Zero Click SEO for AI Search, Google Rankings & Agentic AI

E80 ZERO CLICK SEO FOR AI SEARCH, GOOGLE RANKINGS & AGENTIC AI

Transcript

Dalton Anderson (00:01.742) Welcome to Vich Step Podcasts, where we discuss entrepreneurship, industry trends, and the occasional book review. Today's topic is something that I've been curious about for some time, and it's becoming to be more developed and I think going to be a talking point in the industry of search engine optimization, and not only a talking point, but also

a required set of materials and skill sets going forward in the future. But I think everyone's caught a little off guard how quickly this technology is being adopted where

the expectations of where a LM would be now or the adoption from the user base or the movement of enterprise.

customers towards this AI centric workflow or products. They caught a lot of folks off guard, even the people that are providing the services and I think

I think it's just interesting, right? Where for 20 years, well, maybe backtrack for a second. This whole thing that's been going on for the last, mean, AI stuff has been around for a while, but these LLM products and the original beta version of ChatDBT and where we are now, I've mentioned it many times in the episode, blows your mind. Like how quickly things have progressed is insane.

Dalton Anderson (01:48.014) people can predict that. No one said, okay, we're to be where we are two years from now. People are like, this stuff's going to take 10 years. And then we're there in six months somehow with some massive breakthrough or there's just a lot of limitations that people had on their assumptions. And those limitations were broken through these.

just focused outputs or focused inputs to get an optimized output.

Dalton Anderson (02:22.892) And so what I think is interesting is that just the amount of adoption and change, not only for your industry, like my industry of insurance is seeing massive changes with AI. And if you aren't embracing AI, then you're going to be left behind. It's simple as that. And I think companies that are in denial of the importance of using AI and their workflows on a daily basis are the companies that will really struggle.

to progress because something that would normally take a couple bodies, throw bodies at the problem, just spend 60 hours building out the workflow using an AI agent or whatever it may be, some kind of generative solution, generative AI or an agentic workflow to build what...

you need to do. And if it's a repetitive manual administrative task, this can be easily optimized. And that's not to say that we didn't have solutions before like these RPAs and different things. The difference is that these AI systems take a lot less technical skill and they also have flexibility on different formats that are coming in. It's just a lot better. It's just a way smarter way of going about

optimizing and automating things. That being said, there's also some changes to like industries that haven't changed in a while, mentioning mine insurance, as well as the search engine optimization industry. So SEO has always been a foundational piece of ranking on webs, not websites, but on these search engines for the crawlers. And it's been the same way for

20 years, there's been slight changes to the algorithm, but basically it would be like you play by the rules, right? Where you, you have on page SEO, off page SEO, and then you have the technical SEO where you have on page, you have this like hierarchical, hierarchical usage of information to where you're using headers and sub headers and using H1, H2, H3, H4s. And then you have this

Dalton Anderson (04:44.813) contextual information below and the crawler and the engine is looking for these keywords that the user prompted into the system. And then,

that combined with off page SEO, which would be your authoritative rank, where you ranked compared to your peers in that cluster of information. And the way to get that is through basically like these backlinks. there's a little bit of contention, like Google says we don't use backlinks, there's proof that they use backlinks.

No one really, really knows, but it doesn't hurt. The back, back links gives you this, this authoritative rank as well as other things like how, how readable and usable is your site? How easy is it for search engines to crawl, which would be like the technical piece, like making sure that indexes or crawlers can index your website. All of that combined allows you to be a high ranked

website in your space. And so that is the entry ticket now for you to even be considered. Because when I was talking about fast adoption, the adoption of these

AI agents slash LLMs, however you want to phrase it. It's not really agent, but like a chat bot. LLMs searching for the user has really increased. as of, as of 2025, I'm reading this off as of 2025, 52 % of us adults report, if I could read, reported having used an LLM signaling mainstream awareness.

Dalton Anderson (06:43.117) Critically, two thirds of these users report using them like search engines for information retrieval indicating a direct substitution for traditional search habits. Next empirical information is while still a fraction of Google's total search volume, LM-based searches rapidly gaining ground. It accounted for approximately 5.6 % of US desktop search volume.

As of June 2025, a figure that has roughly doubled from the previous year, more recent studies suggest LLMs now account for as much as 27 % of all searches in the United States.

Google AI overviews are designed to provide a comprehensive synthesize answered directly within the search results.

for the user without needing a click to retrieve from an external website. Studies confirm that the impact of this model with nearly 60 % of searches in the US and EU now ending without a click to a non-Google property.

So basically, these AI overviews or Google AI overviews are resulting in what is being deemed the zero-click search, where people are searching for information and they're finding the information they need, and the answer is rich enough for the user to deem that they don't need to click anywhere. So that's kind of a problem if you want clicks on your website. And then another thing I found was interesting when

Dalton Anderson (08:28.855) I was doing a little bit more reading about this was that the models, each model and company has different parameters and thought processes on who they cite and how they cite and how rich are their citations. So Google in perplexity, which perplexity is a little less well known perplexity is a AI company that specializes in search engine stuff. And so they're trying to

replace Google as the AI native search engine. Still think it's gonna work out that well, but good luck to them. It's gonna be really tough. It sounds like a...

piece that you hear all the time where is your product a feature or is it a product? Is this a product or a feature? And you've seen repeatedly over time with these massive companies like chat, GPT, Google, Microsoft, implementing these features that just wipe out just hundreds of startups because they were a feature, not a product.

And so for Plexi is trying to become a feature, not a feature, become a product. I just don't know how they compete at the scale of Google.

maybe they could have a niche out there. But anyways, don't want to get us too distracted. Just trying to explain perplexity as a company. Back into what I was talking about. So when I was doing some research, the different models have different parameters of how they evaluate sources and site. So studies showed that perplexity and Google cited more often with a roughly

Dalton Anderson (10:23.481) six citations per query, whereas Chat GPT had less citations at three. So Google and Perplexity had nearly double, I say nearly, but these are all, I say nearly because it's an estimate, because it's a couple thousand queries, they're basically taking the average, so we can just call it double, whatever, who cares. So they have double the query, they have the double the,

citations for query, which was interesting. And then there's different models, right? So chat.qbt4o is authoritative seeker is how you could describe it is that it heavily favors factual information from authoritative sources. And one of its main sources is Wikipedia.

alongside major news outlets. So it's trying to stay away from things that might cause hallucinations or false information. So unless you're like heavily vetted, you're probably not going to get cited by Chachapi 4.0. So the people on top stay on top in that scenario. And then Google Gemini AI overviews has broad aggregators. It is a broad aggregator where

Most of the time it will reference wide range of sources from community forums like Reddit, Quora, social media platforms to like random product blogs, such authoritative experts in a certain area.

Dalton Anderson (12:10.253) with the vast majority of those queries 82 % of the time being what is deemed rich content pages that are cited versus the homepage. Perplexity has it broken down into two categories, which I thought was interesting, where it has a personal search query and then a enterprise search query. So the personal search query has emphasis on community forums, likes to reference Reddit and these other forums.

like websites like Quora.

Then businesses typically reference industry niche publications or reports or data-driven blogs, vendor blogs. Whereas the personal version references forms and opinions and gathers that information. B2B searches, it's just, okay, here are the...

Most useful information I deem to be useful, and that's how they understand, this is a business search from an enterprise. We then need to reference these industry publications or these data-driven blogs from vendors, and that's where we'll get our information. All of that, I find, is very interesting because

it definitely disrupts what was the three pillars of SEO. Like for the longest time, as I talked about, is on-page SEO, off-page SEO, and 10 new SEO. So the old goal was just to be part of this high-value keyword. And so you'd have these high-value keywords that you would try to get authority in. And by doing that, you'd have to identify the low-value keywords and avoid those. And then you'd have to see what the high,

Dalton Anderson (14:12.311) cost per click keywords were and then use those in your site but also there's people lot of people paying for those so they're very competitive so you'd also have to attack the medium and the lows and then slowly gain your authority and hopefully you're ranking naturally for high value keywords that are expensive.

So as I mentioned, there's just like cracks in the foundation where this AI is synthesizing this information where the user is no longer clicking on the site. And if you ask questions, it surfaces things either with a mention, like a direct mention in the query or a reference. And these things will look a little differently depending on which app you're using. But if you ask a question like, what are the best shoes for running?

it might tell you brands or it might give you a direct link. Like here are the top 25 shoes for 2024 or 2023 or whatever. They might reference some random year and you'll have to tell it like, I actually need for 2025, but I'm saying that because I recently have it to be, I was so sad. was devastated.

But in certain scenarios, like Chatty-Patee or Plexity, they'll have the actual link to the website and it's kind of like a shopping square where you just click on it, like a, what is it, product card. It's a product card, like if you would see on a website. And it's similar for other sites, but I think that a mention or reference is more valuable in certain scenarios to where as if you're referencing for like shopping.

If you're trying to buy something, think a reference would be better, right? Whereas you'd want to rank for both, but I think in certain scenarios, a reference or a mention might be more useful.

Dalton Anderson (16:11.105) Okay.

So we have defined SEO, right? We have defined that, or we have not defined, but we've illustrated that there are some serious growth trends for these LLMs. We are almost reaching critical adoption in the United States. There is citation stating that majority of user queries

in with no clicks to the actual website.

Dalton Anderson (16:48.654) So this seems to be a paradigm shift where you're no longer optimizing for the user. What you're really optimizing for is for the engine, the machine, the AI machine, all hell, the machines.

Dalton Anderson (17:10.822) man, I crack myself up sometimes. But in all seriousness, you're optimizing for the machine now. You always were optimizing for the machine at one point, right? Like for the crawler, for the technical pieces, making sure their website can be indexed and all these other things. But now it's integral that the machine is first and foremost. And that makes a

difference where you're trying to become the foundational source of truth for AI not for a human because for you to get in front of a human you've got to pass the vibe check of the AI and if you don't pass the guidelines parameters of what the AI is looking for then you'll never get referenced and you'll never get mentioned.

And if 60 plus percent of people aren't even clicking a website, then your outcome is pretty slim. It's pretty low. And that's a grim thing to think about. So what you really need, you really need to optimize for the machine. I like to say it like that because this is so funny in this episode. So the machine.

So there's this shift, right? Clearly define it's there. You know, people that have started using LLM, like your grandma's probably using LLMs. My Nana does sewing stuff with LLMs where she asks it to build designs and help measure different things and make blueprints. And obviously she had somebody cool in her corner, but I just told her what websites they use and gave her some examples and she just went off and ran with it.

She doesn't need much help and it's self-explanatory. She's very technical though. So like she's kind of a beast when it comes to computers and phones and stuff compared to her peer group. She's a beast.

Dalton Anderson (19:25.602) Okay, so now we've reached this tipping point. This isn't a trend, this is a complete change in customer attitude and perceptions in usage of how they're behaving. recent studies show that, like it's massive. It's there and it's not gonna decrease in usage.

Right? This AI overview piece is here to stay. That's very clear. So then what do we do in this zero click world? When we're in the zero click world, how do you become relevant if relevancy is determined by the machine? Well, you got to start optimizing for the machine. So what is that called? That's called geo generative engine optimization. The generative engine optimization.

is optimizing for the machine, but there's a couple things that you need to do, like guess key tactics, and this is still developing, so it's not a for sure thing, Like it's advice and studies and experiments, but this could change on the flip of coin in two months from now when they do these massive updates on the models and then this could all change, who knows? But as we speak,

Key to act is the E A T. Experience, expertise, I said A right? Authoritativeness and trustworthiness. So basically, what's the experience, expertise, how authoritative are you and what's your trustworthiness? Similar to SEO, the difference really is the conversational language, like making sure that your information is conversational and easy to source and it's

in a structure and a manner to where there's not information hidden and everything's there. And so it's going to be a lot more information than typically a human would be interested in because what's different from SEO and geo is large language models use semantic search versus SEO uses keyword search. So keyword search is a word, right? So you have a word. it says, I want to play baseball.

Dalton Anderson (21:53.864) at or I want to see baseball at Fenway or I want to watch soccer at Anfield. And so there's keywords that are associated with soccer and Anfield and watch, right? So it knows that you want to watch a game at Anfield, which is owned by Liverpool FC. And Anfield is in Liverpool, England. Or yes.

I was thinking like, as I the other way, ingloriful, that doesn't make any sense. So all this being said, those are things that are linked together, right? So they're linked together. So the machine knows in this example that these keywords come up together often, like a keyword pairing, whereas semantic search understands fundamentally

the user's intent, what they're trying to say, even if it's misspelled, what they're trying to do, even if they don't explain it well, and taking the context of not only the words that are keyword specific, but also the context of the full query that was sent to them. And then the context of the chat that they have. So it has a way better understanding of what is going on around them, or her, I don't know.

The machine, I like the same machine, but whatever. The machine understands your thoughts, your life, everything. I'm feeling pretty funky today, by the way, I don't know. I've had the most productive day I've had in months after getting sick, so I'm just buzzing right now.

But okay, so the machine understands the full context, whereas SEO just understands the words and they're linked together, but doesn't necessarily understand why they're linked together. No one told them why. So the machine doesn't know. Whereas semantic search, it understands that all that stuff. That's part of the LLM. The LLM can link the different words and the intent to understand what's going on. And so it...

Dalton Anderson (24:14.923) unlocks a different perspective on this lexicon or like this lexicon. I don't know if that's a word like a lexicon search, like the lexicon where as before couldn't do that. And that's why it's important to provide the machine with full context on what you're trying to do or what you're trying to talk about versus just talking about it. And

it's a little different where as before you were kind of just, I'm, I'm trying to get these, these keywords on my page and reference them and they get some backlinks and then build a whole bunch of different posts and that builds my authority and get people to visit there. And then if they visit, they don't immediately leave. And that builds my authority some more. And I've got some more references of these keywords I want to rank for. And then I'm Whereas this is you need to provide

useful information that has all of the context that the LLM is interested in. And then they need to have a little bit more context on what's on the page. You can't just list out stuff. It has to be made in a way that

is structured and the machine can understand reference and compile your information on your page.

Dalton Anderson (25:48.93) So the next piece that's kind of part of the zero click era is this agentic AI optimization, AAIO.

This is a newer segment, even newer than GEO, whereas AAIO hasn't really started as much yet, but it is starting where you're seeing that agentic workflows are being offered at an enterprise level from Google, Microsoft. OpenAI has some stuff that they offer at the enterprise level with their Ultra, or they don't have Ultra.

I'm not sure, I think there's this called Pro, Chachipati Pro, and then Google is called Ultra, like their highest tiered model.

offers the ability for agents to do stuff on your behalf. They just can't do things that require them to log in. And if they have to log in, they require you to approve the login and type everything in, which is a really a bummer. Like you can't give your passwords to the agent and then have them do stuff for you. So that kind of sucks. But looking things up and researching and doing those things on your behalf and like typing in information that doesn't require login, agentic workflow all day.

Okay, so this is a little bit newer, but it's going to become more popular in the future. Like there's going to be a lot more AI agents doing stuff.

Dalton Anderson (27:26.255) which is a little bit different, right? Where it's like geo plus this other stuff. I'm not as familiar with it because it doesn't really exist. mean, it exists, but not built out as much as geo is. And all of these things are rather new. But they're stating in what I was reading is this deep schema markup.

Actual micro content and then APIs for your agent to reference the content, which would be interesting. Why wouldn't it just go to the website?

and then hierarchical information, which makes sense. So the way that I was reading it is, I think the biggest one would be the schema markup and then this hierarchical SEO. So think about an agent as a person. And so if you go to a website and your website's just misorganized, can't find anything, search doesn't work, you don't know what really is going on.

agents not going to know either. Sorry to tell you. if you have, if you have a hard time doing it, then the agents probably gonna have a hard time doing it. But if you go to a website like dictionary.com and they're looking for random letters that start with F or letter, I'm sorry, letters, random words that start with F. So when you go to dictionary.com,

It's in a structured schema to where when you get on the website, you see Fs and then you click on F and then they can scroll down. They could search all the words they want that start with F, extract the information and come back. That's a way different scenario when you have all these random web pages that have words that have F, the Fs on them. And it doesn't allow the agent to parse the structure.

Dalton Anderson (29:32.943) of the website and I guess you'd have to explain it in a way or I need to explain in a way where you need to structure the website where it has ease ability of understanding by an agent and the easiest way to do that is to have it structured in hierarchical so the agent can understand the semantic context of your website and where it needs to go. So there's a kind of a lot there but

I just thought this topic is so interesting because I'm working on these blogs. Checking the time here. I was working on these blogs and I was like, hold on, hold on, hold on. I'm spending all this time writing these articles, but who am I writing for? Who am I writing for and what am I trying to do? Well, I'm trying to become a Spotify partner. I'm trying to grow Spotify, I'm trying to grow YouTube.

So for me to do that, I need more visits. And the good way to do that is to repurpose the podcast content and make articles so then people could search stuff online, they could see it, and they could find what I'm talking about interesting, and then maybe they watch this video.

Dalton Anderson (30:53.423) And maybe not, but it doesn't hurt. And also you can get referenced by AI and these other things. But then when I was really thinking about it, I was like, okay, in the future, who am I writing for? It's not a human. It's gotta be AI. And I there's been talk about the geo and these, this like,

other type of SEO, the AI optimization piece, the, and this agentic AI optimization, he's been talking about it, but it's becoming increasingly more important. So then I was like, okay, well, if I'm running for these folk, what do I need to do? And so then I just did a crazy amount of research, then I made AI gems.

or Google gems like these AI templates that structure my transcript of my data into, you hear this, you gotta hear this.

Dalton Anderson (32:07.535) a readable file that allows the LLM to understand the full context of the content that they are referencing or searching for to give me a higher likelihood of a mention or a reference.

Dalton Anderson (32:30.169) pretty neat and I made these little see if I could share this screen.

screen.

Dalton Anderson (32:41.903) That's kind of a bummer.

Can't share that screen.

Dalton Anderson (32:49.379) Huh, all right, well, we'll do it anyways. I'll just have to share my window. So I'll share everything with you. So let's do this. wait, I can share, I can share, cool. That's pretty sweet. All right, nevermind. I take that back. Okay, so these are my little gems that I have. I like Google, I like OpenAI. I lean more towards Google. I switch back and forth.

but for gyms or these templated.

workflows. I think Google is superior in many ways than OpenAI. Okay, so I had built some things prior and you can see that I have an Obsidian VentureStep Templater. I've got the VentureStep Episode Engine, which is what I do when I put together all my notes and my research. I put it all together and I give it to my Templater or my Episode Engine and then the Episode Engine formats my

notes and research all into an outline that's structured and is in the same format every time. So when I do this podcast, I can, if I, if I need to, can just quick glance. Normally I have stuff printed out and I could just reference it, but that's a different matter today. I wanted to show, so I've got this VentureStep image generator, and then I've got the Ventures VentureStep article engine.

And then I should have one more.

Dalton Anderson (34:24.399) Oh, well have this template so that's why I have three. Okay. So if we look at this, I'm going to share my secret sauce with you. See.

Dalton Anderson (34:37.583) I should be able to edit that though. Okay. All right, so let's do edit. So I'm show you. All right, so it took me four versions, took me a lot of tries, a couple of tries, more than I would like, but first we define the role of who we are. We're the SEO specialist and our primary mission is to convert raw podcast transcript into a comprehensive high value article.

for the ghost blog. Ghost is a company that I use. They're an open source project. They share all a hundred percent of the revenue proceeds directly to the content creator. And it's overall a great project. If you want to support it, feel free. That's who I subscribe my podcast to. Didn't want to use WordPress because of all the stuff they've got going on with their person trying to take over the nonprofit and such. Anyways, so

I've got this system prompt that it follows and then the output looks like this. So go to recent one I did was robots.

robots, brains and future independence, which was a.

a episode about.

Dalton Anderson (35:53.912) NVIDIA, Neuralink, a couple other things. right here it says keywords. NVIDIA, Project Groot, Neuralink, Human Trials, Robotics, AI, Foundation Models, Brain, Computer Interface. Interesting. Talked about that a while ago, like over a year ago. So this first components, this is my SEO title that it's creating. This is the meta description for the crawlers and such. Keywords is more of like a reference like

If I want to use these keywords on ghosts, could, don't necessarily. And then this is the URL that it's suggesting for me to use. I change this stuff sometimes. Most of time it hits it on the mark. So then I have Obsidian Knowledge links that I have to remove, but I like Obsidian and I like to link stuff. So I try to link things. And if I don't have links, I make links. It structures your article. I've got the article.

and then the too long didn't read for social media. And then here's where we get into it right here. Introduction. So it talks about what the content's gonna be and then has all the key takeaways, key takeaways for the AI to reference. And it's got the full conversation of the transcript, but it's enhanced where it's structured. So it has the intro, it has what it's talking about, Introduction Project Group Neuralink Human Trials.

It says exactly what I'm saying. And then it pulls out little snippets for me. So it says, understanding foundational models with a

with a Lego analogy. And then it pulls out a quote, a foundation of models where you have these three basic Legos and you can build a Batmobile or a Princess Castle without having the instructions. But to do that, you need a solid understanding of your environment, a solid understanding of physics, and a grasp of how your body interacts with external environments. These are crazy things like when you step as a robot, how long should you wait until your next step? Or how much force is there in the rotation between steps?

Dalton Anderson (38:00.721) How do you lower your center of gravity on a slippery surfaces? All these crazy things that you wouldn't think that are that complicated are pretty complicated for robots.

Dalton Anderson (38:12.068) Super sick, so cool. So I'm structuring my transcript. I'm keeping all the content for the transcript, but I'm structuring it and making sure that it's useful, readable, and accessible with the full context of what I'm stating, why I'm stating it, and why it's impactful. And not only am I doing that, it has these key call-outs so the AI can know that like, hey, this is another quote.

or this is another crazy thing that he's referencing.

And then all my references are at the bottom.

So.

Yeah, very, very interesting stuff. Very interesting stuff. I just thought I'd share that. That's what I've been working on. Why I've been so interested in this. And I've been thinking about it for a while. Just wasn't at the point where it seemed like it'd be important. It's more like a pie in the sky thing, doll, and you're always in the pie in the sky. But now it seems as though way more important than it used to be. Stop sharing.

Dalton Anderson (39:27.704) which is a way different structure than what you're used to when you're looking at these articles. Whereas an article is more traditional, it's got these hierarchies and it's got some information where this is like 5,000 words of just very dense material that has structure to it, but it's not meant for humans. It's meant for the machine. All hail the machine. But.

That being said, I really hope that you enjoyed this episode. Bringing the energy, trying to get a guest on the show for next week. Would love for Luke to get on the show and talk about how you guys influencing his industry. Think that would be very related to this content right now. But of course, wherever you are in this world, have a good afternoon, a good evening, a good morning. Thanks for listening and

I'll see you next week. Goodbye.