Hello everyone. Welcome to the AI SEO webinar with Ethan Smith. I'll go over just a few things. We're going to do about uh 40 45 minutes of a presentation and then we'll have plenty of time for Q&A at the end. Excited to have this session. So I'm Ethan. I am uh CEO of Graphite which is which is our company. I teach the Reforge SEO AEO course as well uh if if anyone's taken that and excited to chat. I've given a few presentations on this subject but I'm going to share uh specific case studies and then I'm going to share more around AI content but uh a lot of this is going to be around specific case studies. Also people have questions in the chat please feel free to ask there and then Caitlyn will summarize anything. Uh but uh if anyone has any follow-up questions, I'm happy to make this interactive. Cool. So this is our company which is Graphite. We're an AI powered growth team. So by that we mean we're a growth agency and we we uh use AI to power our our work. So we we specialize in search engine optimization, answer engine optimization, paid advertising, content marketing, uh work with companies like this. I'm going to talk about Ripling and Web Flow uh case studies in this presentation. And um when people ask me what's the best way or how do you hire an AEO SEO company or how do you hire a person? My answer is always ask them who they worked with when they started. Then look at their chart in AHFS and see if it went up and the more it went up uh the better. And if it didn't go up then you should probably be concerned. So uh and also the uh the best way to know what works is to reproduce this the uh reproduce things over and over again. So reproducibility. So these are SEO charts. These are not AEO charts. But I will show you some uh AEO charts. Uh okay. So the first is that AI is growing. It's it's grown a lot even in the past few months. It's especially been growing since January of this year. I gave a presentation on this I think about a year or a year and a half ago but the interest wasn't quite there but it's blown up since since January. Uh statue PT. So this is a breakdown of which surfaces are used the most. Which surface is used the most depends on the category. So engineers and people in SF tech are tend to use perplexity a bit more but uh in general chatbt is by far the the largest and now we have real data. So in previous presentations I presented ideas that I thought would be impactful but now we have uh actual uh case studies and examples. This is web flow. So web flow now 8% of web flow signups come from LMS which is huge and that's grown a ton over over the past year and uh the other interesting thing is that the visits from LLMs are more qualified signup rates higher so the quality of the visits are are really high so now we have real data showing that LMS are actually causing uh conversions and outcomes okay so AI frequently starts with search. So what I mean by that is if I say something like what is the best website builder for designers, it says searching the web and then it's going to get a set of search results and then it's going to summarize those search results. So you'll see here when I go down here uh well anyway you kind of okay yeah so it did a search this these are the search results. So for Google, you would get these links here. But for LLMs, this is a summarization of the different search results. And you'll see here that Web Flow is number one. Web Flow is number one because it's mentioned the most. So this is a summarization across all the different citations. So even if Web Flow had their URL ranking number one here, it would not win uh if if that were the only thing uh that happened. It's winning because it's mentioned over and over again by all these different sites. So why that matters is that uh answer engine optimization is very similar to search engine optimization that the core principles uh overlap heavily and so you can kind of think of this as a holistic strategy across both SEO and AEO and then chat GPT is converging with search perplexity is converging with search and uh when you ask questions like what's the best TV you'll get a list and you'll also get these rich car uh rich cards. So when Google started, you would just get 10 blue links and then Google had uh shopping carousel maps, local uh rich modules. We were seeing the exact same thing happen with chat. And this is also part of why uh engine optimization has blown up since January is because you didn't see these things prior to you didn't see these things often prior to January. But now you're seeing these rich modules and the uh the answers are now very clickable and so people are getting more uh referral traffic from LMS. And I also think that Google and and Chachi and Perplexity on these different surfaces, they're probably all going to converge and have a very similar experience. Okay. So, one thing that's different about uh search and LMS is for Google again, they used to have 10 blue links and then they started adding all these different modules and uh and a lot of Google clicks don't go to organic, they go to ads or they go to their own products. So, if you do something like wedding photographer San Francisco, all of these are ads, more ads, Google map and then you have to scroll all the way down to actually get a an organic result. And so a lot of the clicks don't go to organic in in Google search. A lot of them go to ads or they go to Google products. And that is not the case for for LMS. So LMS don't have ads or at least they don't have many ads. So you'll see here on the left uh we did a study looking at above fold screen real estate and over half of it's not organic in Google and only 19% of uh Bing above fold is organic. Whereas for Chachi Betain Perplexi, it's all organic. Both companies will probably introduce ads, but um just like in YouTube, if you subscribe to YouTube, uh you don't get ads. You only get ads if you have a free tier. Probably the same will be true for for the LMS. And many people subscribe to the LMS. So, I think that the overall click uh click share is going to be way more organic and and far less ads, which is great. Okay. So we uh we introduced the concept of topics in the reforge SEO course. So a topic is a cluster of keywords. This is in search. Topics are clusters of keywords. So this one URL here, best website builder is targeting 2400 different keywords. So you're not targeting one keyword. You're targeting a whole cluster of keywords in search. And then similar with uh with uh AEO. So you're not targeting one question. What's the best project management software? You're targeting a whole cluster of questions. There's many different ways to ask the same question. And so the topic concept translates over to answer engine optimization. So a topic in AEO would be a cluster of many different questions. So for project management software, what are some of the most effective tools? What are good free tools? Which ones are good for for development? All a single page can target all these. So one page is targeting one topic which is hundreds of uh keywords or hundreds of questions. And then there's a long tail. So uh and also the tail in chat is larger than in search. So the average uh I believe of number of words per question is around 25 whereas for search it's around six words. So the tail is much much larger. And people ask lots of follow-up questions. So they'll ask what's the best tool for X and then they'll say well what about these attributes? What about these features integrations? And so the tail is much larger. And there's different strategies for both. So when you ask something like what's the best website builder for designer? So again I showed you web flow but it's not because they it's not because their URL ranked number one and in fact it it doesn't rank number one. Techraar Reddit expert and so they win because they're mentioned mo many times and more than others. And so the strategy for this to to win this is different from uh Google search. Google search is my URL shows up number one. Here it's I am mentioned many times. And so we call this earned AEO. And uh if you look at this one here, there's different strategies. So here uh web flow is mentioned in the Reddit thread and there and it's mentioned in tech radar and therefore it's uh therefore it's the answer. Then for tail, as you ask more follow-up questions like what about these extra additional features, then your URL is uh how you win. But there's these two different strategies and I'll cover this. I I'll go I'll go through examples of this. So we uh we're going to publish a study on the most cited domains and uh this is for chat GPT. But you'll see again yes I want my URL to win. But if I want to win these ad questions I want to get mentioned many times. These are the sites that are mentioned the most often. This is probably not too surprising, but Wikipedia, Reddit, YouTube, and then uh media sites like uh Business Insider, Investipedia, stuff like that. And then this is Perplexity. An interesting difference between Perplexity's most cited domains is that YouTube is so much more heavily cited. Uh but then we, you know, we see a similar set, but YouTube is a huge channel for uh for citations. And I think people focus a lot on Wikipedia and Reddit and underfocus on YouTube. So a and also YouTube is something where you can control what videos are there whereas Reddit and Wikipedia is heavily heavily uh uh managed by the the community. So YouTube is a very under underutilized strategy. Okay. So then how do you win? You win it's pretty similar to to Google. You win by answering every question. So why did this URL rank and the reason why it ranked? If you look here uh this so again your cluster keywords your clusters of questions what's the best website builder for develop uh for uh small businesses for e-commerce and this one wins because it has stuff about small business and e-commerce so the exact same thing is true for for uh AEO your URL shows up the more that you answer every question the more more uh your page shows up just like just like a Google search okay so then I'm going to go through a few different uh categories for how to do AEL. So, I'm going to start with B2B and uh similar to what I described before, you win both with your URL showing up and uh owned and then earned which is other people mentioning you. And this is a Okay, so this is uh most cited domains specifically for B2B. This is perplexity and you'll see YouTube Atlassian uh betterup which is a company uh we worked with Google somebody posted I think posted on LinkedIn recently stuff around Google being part of the citations and we also have seen that uh but but these are the most out of domains for perplexity that we've seen specifically in productivity okay so I'm going to go through a few examples so example one is uh what's the best payroll management software and you win by being mentioned by others and then as you ask these follow-up questions like well which of these have which of these integrate with HR IT and finance and now Ripling wins and Ripling wins because of a URL on rippling so this is an own strategy so why did this one show up this one showed up here introducing uh which one is this here yeah so this one. This one won because it listed the fact that it does all of these things. It does HR, it does it, it does spend management, and therefore it wins. And it wins similar to to why it would win in Google. But um uh for Ripling, there's all these questions about which payroll management software has these features, these use cases, these integrations, languages, regions. And so the more that we have these pages about uh about all the things that Rippling does, the more likely this to show up. And this is also a bit different because in traditional search you might have things like how to do X but you wouldn't necessarily focus as much as as much on which what are the things that uh my tool allows you to do. You would you would have a few pages but in in chat you can have way more pages about these are all the things that my uh product can do. Okay. So this is HR and it is yeah spend management. This is another another example. What are the best payroll tools that do all of these things? And uh I'm guessing many people on this uh on this webinar, you're probably if you have a simple question like best project management software, you can probably do that in Google where chat outperforms uh a Google search is asking all of these follow-up questions and having this long ongoing conversation. And so you can have landing pages to to target all that stuff. So spend management, why does Rippling show up here? Ribling shows up here because it has a page saying yes, we do spend management and therefore it shows up. Uh let me show you this other one. So yeah uh okay well which of these payroll management software solutions which of them integrate with paid? And then we have this example here. Will we integrate with paid? And so um uh another example with this is with Otter and with Otter we want to ask the question which of these integrate with Zoom or sorry which meeting transcription tools integrate with Zoom and Otter is an answer and Otter is an answer because there's multiple pages talking about how they integrate with Zoom. So this is the Zoom product page. This is an article about how they integrate with Zoom. This is help center. And so all these things can again say I my product uh fulfills this use case. This is how my product fulfills this use case. These are all the things that we integrate with. These are all my features, all my use cases, all my languages. And so my general recommendation for B2B is list out all of the use cases that your product is a solution for. All of its features, the integrations you have, maybe even integrations that you didn't know you have. So uh another example here is which meeting note uh transcriptions integrate with looker. So we have sales calls and we want to transcribe the uh the meeting and see who was in the meeting and then chart that. So we want to put that in our Looker dashboard. And this says uh I believe this says yeah tools with transcription but not known native Looker and it says Otter and Otter actually does sort of integrate with Looker because Otter has a Zapier integration where you can grab the meeting send it through a zap anywhere including BigQuery and then uh do a Looker dashboard. But if we don't have a page about this is our integration with Looker and so we're maybe less likely to show up. Uh but yeah, overall for B2B, yes, do things like how to do X, but also list all of the use cases your product is a solution for, all of its features, languages, integrations, maybe integrations you hadn't even thought of. Uh have a bunch of landing pages around that help center optimization. Uh those are my recommendations for B2B. Okay. So, I'm going to jump into Web Flow and I'm gonna switch my screen. Okay. So, web flow. Okay. So, for Web Flow, I showed you that chart where we're driving a good amount of or 8% of our signups now. And so, uh, we've done specific things at Web Flow to cause that. So, here are the things that cause that. The first thing that I mentioned is YouTube is a huge source of citations and Web Flow has a ton of videos. So, Web Flow has 800 different videos and there's a bunch of videos about how to accomplish a goal like how do you do how do you do uh yeah create and style collection list uh yeah like enterprise custom roles etc. So there's all these different uh ju just like I showed you the use cases for rippling uh which of these have spend management you can have videos around that and the nice thing about this is that if you're an early stage company who doesn't have a lot of authority uh it's maybe hard to rank but you can immediately rank in YouTube. So if you have a YouTube video about some high LTV topic like how do I do X that can rank right away in YouTube. Uh and and also many B2B companies aren't actually building or making uh YouTube videos. So it's a very under it's probably the most under uh utilized strategy is just making YouTube videos about how to accomplish a goal with your product. Okay. So the second thing is Reddit. And Reddit's one of the things that we're asked about the most and it's very hard. We've played around with uh uh adding comments to Reddit and the community is a strong community that places the content and so it's pretty hard to optimize for Reddit in a sort of uh parade hat way where you say my tool is the best you know my product is the best thing for this for this particular use case. What does work is uh being authentic adding actual value saying who you are uh and then people don't get upset. So this is an example of Vivian at Webflow where somebody asked a question about Web Flow. Vivian said who she was. She added actual value. This is another example for not explicitly about Web Flow but what CMS do you recommend and then uh somebody at web flow says this is who I am. I worked at these companies and web flow is great and it it gave a useful uh useful summary of you know why it's better visual canvas uh etc. So these we're seeing these work. So the the general recommendation here is yes, Reddit works. Uh to make it work, say who you are and add value. Don't just promote your product or sell something. Uh and build a YouTube audience. Another uh thing I've had good success with is buying a small amount of ads for YouTube videos. So, we take webinars like the one like this one right now, post them on YouTube, and you can boost them with small amounts of money and actually get a pretty decent amount of views and get like 2,000 views for a few hundred. And now your video is showing up in citations. So, this is this is probably my number one recommendation. This and doing landing pages for all of your use cases and features. These are the two most impactful things uh probably to to do B2B uh AO. And then uh good content answers every question. So same as with search, but uh the questions are probably more longtail in AEO, but for your URL to win, you need to answer every question. This is an example where we make content briefs, which is sort of a summarization of all the different keywords and questions we're targeting. And then we make landing pages to fulfill all that. So this is uh this is what do I need to become a freelance web designer and part of the answer is host your portfolio use portfolio websites like behance dribble and web flow and web flow has this landing page here freelancing for web design and you'll also see that part of the answer from web flow was dribble and behance So chatbt is directly taking stuff from this article from web flow and also including web flow as part of the answer. So go to web you know go to web flow and make a website uh to uh create your design portfolio and so so the result so we're seeing really uh really really good impact from this 8% of signups. So these are actual strategies that we know have caused outcomes. Okay. So let's chat about consumer and uh these this is the breakdown of most citations for consumer websites. You'll see a bunch of actual stores as part of consumer website or as part of consumer. This is for chat GBT which is interesting. Uh so target. So if you're out if you're curious who's doing the best in these different categories, you can just look at the most sighted domain. So who's doing the best? Looks like Target's doing the best. It's actually doing better than Amazon, which is interesting. Doing better than eBay uh and Best Buy. So if you want to model yourself after someone, Target appears to be doing the best in AEO for commerce. Okay. So then what are different strategies? So the same strategies that I mentioned earlier also work for any category including commerce. An interesting strategy for consumer is affiliates where I will show you this. I have many tabs open. Okay. So, if I ask the best best credit cards for travel, you will see Capital One, you'll see Chase Sapphire. And if you look at the citations, it looks like they're mostly affiliate sites. So Barons, The Points Guy, Market Watch, these are all affiliate sites. And so you could basically pay them to include you. So here, this is one of the uh citations. Forbes, best credit card for travel. Guess what? It's a Chase Sapphire. And that's because they paid them. So uh uh all these these affiliate sites are pay to play. Uh one interesting thing is that affiliates are their monetization is based on getting clicks. But uh for chat you don't actually need a click, you just need a mention. And so if you're down here and you're not actually getting clicks and you're not paying for the clicks, you might actually still win just because you got mentioned. So what matters more is getting mentioned versus appearing high or getting a click within here. Okay. So AEO tools were asked a lot about this. There's a bunch of different tools. We have a landing page that we made with many different tools which I'll add in the chat here. Um but there's a lot of different tools. I think there's over 60 tools we have here. And um they mostly do tracking like this and uh I covered in more detail how to do measurement but how we're generally measuring the impact of uh answer engine optimization. The first thing is conversions. So I showed you the chart with web flow measuring conversions. That could be last touch or that could be asking people how did you hear about us post conversion and then uh you can look at traffic via last touch. Uh, and then you can look at am I showing up for the questions that I want to show up for. And that's what these tools are generally doing, which is you put your questions in and you can track your performance over time. We have a tool that does this. Uh, there's a bunch of others. They're all pretty similar. They're all kind of like keyword tracking, but question tracking. So, my general recommendation to people is uh, pick whatever tool does what you need and and is the cheapest. Okay. So, uh, we get asked a lot about AI uh AI generated content. I think somebody in the chat actually said that they're building programmatic AEO with AI generated content. Uh this is a study we're going to publish showing that the cont there's more content published that's generated by AI than humans now. So this is looking at common crawl URLs over time and uh looking at how much uh is AI generated and now there's more AI generated content than human generated content or human created content on the internet. So AI content is taking over gaming content and uh but it doesn't work. So uh we published a study last year, we're republishing it this year showing that uh AI generated content is not performing in search and it's we also looked at is is AI generated content showing up in chat GBT citations like is chat GBT generating derivatives that are then fed back into its own uh into its own uh model and the answer is no. We're not seeing uh AI generate content work. So it's true that people are using this strategy. It's not true that it's working. Now, that doesn't mean that it can never work, but generally speaking, it's not working. So, my general advice to people is to not do AI generated content. Even if it's working now, it's probably not going to work in the future. So, I would generally recommend against AI generated content. Uh, using AI to produce content makes a lot of sense. And the f I mean, everyone in in the future is going to be using AI to assist them in creating content. and and uh AI is great at giving ideas, giving suggestions about what an outline might be, what questions do people have. AI is great at that. I just wouldn't use AI to write the entire thing for you. I would use AI to assist, but you human uh is the one who's actually producing and writing the content. Uh this is a correlation study. So human is on the left, AI is on the right. And this is showing again that uh strong correlation with AI generated content does not perform uh uh in Google search and chat GPT. The more AI you use the worse. Uh how to think about that is you know maybe if AI generated 10% of the page that might be okay but the greater the percentage of the page that's AI generated the worse it's going to perform. And this is uh clearly showing that that's true. Okay. So key takeaways are um AEO heavily overlaps with SEO. Many of the core principles are the same. The two main differences are that the tail is much larger in chat and owned uh AEO matters where it does not matter in SEO. Uh second thing is topics. So the topic concept topics are cluster of keywords are also cluster of questions. Uh how you optimize your page is the same as Google which is uh answer every question. the more the more comprehensive your content is, the better you're going to do. Earned is the new thing. Uh that is not uh not something in SEO. And earned, there's YouTube, there's Reddit, there's affiliates, all of these are new. And if you want to own, if you want to win the head questions, you need to get mentioned multiple times. And so earned helps you win the the head questions. For tail, it's all of the follow-up questions. for B2B what are all the features that I have use cases that I fulfill integrations that I have languages regions and and filling in that till uh and and don't do AI generate content use AI to help but not to generate. So I post a lot of stuff on LinkedIn. I'm posting several of these studies uh over the next few weeks on LinkedIn. So just add me on uh on LinkedIn. This is my email and uh happy to answer any questions. Awesome. We have a bunch of questions. So, we can uh start with the ones that are in. So, on the most recent topic, AI generated content. We have a few follow-up questions, which is people are wondering how they detect AI generated content or know that content is AI generated. Um, and if you have an hypo a hypothesis on why it's not performing well. Yes. So for this study we used surfers surfer SEO AI detector. We use this and uh I will paste it here. There's others. Grammarly has an AI detector. Originality has an AI detector. Uh we used this one. We also evaluated how accurate is surfer's AI detector. And so we did that a couple ways. The first way is we AI generated a bunch of content and uh surfer was I think it was 99% of the time was correct and then we took our own human written content uh that that we make said dropite and the false positive rate was around 8% or so pretty low and then we also took content that was created before chat GPT existed so I showed you the common crawl example and uh the the uh false positive rate was also like 8%. So, Surfer SEO's AI detector is very accurate. Uh I assume Grammarly and and Originality are also similarly accurate. I we just haven't done, you know, rigorous evaluation. So, we we use surfer SEO just because we're we're friends with them. Um so, anyway, so that's how they how they detect. How do they actually build the algorithm? I don't know. Uh why does AI generate content not work? The reason why AI generated content doesn't work and or rather why did Google and chat decide that they don't want it to work. The reason why they decided that is the same reason why scrape content doesn't work. So in in the uh late 2000s and how I started my career was scraping other people's content and then remixing it and creating a bunch of landing pages with other people's remix content. that worked really well and that's how longtail SEO that that was part of longtail programmatic SEO and um uh everyone was doing that then so then all of Google search was just other search engines or other search pages so it was a search after a search and uh so Google created algorithms saying I don't like scrape content and one could argue that the scrape content that we were doing was actually good it was summarizing other people's useful information so you know one one could say well it it's good but for Google if Google is just a search engine for other search engines then there's no reason for Google to exist. So similarly with uh with AI generated content if AI generated content works and there's all this AI generated content then when I do a search on Google it's just uh a search engine for chat GBT responses which is not valuable if it was if if all Google is just chat GBT then I could just go to chat GBT there's no reason to go to Google and then similarly with with chat GBT if chat GBT has answers that are uh derivatives of its own derivatives then you can have an infinite loop of my own you know derivative after derivative and then you have this really noisy uh you know very quickly the the uh the answer becomes not useful. There's a paper called model collapse about using derivatives of derivatives to train the model and all the problems that that causes. But basically it's it's it's a it's if AI generated content works then Google and chat and all these just very quickly become uh not useful. Um so so that's why those companies have decided to make it not work. Awesome. Okay. Next question asked a few different ways by a few different people is how do we go about finding AEO topics that we should target? So obviously in SEO we have keyword volume. How should we figure out what questions to track and what clusters of questions to to track in AEO or to target track and target? Yeah. So how do we get search volume for questions? And the answer is you can't yet. Uh but what we do is What we are doing is we're taking keyword data, Google search keyword data, and then we're transforming it into questions. And the assumption there is that Google search Google search behavior is is directionally similar to to question behavior. We don't actually know if that's true, but it probably is true. Uh so we are we built this tool. This is an internal tool but we built a tool that basically takes uh Google search keywords and then transforms them into questions. So I would do the same thing uh here. So you'll see here payroll management software and then source keyword is here and then we have the question here. So that's what I would do. I would just take your SEO keywords and then make them into questions. What you don't know is the tail. So I think that a way that you can get the tail which is a a good idea that uh guy at Webflow had had which is like what are all the questions that people are asking about my product and what I would do is go through your uh sales calls and see what questions for B2B go through your sales calls see what questions people are asking in sales conversations customer support if you're a consumer app what are the questions people are asking in reviews but try to get uh some information from your own own internal data about what questions do people have about your product. Awesome. And next question, a few kind of what I would call tech AO questions. Um, but people are asking about LM.ext. Um, is it important? Does it have impact? Asking about optimizing robot.ext for the same reason and also crawl optimization. So what like kind of more technical um strategies have you found to be either zero impact, low impact, high impact or we don't know yet? Uh generally speaking, all those things are not impactful probably. So uh LM.ext, as far as I know, no one uses LMS.ext and it was an idea that somebody proposed, but none of the LM companies are using it. That's my general understanding. It's possible I'm wrong, but uh Google has said that they don't use it. I I believe Perplexi said that they don't use it, but there's no evidence that I've seen that they use it. There's evidence that it gets crawled, but any page can get crawled. My my help center can get crawled. That doesn't mean that it's it's it's used for anything. So, as far as I know, given the data that I have, LMS.ext is not used at all and and also has no impact. Uh the other question was about robots. So, robots doesn't cause traffic to go up. Robots blocks part of your site. uh which could be useful, but it would be useful for something like I don't want the LLM to train on my data. I want to get indexed, but I don't want it to train on my data. And I believe Cloudflare announced something around this where they're going to block uh the training bot. So, if you don't want your data to be used without your consent to be part of the model, you can do that, but it won't cause traffic to go up. Uh the other so so what what is impactful for uh for technical AEO? There's two things. The first thing is make sure that the page can be uh crawled and indexed properly. So avoid weird JavaScript implementation. A second thing is uh cross syncing. So Google and the uh chat GPT search engine are crawling your site with links. So make sure that you have links on your site so that the bot can crawl them. I'm going to show a quick example here. So for master class we have these links here. This is very impactful. We see this over and over again. Most sites are not optimized for their cross syncing. So have better uh have better cross syncing, have better spread cross linking. I have a we have a study on my LinkedIn uh on this. Those are the two most impactful things. There's oh and schema. So add schema for I showed you the the shopping carousel and maps and all that sort of stuff. ad schema, especially review uh rating and review schema, product schema, location schema, uh most other things in uh SEO and in AEO that are technical, most other things do not have an impact. Um great. And you you uh next question that's that's being asked by multiple people is also uh you covered this a bit, but attribution. So um how are you thinking about attribution or tracking what conversions are coming from LLM if there's not always that refer data you know how are you doing that with clients we are asking how did you hear about us questions so post conversion we will uh we'll say how did you hear about us what's the source especially for B2B I strongly recommend this for attributing AEO for attributing any other marketing or you know when you sign up for web flow or you sign up for Ripley, you probably heard about them 50 different times. And if you only look at the last touch point, you don't actually know the other 49 touch points. So I would definitely ask how did you hear about us uh for for any B2B company. Uh you can also do last touch attribution where it's just uh use G4 and see where the referring domain was was uh chat GBT. Some people you saw you may have seen in the URL are putting a URL parameter saying it's from Chat GBT, but that's only going to tell you probably onetenth of the actual impact that you're getting. Uh most of the impact is probably not going to come directly from a click. And so the only way you know is how did you hear about us? Awesome. Okay. Next kind of um category is around the um earned optimization strategies you mentioned with YouTube and Reddit. Folks are wondering if you know on the YouTube side does the quality of the video matter and on the Reddit side does the number of up votes matter? Um and if not today would we expect that matters in the future? Um what are your thoughts on that? I don't have clean data on that but what it appears for YouTube videos to show up is more views and so the quality would be a function of how many views does it get. Many of the videos that I see are not high production value uh videos. It's, you know, a video like this one or, you know, a loom of me talking about uh payroll management software, you know, this use case. So, it could it could be a pretty low budget video as long as it's useful. It doesn't need to be high production value and you need some decent number of views. I I I'm not seeing like a very clear correlation with uh for which YouTube videos are showing up, but in general it looks like more more views uh helps it show up and probably more recent for Reddit similar. I don't have super clean data on it. It does appear that uh there's a correlation with number of up votes, but I don't have it's not like a a highly predictable uh model that I've seen. Got it. Okay, let me find you the next question. Oh, um, someone was mentioning and I think this is someone was mentioning that they were seeing different results logged in versus logged out. So, I think you know um maybe covering what we've seen with our studies on that and also how that affects the different how you might choose a tool in the space. Um, so logged in, locked out, etc. and the different variations. Yeah, there's there's a few different variations and I will show you. So, there's logged in versus logged out and then there's also with search and without search and you'll get different answers and you'll even get different answers if you ask the exact same question logged out. You'll get different answers. I think we looked and uh you need to ask seven to 10 times or so to to get an actual distribution of the answers and that's just for logged out. So uh uh we've been looking at the difference between logged in, logged out and with and without search and uh historically it looks like it's very different and the citations are very different. Uh so we do both logged out scraping and logged in scraping so that we have both. It does seem anecdotally like they're actually getting to be more similar. Uh I don't have clean data on this but uh most of the almost all the tools are doing API. So sorry there's logged in, logged out and then there's the API. So most of the tools are using the API which is actually different from logged out. And the API response tends to be quite different from from your personal experience. Most of the AEO tools are doing API scraping and not actually scraping the page. I believe a couple are doing scraping, but it's mostly API responses which are very different, especially for the citations. However, it's possible that they're getting more and more similar. So, sort of an ambiguous answer, but uh you know, if you want really clean data, you need to use a tool that is doing scraped and not API uh tracking. Great. Um, we just got a great question which I think you know there's a lot of similarities between AO and SEO that you just went over. What in your mind is radically different or are there even radical differences between the two things that we should really focus on when it comes to strategy? There are two main differences. Earned is one difference. So for the head question to get mentioned many times and not just a URL on your site. So Reddit, YouTube, affiliates, that's thing one. The second thing is that the tail is larger and uh uh listing out all the features that you have the integrations like the the uh rippling with paid integration. I don't know how many people are searching for that but there's probably a lot of people asking about which tool has these integrations. So having uh filling in the tail is different and for B2B in particular it's all the features that you have the use cases the integrations maybe the integrations and connections that you didn't even know that you had like Otter plus uh plus Looker help center optimization those are the two main differences. Great. Uh next question we have here is so you showed some examples for B2B and consumer products. Um, what about marketplaces or storefronts? You mentioned like Target was specifically successful. Do you have any is a strategy change at all for a marketplace or a storefront or more programmatic pages like that? Yes, it does. So, let's take an example here. Uh, what is the best dress for dinner parties? Okay. So, if we do this, best dress for dinner parties. So, you'll see a bunch of category pages like this. And uh and similar here, you'll see the same thing. So, let's see here. Sources. Uh actually, you're not. Yeah. So, birdie here. So yeah, actually you're seeing a bunch of editorial pages, but um let's say so again chat is follow-up questions. So let's say I wanted something like uh I want the I want this feature I want wrinkle resistant. So, which of these are wrinkle? Which? And what you will see is that all of these facets here, there's a bunch of information about these dresses that is not on this page. So whether it's reflective or wrinkle resistant, that that tag is not here. It's on the product page, but it's not here. And so all of these followup questions about which dress has this feature, this color, this other stuff, it's not here. And uh and so the LM can't find it. Which of these? Could you clarify which specific address? Yeah, this is maybe not a not uh not a great example, but um essentially the the main difference with marketplaces and commerce is to expose this facet data here. So, how I might approach that would be the LLM version of this page maybe has a bunch of tags here or, you know, you're essentially exposing a lot of the product page data. You're exposing that all this stuff here. You're exposing that on the category page. I would not do that for SEO because Google would would not like all that boilerplate content. But for the LMS, I don't think that I'm not seeing individual product pages show up in LMS. I'm seeing these category pages and the tags and metadata are not present here. So I would I would consider having a different version of this page with a lot of the metadata exposed uh here. Uh the other the other thing for for commerce is listicles and uh articles. So here you saw it's a bunch of listicles. It's actually not category pages like this one here. And they can be similar. This is essentially a category page uh reframed as an editorial page. Uh but but those are the two strategies strategies I'm seeing working. Great. Um, next question is, how should really early stage companies be thinking about AEO? Should it be high priority? Should they be investing in it? Um, what are your thoughts about that? I think that you should do a little bit, but not a lot. So, let's do something like what is a good payment API alternative to Stripe. So I see a bunch of uh very specific niche B2B companies doing some sort of AI for X use case uh like you know best AI uh payment alternate or whatever and um generally speaking I would say like when people ask me early stage company asks me what they should do for SEO I say don't do anything at all just do other stuff do paid advertising do other channels wait until you have more traction to SEO for AEO, you can actually win though. So for AEO, you can win by getting mentioned. So you don't necessarily need a lot of authority. You can uh you can win by getting mentioned here. And so for this one here, these are large companies. Uh but uh but if you're a new company, I haven't heard of this company, but uh if you get mentioned by these, you could win tomorrow. And how you could get mentioned by these would be the same stuff. YouTube videos and Reddit and YouTube is especially interesting because how many YouTube videos are there about uh you know very specific niche uh 7.8K views. Yeah. Like I I would make YouTube videos about your money terms like how to do X and there's probably not a lot of competition for that. But I would stop there. So I would if I'm an early stage company and and I have these five key use cases, make landing pages for those five key use cases. make YouTube videos and maybe comment on Reddit. You can potentially win quickly and then I would stop. I would not do all the other stuff that I talked about and then I would move on to the next thing and then as you build more authority then you can have your own URLs showing up uh here but I would exclusively focus on earned. Great. Okay. Next question is somebody's kind of wondering if you have a sense of how the models are actually uh searching. So it two options. Option one is you know they're they're um you're asking what's the best builder for a designer and are they running a bunch of queries and search engines and summarizing all those different runs or are they just going right to Bing or Google and asking the question and giving and summarizing the results they get. So basically do they run variations through search or do they just go straight and take the first set? And do we know? I'll tell you what I think. So, we know that they're searching the web and we know that some of the search results are listed here. So, we know that they're looking at this here and we know that they are looking for mentions and that we know that the thing that is mentioned the most seems to uh rank the highest. What we don't know is is this the comprehensive list? Are they running other uh other searches? I believe we actually do have data. I think we have data on Google AI overviews where Google appears to be doing a fan a fan out where they take the keyword and they basically create uh other uh refinements or variations of the keyword and I believe we have data saying that AI overviews is doing that. So is this is chatbt also doing that? Uh I don't actually know. I'm not sure. My guess would be that they are I think that uh they are probably running a lot of these things in caching results. They're probably not doing a live search every single time. But yeah, is this comprehensive or are there even more parts of the search? Uh I don't know the answer to that. It would be very expensive if they did. It probably also depends on if you're paying and if you're logged in. Got it. And then um this is switching gears a bit, but I thought this is a great question. There's an industry-wide push back against top and middlefunnel content. What's your take on that? There is push back uh push push back on like what's the definition of the cont middle of the funnel content basically publishing that and even investing in that. And do you think people should be creating top of the funnel content or middle of the funnel content or just focusing on that conversion content? I would definitely do uh top of the funnel content. What I might not do is really low intent what's the definition of a word type content. Uh but I would for sure do mid and longtail and like uh listicles and all that sort of stuff. I'm seeing that work really well. Um okay, this is specific one that has some up votes. So I'll go ahead and ask it. Instead of pouring all our effort into a single highly AAO optimized website, what's your opinion on launching 10 plus leaner sites each tailored to a different audience segment to multiply our chances of earning references for the same topic? So basically a bunch of little sites so that you can have a higher probability of being cited um multiple times or at all. What are your thoughts on that? I would not do that. And the reason I would not do that is probably none of the sites will rank. Uh so if we go over here, if you have a brand new site, it's probably not going to rank for anything at all in Google or or in chat GBT. You only start ranking when you have enough authority to rank. The exception would be something very long tail where there's zero competition. But generally speaking, you need some authority for the search for the search engine to rank your URL URL. What I would do is go to a bunch of Reddit threads, make a bunch of YouTube videos. You could have different YouTube accounts and make YouTube videos from those different accounts. You could uh get mentioned on all the affiliate uh sites. That's how you would get, you know, create 10 different mentions of yourself. But if you create 10 brand new sites, probably none of them will rank. Got it. And then there's a lot of people asking about um you know specific AO tools and and also just generally what's your opinion on the best tools. So maybe we could take a look at that list and just get your thoughts on differences, similarities of the tool, what to look for, how to pick one. Yes. So the tools do the following. They all do tracking. Some of them do API uh they ping the the API and get answers that way. Some of a small number of them scrape. Uh but that is a key difference. It obviously costs more uh if if you're scraping than if you're using the API. That's a key difference. The second thing is some of them will do crawl crawl stats. So what which pages did the LM crawl and uh and how much I believe that's basic and you know some are built for enterprises. Those are the main differences or sorry those are the main features. Uh I'll describe a few. So, Scrunch specifically is built uh they have a uh API access and they're more enterprise uh devel or they're more developer friendly. Uh if you had to pick I would probably pick from this table. This table is basically companies who raise money which is not necessarily causal. They're better but this is probably probably a pretty good short list. Uh Everune, Athena, Peak. I I know a lot of people are using Peak. Uh I think peak is also less expensive than than some of the others. Uh so that's my general set of recommendations. But yeah, I would pick one from this list or you could we we also have a tool. I would pick one from this list and uh if you're a developer, this scrunch seems to be the uh the favorite among them, but I think that they're all pretty similar. Great. Next question is, when do you think LLMs will start approaching search like a business the same way Google does? And do you think Google search will become more irrelevant with time? No, I don't think Google will become more irrelevant with time. Uh Google and and and Chhat will and perplexity will all just sort of merge and they'll all be similar and people will uh you know Google will no longer have their 95% market share but I don't think Google will go away. Uh the other question so what was the first part of the question? Uh oh yeah uh yeah what was it? It was when will they start approaching it like a business which my take was when are they going to start running ads? Uh so uh I believe Perplexity is already starting to test ads and I believe uh OpenAI hired people from different ad uh other companies that ran ads. So I'm assuming OpenAI is building ads right now. When do I predict that will appear? Probably like 6 to 12 months if I had to guess. But yeah, I I think that they're both building ads right now. Great. I think we have time for one more question. Um, how about um, someone was asking specifically for marketplaces? So, a follow up on the marketplaces strategy you were discussing. Is there any benefit in adding in question answer content on category pages? Definitely, and there is for Google search as well. So if it's something like best uh best dress for uh for dinner party uh and then let's let's take a look at this. So let's do uh used or uh Santa Monica. So this is a common strategy for both SAT and for search where apartments in Santa Monica and then there's an article with a bunch of FAQs. And so we we generally see this work pretty well. And I would do I would do the same thing for uh for uh LLMs. And I would also consider doing the thing that I mentioned, which is bring some of the metadata in that's hidden here. Bring some of that in your FAQs. Like maybe you would have you you could probably even have an FAQ about each of these different facet categories. Like a section about the closure and the fabric and the material and the neck style. like you could have FAQs about that because I think a lot of the follow-up questions are best product with these attributes, this color, this material. So, I would have FAQs around that. Awesome. Thank you so much, Ethan, for that awesome presentation and answering all those questions. And just so everybody knows, this session was recorded and will be shared out and we can also share the slide deck. A few people were asking that, but we appreciate everyone being here and thank you for your time. Have a have a wonderful day. Please add me on LinkedIn. I'm publishing more as stuff there. Thanks everyone.