Inside

SEO Week

Episode 23

Metehan Yeşilyurt

Co-Founder & Chief Growth Officer // AEO Vision

Metehan is a Growth Marketing & SEO Manager with 10+ years in SEO, AI, and digital product innovation. He regularly speaks at conferences, meetups, and team events about the future of search and AI visibility.

Key Takeaways

  • Training data is becoming a critical factor in AI search visibility, not just real-time retrieval.
  • SEO is evolving into reverse engineering of LLM behavior and data pipelines.
  • Agentic systems will fundamentally change how users interact with search.
  • Google’s distribution advantage remains its strongest competitive edge.
  • Prompt injection and system manipulation are still possible across many LLMs.
  • The fastest way to learn AI search is through hands-on experimentation and technical exploration.

Transcript

Garrett Sussman: I can’t believe it. Okay. So two things I am so hyped about. One is SEO Week is next week. It is here. I can’t believe it. When, when, not when we’re recording this, we’re recording this a little before, but like when you watch this, this is, it’s coming up next week. And the second part is I am having like, I’m fanboying for a moment. I’m getting to talk to the man here. Metehan, I know you’re going to blush or whatever, but Metehan, who is the Co-Founder and Chief Officer, Chief Growth Officer at AEO Vision. And you’re just a dude that I just love. I love what you’re doing. I love talking about this stuff with you. What’s going on, man? How are you? 

Metehan Yesilyurt: Thanks. Thanks for having me today. It’s going great, actually. And I’m super excited for the SEO Week, you know, and currently working on new experiment, new stuff and trying to manage the clients requests and everything is happening fast you know because of this AI hype. 

Garrett: It is nuts. I mean you are in DevChrome tools reverse engineering ChatGPT, AI Overviews, you know everything like AI Mode, Gemini, kind of getting into the weeds. You’ve written all these amazing insights of your discoveries, doing really fun experiments like the ridiculous prompt injections one, which I put on the iPullRank blog – I’m sure I got in trouble for that for a minute. But you’re doing the work, and that’s what’s so interesting about our field right now. So I want to ask you, Metehan, what is top of mind in AI search? How are you thinking about our industry right now? 

Metehan: Yes, everyone went crazy. You know, on the other hand, there is a new, it looks like a new market, but I still believe SEO is king. SEO isn’t that, but that’s another debate topic. I will skip that part. And, you know, every development is exciting at the moment. So I really love working on ChatGPT, Perplexity, Gemini, Claude. And now we have some agent stuff, some coding models, you know, and everyone, actually every client, of course, you are also facing this situation, demanding too much stuff in very short deadlines. And I’m right now currently trying to build the best agentic SEO and AEO orchestration for my work and for the future. But yes, everything is so hype and I’m really, really glad to be in search industry. 

Garrett: I can’t wait to hear your presentation. I feel like the next three months, six months, 12 months, there’s going to be a lot more noise. There’s going to be a lot of changes. What do you anticipate? How are you trying to stay ahead of what’s happening in terms of consumer behavior, what the different LLMs are doing? How are you thinking about what’s coming up? 

Metehan: First, I always try to take a step back as a user. Okay, just leave the algorithms leave the probabilistic design of LLMs because every every – even you ask the same question, okay context stays the same but citation changing, sometimes clients just ask okay we are we can see our brand in this prompt can you check I type it but no response etc. etc. But as a user, I’m trying to understand also these developments because it also benefits my work and my life. Suddenly I find myself in Google’s AI Mode, ChatGPT or other models or AI Overviews, I just start to ask too many questions on internet, so – it’s it’s super cool by the way – and today I just asked a question and AI Overviews just came up. Then I felt myself, okay, I need to ask one more question, but how can I do it? Then I realized, okay, that’s why Google also released Google’s AI Mode, because I can continue my research or question to get more answers. So everything is fascinating, of course. 

I’m not happy with the e-commerce developments with the AI. You know, ChetGPT just came out with another exciting idea. And they basically told us, okay, we also want to disrupt all e-commerce area with the agent stuff. Agents will start buying too many things, etc., etc. But for the next month, for the next weeks, I’m super excited for my talk at SEO Week. This is my first priority at the moment. Still working on my presentation to deliver the best, one of the best presentations in the event. And I know too many cool names will also speak. And second, I’m really excited to meet you in person. I told you before, I couldn’t find any chance during these years, my previous event engagement, etc., that’s another topic. But for the search industry, probably we will see more agent stuff, I believe. And my current point is I’m trying to build an SEO, AEO, GEO, agentic orchestration. But we’ll see. And everything is happening fast. You know, we’ll see ChatGPT 5 4 3., then a week later, we are trying to deal with, okay, how can we see these query fanouts? And we can see also query fan outs in Google’s AI Mode. Please Google, please share some clue with us. I’m really excited for that. So yes, everything is happening fast. This is my first thought. And then yes, we’ll see. 

Garrett: Let me ask you this. So you’re in Chrome DevTools. To what extent is it, do you have a method when you’re trying to do your research and discover new things versus trial and error? Like, are you just looking at the code and hoping to stumble upon, oh, you know, that’s where, you know, they actually did kind of hide a nugget of query fan out? Or are you kind of systematically looking through different sections of the Chrome DevTools with updates and whatnot? 

Metehan: Yes, first I started clearing my X feed to see more quality and qualified tweets for my opinions. And I believe I succeeded on it. So I see only feedbacks from the search industry, how people is trying to understand these probabilistic design systems. I mean, for the LLMs. And second, I love running deep research with some open source models. I won’t name it, but hopefully in the event. So everyone is invited for the SEO Week. So let’s talk also during the event. Please find me, find Garrett and everyone from the iPull. Yes, deep research. Then I really love digging Reddit. You know, people love sharing their experiments. And on the technical parts, I really started listening to some browser events and server-side streaming events. So month by month, when you find new clues on these events, of course, it needs to be processed. And you need to understand how the mechanism works. But especially if you really, really love listening browsers, actually you can find more clues. So that’s another way to discover new findings on my stuff. Especially in Perplexity, they were just used a very basic crypto schema, and it was really easy to decode and deobfuscate all the code. And then suddenly everything started to appear. Okay, okay. Yes. 

Garrett: What do you think is going to happen? Do you think Perplexity will get bought? Do you think OpenAI will actually challenge Google? Because we still know that Google has 90% of the market share. And like Claude, you know, Anthropic is obviously making, trying to earn their visibility. Although I don’t think anyone really thinks of Claude or a Cowork, despite OpenClaw as like a search engine, like if anything, it has a better chance to kind of get into the agentic flows of things. Who do you, like which organization do you think will become dominant? Or do you think it’s just Google’s to lose? 

Metehan: First, Google has a really, really distribution advantage. If they just want to release a new model, they have YouTube, Gmail, Android, Google Search, Google Ads, you know, too many, too many products. But yes, on the other hand, the other LLMs, ChatGPT, the OpenAI, also Grok, X, Claude, of course, they are very, very successful in the coding space. So it’s becoming like an enterprise and technical people loving place. It’s becoming their second home right now. Of course, it’s also valid for me. Yes, I love using Claude. But yes, these new developments and products models are very disruptive. And it’s forcing Google, it feels like, this is my personal opinion. It feels like, but we also know Google acquired DeepMind more than a decade ago. So they actually, they’re able to release all of these products. And they already, I believe, invented all of this stuff. But these new models forced Google to release them also in the public space. So as a user, yes, I have too many benefits of using these models. But Google has also really has a huge spam experience. So in chat GPT or other elements, you can just play with some prompts and you can be successful with prompt injection attacks. It’s so basic, but Google knows how they can manage it. 

Garrett: I know, but we are so dangerous as SEOs because it’s like it always is the cat and mouse game where it’s like every time Google plugs a hole, like, we find a way to do something, you know, manipulative or, or like “ruin the internet” as, uh, as The Verge said at one point. Um, give me a peek as we get to the end of this, like into your talk, you talk about, um, you know, reverse engineering. You’re talking about the, how it selects different types of content. You’re talking about entities, information gain, um, confidence in hedging. Give us, like, a little peek of what people can expect to hear from you at SEO Week this year. 

Metehan: Yes, there are also some connecting talks with my presentation, but I really want to show everything in depth. Also with some maybe mathematical formulation. I’m still working on that. But especially if you just ask how to make a bread in any LLM, it won’t make a web search and it will bring some, you know, the training data. So training data is very crucial also here in this retrieval pipeline. And I really want to show at the same time how these training data creates a multiplier for your brand and how can be more visible in training data. How can you influence these training data? I still have ongoing experiments with this. And recently, you know, I made a collaboration with the Common Crawl, one of the very popular datasets for almost every major LLMs. And there are some major publishers blocking CC bots for crawling purposes, etc. That’s another topic. But yes, it will be very technical. It will include some training outputs with almost every major model. But yes, for Google and ChatGPT are very popular. So yes, we’ll see many technical retrieval pipeline and actionable tips for every attendee. 

Garrett: Dude, there’s too much to talk about with you. Because it’s like, I forgot. It’s like, if you guys have it, we’ll include the link. You got that really cool common crawl tool, man. 

Metehan: Yes, I’m still thinking on that because I will have a limited time. I know. And I really need to provide the best information for everyone. 

Garrett: Well, you put out a common crawl kind of for a brand to look and see their visibility of how included they are. And you know, like I went with iPullRank.com immediately. I was like, okay, where are we? How do we compare to competitors? Because to your point, it’s like the training data, if the queries don’t need to be grounded in any sort of like, you know, RAG pipeline of like current search, then that’s going to be a major definitive advantage. You know, that you’ve got, I’m curious about your experiment that you’re running with, you know, the AI slop and, you know, how you’re getting more LLM visits compared to, you know, traditional SEO visits at this point. You’re doing so many things. Let me finally ask you before we close out is what are you most excited about for SEO Week? 

Metehan: For SEO Week, first, I will, this is my first big trip to United States. So it’s personal. But it’s really, really exciting because I was biting my nails last year. Unfortunately, because of my calendar, I couldn’t come up. But I talked with my many friends and clients in the United States, and they were all engaged with all speakers, with the event space. You have a gorgeous event space, and, you know, talking with and skip the basics is the best thing during an SEO event. And, you know, you can just sometimes you find yourself, you are just talking about basics. But in SEO Week, you just skip the basics and really talk about the serious stuff. That’s the best. So I’m super excited to meet with new people, seeing friends and try to also visit some popular places. 

Garrett: Dude, you got to do all the touristy stuff and make sure you get a good slice of pizza. And then, you know, you see Times Square. All that is going to be so much fun. I’m so excited, Metehan, to meet you in person. I’m excited to hear your talk. For everyone who’s watching this, you got hopefully by the time this is published, we actually maybe still have some tickets left to scoop up. So buy your tickets the 27th to the 30th of April in New York City. Come see Metehan speak on Day One. Even if you can’t do the whole event, at least come to Day One of the Science if you really want to get your hardcore engineering science technical fill with him and Mike and Scott Stouffer and Jeff Coyle and all the brilliant folks. I can’t wait to see you, man. We’ll see everybody there. Thanks for joining me today. This has been great. 

Metehan: Yes, thank you. 

Garrett: Thank you, and see you in New York. Peace out.