all the same thing. Now what you need in order to make uh topic modeling really work is have a single definition of where the truth is for that thing. So we use at links we use Wikipedia uh because it's an independent source of data. So an entity for us is the Wikipedia entry for the Eiffel Tower. Now, I admit there's one in English and one in French, but it's got a database that explains exactly what the entity is. And what's really important is that it's um not us that builds out that database. Because if I was trying to as an SEO,

you know, optimize through entities and I controlled the underlying definition of entities, the very first thing I'd do is make Dixon Jones an entity and break the system. And that's what SEOs do. We're very good at breaking shiny objects. So we need a third-party system to say this is what an entity is. Wikipedia is great because we we it's open source. Uh but you could use you know the Oxford English dictionary or Webster. Google has its own knowledge graph of course. Um and these entities are basically uh records in a big database that will explain one entity. But you could have many topics uh that are that are referring to the same entity. But you can certainly have many many keywords that refer to um multiple topics as well. But an entity is in a database. And I I want you to write your murder your your your content on on websites and think about your content in the same way that I write about murder mysteries. So when I used to write my murder mysteries, I would have a a newspaper that everyone had. Um and that would be kind of like the front page of

the website really of the the flavor of the flavor of the day. Then I'd have a cast list because every guest at the at our at our murder mysteries was a character in the plot. So everybody was a was a character. So it's kind of like your site map just a little bit of information about everyone that's there. And then we had each person had their own character and it's page of A4. So think of it as a page of information on on your blog for example. But your character had all sorts of different plots. So you so Linda Love Lace knew something about the vicar, the vicar knew something about the police officer, and the police officer knew something about the lord and and and they kind of had lots of different subplots within there that explained what their motivation was for the evening. Topic modeling is like having a rounded murder mystery plot because each of those characters dips into multiple plots. They don't just talk about themselves. They have to know about other people in order for the game to work. in order for information to flow. Each subplot has a kind of little focus and it may or may not have its own web page uh or

character, but the plot has a focus that involves lots of the characters or lots of lots of pages um in the al in the analogy. Uh and knowing the subplot helps to solve the larger plot. And I want people to write their content connecting dots in that kind of way. In other words, I want people to model topics for content through entities by doing three things. Firstly, I want you to create a database of entities on your website, things that you're talking about on your website. So, we're going from the from the esoterical analogy here to to to the uh to the factual stuff. I want to create a database of entities on your website. What is your business already talking about? page by page if you can and then build out a map of all the things that you're talking about and then you can throw all these topics into something like a Google suggest or also asked or keyword insights inlinks does this for you. So good in links will throw all of these ideas and these topics into Google

suggest and it will generate new ideas new keyword suggestions if you like based on the ones that you've already talked about. So, what we're doing differently here is we're not coming up with a keyword suggestion plan based on volumes of people that are talking about uh things on the web. We're coming out with keyword ideas and suggestions that are semantically close or extrapolations to your existing business. So, you model the topics um and throw them out. If you're going to throw them into Google suggest, then what's going to happen is you're going to get a lot of things coming back that are semantically long way from your your business. So you if you start with Mustangs, some of the things about are going to be how much do you buy a Mustang for? And then another one is how do you tailor Mustang? Uh and one's got to be horse related and one's got to be car related. So you've got to remove the ones that are not semantically close or relevant. Uh and then you can model your topic through through entities. So that the thing that you do third is you take that output and you create topic clusters. So you find

clusters of concepts that are related to each other so that you um are not just writing a page about this, a page about this, a page about this. You're writing about a group of ideas and so you're going to get a database of ideas. So why is this different? Why is this different? That's the conceptual bit of how you do it. But this way is built on developing your business. uh it also removes human bias as well and it connects concepts and helps you go deeper into where you need to go. Importantly also it avoids too much repetition. It should avoid that problem of writing about the same thing five different times with slightly different um keywords try that you're trying to optimize for. Uh it should be very much topic focused. So let me take an example. So team building games and murder mysteries was a version of team building games. So I took a popular keyword tool uh and out is spat a huge

number of uh volumes around uh different different phrases here and the intent is very very narrowed informational commercial transactional you know these kind of things that it's you know whereas for me intent is not about is not about that at all and we'll see in a couple of slides why we're a million miles away from commercialformational that kind of idea. We've got a bunch of keywords uh and and a bunch of uh keyword volumes and uh and and some keyword difficulty. So the spreadsheet is well known. We've probably all seen these. Uh it's a well-known sort of uh output, but it's it's not really in context. So when you start writing about that, you are going to be writing fixed pages. But what I want to do is you come up with websites that are a bit like this basically where you've got got your parts and your screws and your nails and you've got a page on screws and a page on nails and a page on war washers and then a section on a pillar page on power tools and source and drills. That's not how we as human beings understand con

content. We listen to stories. We want to we we we want to take in concepts and ideas uh a bit more of a tumble weed sort of idea where you know you hear a story and it threads through the different ideas. So instead of talking about screws you're talking about problems that that people want to fix. How to uh hang a wall hang a door on a wall for example. So going back to team building games, if I type g team building games in to Google in the UK, you can see how Google is starting to move away from understanding individual pages and really trying to do two things. It doesn't like that I've typed in team building games. The first thing it's done here on this screen is it says, "Hey user, who are you? Are you asking for team building games for adults, for work, for students, for kids?" It's trying to narrow it down. team building games is too broad as a concept for Google to come back with uh meaningful information. And I found this really interesting when I saw this here.

And then even if you're not going to do that, it's going to give you uh I've expanded this so it's taken up the whole page, but it's it gives you a bunch of examples of team building games. Scavenger hunts, escape rooms, you know, um uh murder mysteries are in there. So Google is is is saying firstly, I need to know your audience before I can do anything here. And secondly, team building games is very generic. And I think you actually want a team building game. You don't want to you you don't want the concept of team building games. You've got to dive into something a little bit more specific here. And these are very topic focused ideas. What we're trying to do moving forward with uh topic modeling is to create a system based on the principles of understanding your website, extrapolating your website topics and concepts, taking more ideas from from Google or suggest or other data sources to extrapolate from your existing ideas to come up with a topic plan that is going to cluster ideas into sensible suggestions. So around the

concept of team games for example we've got a volume of how many people might be talking about team building games but you know maybe we want to write about 10 pages in there and we've only written two of them. So what we're trying to do is to automate this process very manual process of um of diving in finding out what are the gaps in your topic knowledge what do you need to write about to complete the circle of writing all the things that are relevant to your business but nothing more but nothing more because anything more starts to dilute the authority in the things that you want to be authoritative in then saying right okay we want to write about team building games then basically we need this stuff about outdoors or we need this stuff about uh a sub subtopic of youth games for youth for example. So we're saying right you can see on the right hand side here our topics our subtopics within the concept of team building games maps quite nicely the audience that we had here at the top of

this slide here we got for adults for work for students for goo for for kids online and the way that we ended up coming out we came out youth kids ideas outdoor employees etc zoom uh because uh obviously people were uh doing a lot of things during covid and so zoom ends up as a subtopic as well So we're trying to create concepts that are extrapolating from the original idea. So if we go back to a more traditional example of the DIY site, it means that when we can start writing content, we want to not just write a page about screws, a drill, paint, putty, whatever it is. What we want to do is write a story that solves a problem. So for example, we might want to take the story about hanging a door. How do you hang a door, a new front door? And you might talk about drills within that. You might talk about masonry holes and raw plugs and drill bits. And in doing so, you connect a story that makes it obvious to a machine that you are an authority on choosing the drill. Of course, you are going to need landing pages. You're going to need

a landing page for the drill. You're going to need a landing page for the uh for the raw plugs. So, you do want to say, right, okay, this screenshot here is around keto diets, for example. So within keto diets, we want to talk about things like uh best diets, diets for beginners and diet pills, for example. They are um subpages of the concept of uh keto diets. But protein suddenly becomes a totally different pillar page. It's got large amounts of volume, but it's really related to protein drinks and keto in relation to protein drinks. You end up with a different pillar based on the volume. So what we can do is take the relevancy of the term and the search volume of the term uh and the clustering of the term to see uh what we should be writing next and whether it should be a new page, a new pillar page or just an extension to an existing page and you just put a paragraph in about a particular topic. Uh and this is a different approach that I think is is great. Now, it's different and I've

talked about uh about my my my process in an abstract way, but the takeaway is not anywhere near so abstract. The takeaway is is really this. Um so, if you're not a uh if you're not a you know, a murder mystery context kind of person, this is a dry version of what I've just tried to say. Cluster topics as an extension of your current business and your your content. I don't want you to cluster around keyword volumes but in terms of your your existing business. Then I want you to make the topics answer questions and solve problems like hanging a door. And then I want you to define pillar pages for a topic and then create sub subpages off of that topic. But you connect them up with stories which is why you need a blog. Pillar pages are all about topics that are most important to be your business. A subtopic has enough volume to create a new page. But if a subtopic has really low volume, then I would just make it an H2 or an H3 paragraph of a pillar or a subpage. So, uh, what you're trying to

do is not create millions and millions of pages. You're covering your bases. You're doing what once and you're doing it right. You they've got a few other bits and pieces here. Um, I a page can be a pillar page for one topic and at the same time it can be a subtopic for another another pillar page. Now, that's a a bit of a a long-winded way of saying something, but you can essentially have a page about Mustang cars, and that can still talk about uh electric engines versus gas engines, and they can link through to the engine page and to the uh to the electrics page. Uh and that's absolutely fine. I think there's a feeling amongst the SEO industry that, you know, you've always got to link to link into your pillar pages and never out of them. And that's a big mistake because the world isn't like that. Uh somebody can be an expert in one thing and comment about another thing and it's the same in content on the web. And so in internal links complete that referencing between pillar pages and subtopic pages. So internal linking

makes that those links uh makes those connections exactly like the murder mystery plots do. uh and they tie everything together so that uh a search engine can eventually understand what it is that you're an expert in. Another way to say it, if you want color in your content, then the big terms at the top need authority at the root. Get the base topics right, get the underlying topics right, and then you're going to be able to rank for the bigger terms. That's kind of what I've got. Uh if you would like a copy of your own topic wheel um then feel free to email me on this address um preferably fairly soon um and uh mention search engine ranking so that I know um which uh which presentation it was. Thank you so much Dixon. Uh it's always great to be able to hear you go through these key concepts. uh you've taught me for a number of years and uh I in turn have taught my own team the uh steers towards entities and not keywords as the the golden chalice. So you got to reverse this all around and be able to create your own murder mystery on your

own website. Thank you very much for the classic scenes as well. We got a couple questions here. Well, Steven Grales asked a question here. Dixon, how do you spot emerging subtopics in your niche early and build authority on them before competition catches on? I think that's one of the key elements of being aggressive at building authority and and seeing that it's very topical and trending in nature. So, how do you go about going after that? You know what I do right now? I'm going to get hated for saying this. I'm going to hate Yeah, I'm gonna get hated for saying this. So, I've been I've been looking at all the LLMs out out there and uh testing them against each other. And I got a a little survey I'm doing on my on my on dixenjones.com. If anyone wants to jump in and just look at six different LLMs together and and see what uh see which one they think is best, then I'll tell them which one's which one it is. But for Steven's question, I'm not saying this the only way to do it, but it's a new way to do it. Grock is is X's X's LLM. Now, the thing about Grock, it's

not the best. However, it's uses Twitter's live information. So, probably I would consider looking at Grock, which you can get in the you got to you got to pay for your blue dot bit and you get Grock uh in your in your Twitter, whatever it's called. But if you ask um Grock what's emerging, then it literally will come out with interesting things that have happened in the last 24 hours. So to Steven's point, I would say it's a pretty good place to go to come out with concepts, but it's not connected to what I was suggesting we should do in the presentation. Because in the presentation, I think you should do what's semantically close to your existing business. If you want to start writing about something that you don't do as a business, you will fall down. You need to do the thing. You know, you got to walk the walk before you talk the talk. So, if you just went to X and did that on Gro, you'd get some great ideas, but you wouldn't necessarily get pounds

and dollars and cents in your back pocket in the end of that because you haven't got a product to back it up that you can charge people for and you haven't built a foundation of entities inside your site. So those are going to be subtopics of key foundational entities, right? Yeah. Yeah. Yeah. I think so. Yeah. I mean, you've got you've got to follow the money and I and and SEOs are really bad at following the money. It's kind of like we're we're we're happy to get our rankings going up. Uh and then if we're really feeling advanced, we're happy to get the click and count the clicks and then we can count the conversions maybe. But really, it's all about the pounds and pence in the in the customer's pocket. So, uh, so we got to know what the business model is and just not deviate too far from that. Roger that. Um, Ron's got a question for you as well. If Well, with search engines placing more focus on topic clusters, what tools or techniques do you recommend for identifying content gaps within a specific subject? I was going to ask that same question. Okay. So, well, so the way that the way that we do it at Inlinks is is that we first build a a topic a knowledge graph of

your of your own website. Literally, that's the thing I was going to like the saying. If you if you want to get one, I'll send you a topic wheel and of all the things that you're talking about on your own website. And it's it takes seconds if you got an account in in in links. If you want to do it by the page, there's a free tool. You don't even need to log in. Go to inlinks.com, go to the free tools and look for the um for the tool on it. Uh and it'll go through a page and it'll show you all the topics on that page. But what we do is then do it at the site level or 100 pages of the site. We'll usually cover all the the the important stuff. Then we take that data set and then we throw it into Google suggest and see what they're throwing out. And then that comes back with these topic clusters and we can see which topic clusters are semantically close and that's what gives us the gap the things you've already written about and the things that you that are semantically close and there's the gap of the things that you're missing and that's exactly what the topic planner in in LinkedIn is trying to do. It's sometimes uh sometimes a bit of a learning curve, but even when we did it, we thought we were doing a good job of planning our content

and uh it turns out I haven't got a page on SEO tools, you know. That's pretty that's a pretty basic finding really, you know. So, so it's amazing the way you see your see your gaps uh when you uh when you actually stop trying to do it yourself uh and uh and allow a machine to um to fill. Absolutely. Absolutely. It's a great resource. It really is. Isn't it? It'll take you down a few rabbit holes, too. Oh, wow. I did mention this. Oh, I Yeah, that is a problem and I'm and trying to bring you action there. We've just introduced some some prioritization to try and prioritize it as well. So, these are the things you should do first. But yeah, it's a Yeah, it's a mind something or other. Yeah. All right. Lastly, Diego actually has a question for you as well. Hey, Dixon, thanks for the classic slides. Are back links essential for building topical authority or should we be prioritizing on internal linking and content structure instead? That's a great question. That's a good question. Uh and of course I am biased because I have uh such a history with Majestic

which is uh an absolute gold mine for looking at backlink data. But I will be absolutely honest with you. I think that the the word backlinks is probably overused. The word citation is underused. In other words, in other words, what you really need is for your brand to be quoted in context in the right place. So if you get back links or mentioned in things that are totally irrelevant to your business model, you're actually doing yourself a disservice because a machine is going to to say, let's say you've got a a link from or a link a link a link from a comic site, you know, Snoopy, and it's uh linking to your website about uh investment, you know, all of a sudden there's going to be a oh this investment company is about comics and it's just going to mess with your head. Now, mostly that's not going to matter because a machine is going to need quite a lot of convincing to say this website is about this. Um, so it's going to need

several citations, multiple citations with um some reinforcement from different locations. But a backlink is a form of citation. You don't have to have a backlink, though. What you need is your brand to be mentioned. It could be mentioned in a video. So it can be mentioned in a in the right context hopefully in the right flavor. So I think um a backlink is a sub is a subcategory of a citation and I think it's citations that are important. Expand your brand get your brand mentioned out there with link or not link but have your brand plus context uh creates that knowledge graph beyond your own site and that's that reinforcing agent right there. Absolutely right. Absolutely. Yeah. Fantastic Dixon. Thank you so much for the presentation and this deck will also be shared after the live stream. Uh huge fans over here Dixon as you well know and we certainly appreciate your time today. Thanks a lot Erin and thanks to theation ranking as well. We have our expert panel the state of AI overviews in 2024 research

insights and future forecasts. And if I could, I would like to have everybody introduce themselves, leading with Anastasia. Give us a little bit of history of of who you are and where you've come from in SEO. Okay. Hi everyone. Uh I'm Anastasia and I'm leading the SEO organic direction here at SEO Ranking. So and having like around 8 years of experience with SEO right now I feel that it's really important for us as an SEO um company and for me as an SEO specialist to really keep an eye on the AI on AI overviews and that's why I really want to talk about this particular subject today and I'm really excited for this Fantastic. Itar, how you doing? All good, thank you. All good. Thank you very much for having me. Um, so yeah, I'm Blau, senior SEO director at Studio Hawk, which is an SEO

agency based out of Australia and the UK. And in terms of how long I've been doing stuff, well, digital marketing I've pretty much done since I was 9 years old. So, it's been quite a long time. Um, and then sort of moving specifically into the SEO industry for a while. But I've I've not been uh an SEO uh any no nowhere near as long as as Barry our our next guest as well. So I'll I'll let him talk a bit about and this guy right here Barry sports we we all know this guy right 41,000 articles 21 years of SE round table. I'm just going to run ahead because I just wanted to congratulate you Barry on that. But what what don't we know about you sir? Oh good question. What don't So I I'm the one that talks the talk. I don't do I don't walk the walk as Dixon was saying. So, um, I write a lot by def. Yeah, I don't call myself an SEO, but I guess by definition, anybody who writes anything on the on the internet is a SEO and yeah, I love talking about search, so I know you got some really good interesting AI data on Google AI overviews and other things. So, looking

forward to this this chat. That's a great segue. Thank you so much, Barry. Um, Anastasia, I wanted to actually tee that up to you because SE ranking has been doing a number of research and analyses on AA overview since the the uh display of that in April of 2024. So, uh, give us a picture of what you guys have been studying over there. Yeah, actually we started our research process like back in January, but we started doing this like seriously when after the official SG AI overviews roll out. Yeah. And uh we right now we have data for like the summer period and versus this autumn November period. And uh as for now like uh as of November 2024 we see that around 20% of keywords that we track and we track a lot it's about uh around uh 100k of keywords that these keywords 20% of these keywords trigger

AI overviews and this data is like almost the same in the US and in the UK. I can say it's a huge increase compared to the August data uh when we saw around 12% of AI overviews in SERs and this can uh show us these dynamics and uh this can also help us to um make some conclusions and make some predictions based on this data. Excellent. Well, that's that's that's a huge jump from 12% to 20% in the order of three three months there. So you have a lot of pred predictions. You also have a lot of industry specific subpoints in there. Can I ask one question? You've observed a strong correlation between longer more detailed search queries and the likelihood of triggering AI overviews. Um they sometimes resemble prompts rather than those traditional search phrases. Right. So how should SEOs approach keyword research and

optimization for these new prompt environments that we're finding ourselves in? I'll toss it to you real quick. Yeah, I feel like the it's interesting when you think about how people have changed in terms of their behaviors. So like search behaviors, we've sort of seen a shift ever since OpenAI sort of I think initiated this shift in search behavior of people utilizing prompts as opposed to more traditional search queries that we've seen. So I think even looking into the data of what we found and in this this research as well, I think it was quite interesting also to mention the methodology here because it's like you took 20 different types of industries, took about 5,000 keywords in each. So you've got a nice sortant sample there. Um, hence the 100,000 keywords in total. And what it seems to me from just anecdotally looking at the the study here is we sort of saw in the beginning the sort of Google's hand being forced into trying to at least provide some

kind of experience that almost not exactly matched but also gave a similar sort of vibe to what open AAI and chat GPT was providing. And then we saw the perplexity claude all these other types of LLMs and some of them evolve further into using RAGs and just how they actually found this information. But it seems as if like what we're seeing to now nearing 20% usage. We're sort of seeing at the beginning Google's like okay we've released this new feature so let's let's roll it out um as much as we can. And obviously this was primarily within the the US before sort of it branched out a bit into the UK for example where I'm based. What we actually saw is in the early days it was actually pretty terrible for a lot of types of of searches or prompts that people were putting in the information was either incorrect or the sources that it pulled the information from were incorrect and then then you sort of see within the data after a couple months it's like okay rolled back quite significantly. And then what we're

seeing now is actually there's higher instances of AI overviews than the initial sort of launch or within the search lab sort of earlier this year. So maybe confidence has gone up now from Google's perspective and things are changing very very quickly as well. So, even for me trying to test stuff, I'm like, I'm having a I'm having a tough time trying to figure out, you know, what's going on, what's improved, what hasn't, cuz, you know, I gave a few talks earlier this year and a lot of them were specifically with AI overviews and how terrible they were. But now I'm like actually, you know what, they've kind of improved. So, like, fair play for Google being able to do that. But it just feels like it's a bit it kind of feels like a bit of an arms race. Um, but you'd expect, you know, a behemoth like Google to be able to at least carry quite a decent amount of leverage will be able to at least improve things relatively quickly compared to its counterparts. Yeah. Yeah. Yeah. And and for somebody who's watching the SERs literally daily, Barry's reporting on this on a regular basis, Barry, how

should website owners interpret the fluctuations of AI overviews presence in search? Because you report on all sorts of search features on a regular basis. Do you believe that the higher number of AI overviews in SERs maybe implies a longer term trend of declining organic traffic to websites? Double-edged question there. Take the first one. No, you're wrong. I'm leaving. I'm insulted. No, I think everybody expected it to show up more and more often as Google became more confident. Initially, Google was super confident. They launched with I thought when they initially launched way too many times that AI overviews came up in in the actual Google search results. Then they got called out for all sorts of fun examples. Um, you know, drinking urine while swallowing stones or something. I don't know, lots of different examples. And it was pretty embarrassing for Google. Um, Google rolled it back a little bit. Rethought things and now slowly is pushing like you said according to the study from like 10% to 20% or so depending on what data you read. It's good to see Google start to

have more confidence in this and in a slow gradual way as opposed to how they actually went ahead and uh initially rolled it out and it keeps getting better. I think you know we I kind of like people the SEO industry kind of like named it the Ray update with the latest uh Google AI overview stuff because she was one of many of who were calling out really bad examples pushing Google to do better. Um but ultimately I think this is like feature snippets rolled out. I mean, I don't think you remember the back in the day where when fe Google started to roll out these feature snippets and there were so many embarrassing examples of how it was giving wrong information, embarrassing information, dangerous information. It was literally like the same roll out, but obviously Google search wasn't as prominent as it was back then it is now. And of course, they didn't have this level of competition with Open AI and Bing doing their copilot stuff. So, it was a lot more eyeballs on what Google was doing because there was no such thing as feature snippets anywhere else. there was tons of people doing AI chat and bots and so forth in a very very visible way. Um, so Google had a lot more views on that and ultimately I think when it came to like the feature snippets we saw the same thing with Google and you see them a lot and they

do a pretty good job now. I think Google's going to continue to roll them out and continue to get better and we will be more confident in those results and I think the results will get better and better. They're not going to be perfect. Even now we can find examples of feature snippets that are wrong or make no sense. And I think even Google when they wrote that blog post about that, they cited some examples of like that I covered of a feature snippets going wrong and getting better. That being said, I think ultimately SEOs are going to be like and publishers and content creators and brands are going to be are going to want to be in those AI reviews if that's what's shown up first. Even though they might send less traffic, I know we ended that conversation with Dixon about brands and brands and brands and everybody's talking about brands these days. You want your brand to be mentioned in there and that's that's the ultimate goal here is I think Ran Fishkin just came out with Spark Toro data around it's the brand stuff. All this stuff is about brand marketing, brand marketing, brand marketing. So I think ultimately you're going to want to go ahead and test these fe these AI overviews to make sure that whatever comes up for related queries that you're searching for um that maybe you want your brand to show up there and how could you get your brand to show up there even if it's not a link which often they are now links Google's pushing more and more links in there. I

think brand mention is going to be very very important there. And I think ultimately SEOs who don't show up there are going to be wondering how do I get my clients to show up in there. Absolutely. Absolutely. And the fear that feature snippets actually engendered that we were going to be losing a heck of a lot of search traffic that kind of blew by the wayside as well. uh and we're predicting that this is going to be the same type of scenario here is that there is a zeroclick concept here that has been discussed and discussed but brand and authority that's happening inside a a AI overviews is is certainly going to be affecting any of the organic visibility that a client has and trustworthiness of that that website as well. Got a question for you, Anastasia. You actually did some research on the uh links that the the sites that are actually coming up in AI overviews and uh you you have some data there. How much um what's the size what's a percentage of the links in AI overviews that are matching the top 100 uh ranking sites? We have a lot of data here and I'll start with the fact that

on average ani snippet has six to seven links need. So these are chances for websites to be visited to be clicked at. And as for the sources uh we see uh we see that around 90% of websites are listed both in the AI snippet and in the regular SE and it is at least one link. But I guess a more interesting fact is that around 60% of the websites that appear in the AI overviews snippet also rank in the top 10 results. I guess it's an interesting topic to cover. Yeah, because knowing this uh we can assume that uh websites that already have some traffic from the regular like organic SERs will probably continue to be getting this traffic and at the same time some websites that do not rank in the top 10 results right now can have

more chances to be visible within these AI results. Have we also seen have we seen the loss of an actual organic position kind of consumed by AI overviews? That's jump ball for anybody. Have you seen that behavior happening or is it is established that we are there's impermanence now in the organic listings that that wasn't before? So I don't think the I don't think any listing would be removed. I think there were some cases with feature snippets where Google would do deep duplication and changed over the years where they don't want the same URL to show up multiple times um in the search results and that was on and off for feature snippets here and there. I think with AI overviews we haven't seen the duplication of the same URL yet. That's fantastic and that's a confidence builder because that was arguably some some thoughts in when we were looking at it in SGE in the lab test that some things were actually disappearing as AI overviews were being tested. I I do have a question um about this the shift of some of the intent of the uh AI overviews. Are you seeing AI overviews affecting the balance

betweenformational and transactional queries? Are there any notable shifts in user behavior or conversion rates that uh businesses should be aware of or are we able to track conversion rates from AI overviews yet? I'm not sure we can track this data because as uh we all know we won't have a direct data source for this kind of data because Google search console won't show us uh data for these particular URLs for this particular snippets. Uh what we can do we can track the keywords and we can see what keywords are listed what keywords trigger uh the AI overviews and for what keywords we have our URLs our website being present and uh you can do this with SE ranking as well and then you can track these keywords and you can match this data with your Google search console data like matching the uh positions, your keywords, your URLs and

your actual traffic data from Google search console and then somehow apply your uh conversion rate on this traffic to see if anything has changed or it remains the same. This is the way I see we can do this kind of analysis but maybe it more have a different approach. Yeah, I mean it just seems very manual at this moment in time and I'd imagine it will get easier further down the line in terms of you know having the the right sort of toolkit to know exactly how you're being impacted by these. I think the interesting shift that I've seen and that sort of is mentioned in the study is we're seeing more and more longer tail types of queries trigger AI overviews and it makes you think it's like well okay if we're taking queries that have you know maybe four words or longer they're more longtail they're searched for more uh search for less sorry so the search volume is going to be lower which actually means there's probably less money that Google needs to spend in terms of actually you know processing power and actually getting

these results to show up so maybe That's a reason why Google is more comfortable sort of showing AI overviews for queries that by nature may require more nuance because the longer your search is, there's going to be more specific types of details the more words you're including in there, which kind of tallies itself quite nicely to show that. I think that it's definitely more informational search focused in terms of the intent of the the search query as opposed to a more transactional or commercial or something where you're trying to purchase an item. I think we saw earlier on this year where there were some really just bad results when it came to specific like transactional types of searches. Um in terms of even the sources, we've seen a shift with that. It's like it'll be really interesting to see the shift of you know there were lots of sources that were shown several months back where it's like the actual domain or the actual page itself either was broken it was a park domain there were examples where the actual content wasn't anywhere on

the page within the source and you're thinking well how did that even happen so you know we we are seeing just such a so many changes in such a short amount of time that I'm not confident we're going to be able to get any sort of automated levels of tracking this um for at least the next 6 months based on just what we've seen because there's just so much happening. But like is saying it's like yeah things are a lot more manual. There may be ways to sort of look for correlations and be able to to do some guesswork and take multiple data sources and try and piece together the puzzle, but we're not going to have anything out of the box that would make our lives as SEOs easier to be able to report on this stuff for the time being. You know, that that also brings up a another factor of just the emergence of paid ads inside of AI overviews. That instability right now uh puts a number of advertisers uh a concern because your ad could very well show up next to something that you did not want to actually be associated to. So, we're not seeing that activation

collectively yet. I think Barry reported on that as well recently. Um it's being tested out in a number of areas. Do you know the percentage of ads that we're seeing right now collectively in the any studies uh with AI overviews right now? Any any any takes on that? Well, I think um in the study they they said um yeah SE ranking said a number of cases where ads appear alongside AI overview fluctuated throughout the year with the latest study showing about 8.77%. Now there's also something about the ads in the UK with 34%. Now I don't know if that specifically looks at the ads on the page in total. I think it does because like a lot of people confuse ads with AI overviews, meaning ads above or below the AO reviews or on the side. Ads in the A reviews are super rare to see. Google's been really really cautious about testing it. So I doubt it's that 8.7 some% probably under 1% in my opinion. I don't have a study on that, but it's very rare to actually see ads directly in the AO a view snippet itself enclosed in that box. I think it

was on a recent report within the last couple of weeks if I'm not mistaken that that's been seen in the wild. I mean the SE rankings is I haven't seen the study but I mean could be but I there are some people who are seeing it in the wild and Jenny Marvin the Google ad liaison said we're being very cautious about how basically you roll it out. Did she say something specifically? I don't remember the actual quotes. I remember hearing on her on stage at Brighton SEO where she said you are going to see it. Um we are testing it and is rolling out but they're being very cautious about it. So, I don't think you'll see it that often, and if you do, you're lucky. And then send me a screenshot so I can add it fast. Absolutely. All right, I'm going to pivot here for a second and uh give a give a a little bit of an insight or a question, I should say, towards some of the more sensitive topics that we're seeing AI overviews come up. Uh we liken it to the YM stuff and being able to have see AIO uh inside of medical advice as well as legal advice and financial advice. Should we as consumers really want to push back on that until

maybe AI is ready for that? Because we've seen a lot of misfires in that in that space. What are your thoughts about the uh those more sensitive topics? Oh, absolutely. I mean like I don't want to get involved in terms of the number of sort of lawsuits and trouble that Google has been in recently but I mean if you're going to be using uh YML topics to show AI overviews especially when the data is wrong you're going to open yourself to all sorts of controversy where somebody might end up losing their house and basically say well I got this information from from Google's overviews which was meant to be a you know credible source of of information etc etc etc. So yeah, it is very very dangerous. I think it's something that unless you are confident but then again it's a question of like who's going to be able to judge confidence. Is Google going to be able to take you know objectivity in terms of what information on a page that it decides to pull out snippets from to display in AI overviews? How is it going to be able to

even determine that? So I would be yeah pretty against uh it's kind of like having like a very confident friend. I mean, I think people need to think about these AI answer machines as like friends. Like you have, you ask your friend, you know, I'm having heartburn. One person be like, "Yeah, you're fine. Just eat more, I don't know, hamburgers." The other person may say, "You're having a heart attack. Go to the emergency room right now." And they're very confident in their answers. I think that's what AIO reviews are kind of like. It's like it's it's kind of like talking to somebody with a personality often time opinionated about what their answer is going to be. Um, and they don't give the disclaimer that like the person your friend's not going to give you a disclaimer that we're experimental and you know, we could be wrong. whereas the A overviews do. So it's interesting from the study like how often it shows up the in these sensitive topics like almost 30% for legal topics, you know, 18% for healthare topics and we haven't seen Google pull back much uh in those areas which is I really thought they would pull back a lot on that and they're not. They're they're they're being confident in their results and they're labeling them as you know this is an AI overview and it might be wrong. So I don't know it's pretty interesting to see that. Yeah, they kind of leaned into it and the study uh of those

different topics showed growth from one study to another study in in the business and legal and and medical environments. Uh right, we uh Anastasia, we talked about that a while back. Yeah. Yeah. It's growing a lot and uh it seems that Google will just continue to do so. I guess uh Google makes itself safe by just adding the disclaimer like this is just forformational purpose and like this uh what like makes this safe I don't know for Google also we see that Google still links to like guess trustworthy sources like medical uh websites like uh some governmental websites and I guess this is also a way for Google to make it safer while expanding the results, the AI results. And by the way, as we're talking about it, we'll flash the QR code at the end of this as well. Uh because there's a QR code to actually hit the the entire study that we're talking about. I failed to mention that before we kicked into

the stream. Final pivot here is really talking about how to optimize to for AI overview. It's too early and very in unstable waters right now. But how should SEO strategies evolve to remain effective in the context of AI overview? Because it's not going away. It's growing more and more and we could probably predict that it's going to go to 30 or 40% of searches in the next couple of years. Don't quote me on that. But how do we stay in that space? It's quite an interesting one. Ju and just because things are changing so quickly, I think the best thing you can do is just try to like use your brain and reverse engineer what is happening and like looking into sources to establish okay what is it about this particular source that's showing prominently what does it have? How is the content structured? You can look further. You can look into the the backlink profiles of the domain. You can look into the backlink specifically to uh that particular page. Um, and so there's a lot of manual things I think at this moment in time that we're going to need

to do because there is no clear answer and things are changing. I I definitely don't think that there's going to be something revolutionary in terms of how people should approach their content and their websites to optimize their sites generally speaking across the board. And I think that even applies to the LLMs as well. I don't think there's anything revolutionary that needs to be done, but I think if you really want to get a better understanding of figuring out why certain results are as they are with AI overviews, it's a case of looking in your niche and trying to determine, okay, here's the types of types of searches that this is showing. And obviously this is where the SC ranking study and other studies on this will will be really helpful for you to sort of know where to pinpoint your focus, but actually just trying to reverse engineer and the old classic, you know, puzzle solving methods of trying to figure out why certain things are happening and then test it yourself and experiment and try to do AB tests and track things over time. That would be my my biggest advice for this. Yeah. and explore uh what who was actually being

listed in that listed in those AI overviews and seeing what other topics they're actually talking about. Going back to Dixon's presentation about entities and topic knowledge, semantic knowledge there, um you may very well see that there's more trustworthiness in that micro site uh or cluster of information on a site. Barry, I wanted to move into some predictions in 2025 because I know you're continually reporting and and end of the year is always predictions, but when we come down to what skills may be important for SEOs in 2025 considering AI and other trends like this, what would you uh suggest to our audience? Yeah, I mean I mean if said use a brain, I was going to argue don't use your brain. You don't need one. No, I'm just joking. Um I mean there's AI that could replace it. So I mean we're all doomed basically. I mean, we should just give up right now. And I don't know why we're even doing this webinar. It's just I'm just joking. Um, all right. There we go. No, I mean, I think it's the the main thing everybody's been saying is brand brand. Um, that's the really thing you have to work on. It's interesting cuz most

publishers these days are switching models. Even the Verge just announced they're going to scription model. They wouldn't do that if they're getting crazy traffic from Google. if their traffic was increasing and they were getting a lot more traffic, they can monetize that based off of ads and impressions. Everybody's getting less traffic on Google. In fact, I reported that my own website got 15% less traffic from search overall in the past year over year. But overall traffic, if you even with a 15% drop in search traffic, which was most of my traffic, I got had a 5% increase in traffic. Why? Because of brands. I'm doing this webinar. I'm, you know, millions of people are watching this right now and asking who is that fool that says don't use their brain and they're going, they're googling me and trying to find out more about Barry Schwarz. And Barry Schwarz that comes up is this professor that's much better looking than me and wrote a book about choice. So, they can't find me. But I mean, I got to work on that. That being said, it's all about the brand. And I think the more you you have to think about the brand when it comes to any of this stuff. Search is is dwindling. People are are in other places and AI overviews are everywhere. And also AI search is everywhere. people are using chatbt as friends and having conversations with them and I think in

in in Japan and on certain countries they're coming out with AI based robots for be boyfriends and girlfriends with. So the world's changing and the beauty of our community and SEO in the SEO world is we we really absolutely love change and we adapt to it really quickly. We embrace it and we look we see that as an opportunity. So again brand is important but embrace the change. Don't be like one of those people says I'm not going to use AI. It's horrible. I mean you hear a lot of people in my community talking about it. It's the end of the world. I'm like, "Guys, you're going to be extinct in a few years if you don't if you don't do it." Try it, use it, and have fun with it. Use it as a tool and don't push back. Yeah, absolutely. We do as SEOs, we do tend to complain about change, but we still do adapt, right? So, there's that. All right. So, uh with that, I do want to toss it over to uh Anastasia for her thoughts on um how she sees because she's been leading these studies over at SE Ranking. How do you see AI overviews evolving in the future based on the trends that you've been observing? As I said, I do believe we will see more of AI overviews in search and maybe we will see more of this trend of zero click

searches when people will find more information directly in the SERs. But uh I guess this will lead to websites and content create c creators to create more bottom of the funnel content like to cover more in-depth more product or brand related topics that cannot be easily answered by the AI and we'll probably will need to create less like top of the funnel really easy what is content. I guess the websites and the SEOs will adapt and we will understand this technology and this new search more and we will find like really uh effective solutions to to get the traffic. I guess we will use more traffic channels uh in the future. Yeah. But still uh we will see more of AI in Google and I guess we will see more traffic from other AI engines. For example, uh we see more uh traffic from

child GPT and Perplexity this month and like this quarter in general and I guess we will see more of like traffic uh from other sources uh not only from Google like like it was before when you get like 95% of traffic from Google and that's all it'll be more diversified. Uh we just anecdotally we just saw our first lead come in for one of our clients from a chat GBT referral and I'm just itching to try to engineer back what was that citation where did it come from and to uh so here's uh reference just dig through find the questions that are being asked answered on that page and you can start trying to find how this actually is being created but boy there's a big black box inside of LLM's that we can't get into yet that will be a source of traffic uh growing already is a growing source of traffic right now. Edimer, uh final thoughts, a uh a thought towards the prediction of SEO in the uh next year. Oh, yeah. I mean, I I think that taking what I think is going to happen is we're going to see,

especially from mid to late 2025, there will be sort of year-on-year declines in Google Search Console for a lot of websites. And I think what they need to do is be better in terms of how they're reporting on overall traffic to their website. So very recent, literally yesterday, I put a a LinkedIn post up sharing uh a Looker Studio dashboard that pretty much takes uh your website. You put in your data through J4 and it will show you the traffic evolution over time from the most popular types of um AI platforms. So your Geminis's chat GPT, Perplexity, etc. that that's the sort of stuff that needs to be taken into account next year in terms of how SEOs are reporting on data to websites to be able to more accurately show that actually like Barry and like Dixon and what we're talking about here is brand and showing actually people are seeing you but it's not highly concentrated and Stasia said it's not 95% from from Google search directly. So I I think that from the Google search type traffic

sources, we're going to see less of that next year. But it does mean that SEOs will need to be on their agame in terms of how the to I guess as accurately as possible report on a more holistic sort of traffic acquisition to your website. Those are answering questions and many more features and facets than Google search right now. And especially with the search GPT that's also u being released. I mean you're going to you're going to have a much more diversified portfolio of of reference that users are actually experiencing. So thank you very much for your time today folks. I really do appreciate the panel's contribution here. Hopefully we answered a lot of your questions listeners and watchers for AI overviews and the future of AI overviews in 2025. And hopefully we're given some good SEO skills uh guidance at least to what you should be doing in 2025. Hey, we wish you all the best our panelists and thank you so much for playing. Appreciate it.