Ep. 29 - SEO in the AI World with Victor Pan | Devs:After Dark

to please our overlords. That's why I always try to start every episode with, when the AI takes over and they go through and they comb through these old episodes, I am completely on board. Thank you for taking over, and I appreciate everything that you're doing, AI. Thank you.

Like mispronouncing the people's names, stuff like that, you know.

Oh yes.

Notorious for mispronouncing like, like Carters, everybody. So yeah, but all right, so here we are, another episode of Developers After Dark. Thank you for all the rock stars out there that are watching and listening. We are back after a little bit of a break. Feels good to be back. Dennis, does it feel good?

Indeed. Oh yeah, man, I'm revved up, as they say.

Excellent, excellent. Now we're doing a little bit of, uh, what I like to call a mini-series, and we're hopefully we'll have a couple of episodes talking with multiple people about this, but essentially one of my favorite topics, and that's going to be SEO in the AI world. Now we have what I would consider an SEO rock star that we're going to talk to about these, uh, this topic and everything that surrounds it. But before we get into that, let me really quickly just tell you, if you haven't already seen the only website that you will ever need, and that is devsafterdarkpod.com, and twice for Dennis, devsafterdarkpod.com. That's where you're going to go once I update it, because as I said, we've been a little bit busy. the last couple weeks. But, uh, you're gonna go there and you can see all of our previous episodes. You can see more about the show. Uh, you know, eventually when I catch up with the show notes, you'll be able to read some of those as well. Uh, but we encourage you, please go take a look at the website as well. But most people, if you're going to be watching this on YouTube, you're probably going to be seeing us from the HubSpot Developers YouTube channel, and that is @HubSpotDevelopers. Go there if you haven't already. subscribe, click like on just about every video you watch unless it's got AJ and Dennis in it. If I'm in there, go for it, click the like, that's fine. But yeah, no, there's multiple topics by multiple not only HubSpotters there, but then you have also the community as well that contribute videos and on just about every single topic. But that's going to be the place you're going to want to go if you want to know about anything going on in the HubSpot community. Now, being in the know also equally important. We also have to our changelog as well. That's developers.hubspot.com/changelog, developers.hubspot.com/changelog for all those rock stars out on the audio. Go there. You can subscribe to that. It's going to tell you about all the things that are recently happening and upcoming things as well. So you'll be able to get you know not only the now. But in the future as well to figure out what's going on in the developer community. Last but not least, of course, that is if you want to get your knowledge and you wanna uplevel, if you wanna go from a level 7 wizard to a level 13, you are definitely going to want to go to academy.hubspot.com. There you are going to have courses anywhere from, you got, I mean, social media marketing to content marketing. I mean, there's SEO courses there you can take too from basic beginner to advanced. And then you've also got stuff too like the developer's certification for the CMS. There's, I think there's version 1 and then there's a 2 as well. And you can get, and the nice thing is too, and a quick plug to the marketplace here at HubSpot, for those who are in the marketplace, right, that list their either their apps or their themes or modules, those certifications carry over to your listing pages and show in the about section for the actual provider what certifications you have in HubSpot. So you get a little bit of a humble brag too as well to throw on top of already acquiring some really, really great knowledge. But nevertheless, enough with the bill paying. Let's get to why we're here today and the reason that I'm super excited about this, because again, a topic I love. But first, let me let our guests introduce themselves. Victor, tell everyone who you are and what you do.

Hi everyone. My name is Victor Pan. I am an SEO at HubSpot. I've been at HubSpot for eight years, and I've been an SEO for a bit longer than that—at a startup for two, at an agency for four—and yeah, I am currently working on this transition that we're seeing with user behaviors. You know, where some folks have started to. really use LLMs to make a lot of decisions versus, you know, back in the day it used to be, let me Google that for you, kind of as this snarky remark, right? And if you're an SEO, what do you do? You care about being a liaison between what people are searching for and what search engines are able to provide. And you're kind of like in that middle as that whisper between the two sides, right?

Yeah.

I have a background in marketing. And I understand how, you know, businesses operate and how, you know, the human psychology behind a lot of the things where search engines really wanna capture what a user is looking for, but at the same time where a marketer can really influence human behaviors to change. And so it's not always an end-all be-all, like look at it from the technical side. It's also a little bit of, you know, real life stuff impacting how technology evolves. So, uh, yeah, that's the general just about me. My pronouns are he/him. I think I might have said that already. Uh, I'm a father of 2 girls, uh, and, uh, I am bilingual. I speak Mandarin and English.

Cool. May I start the bar? Can I start the bar very low? Uh, Nick, just very, very low. Um, that way we can really build easily. You're welcome for that. Uh, you call yourself an SEO, so if someone— I just don't know the acronym. So, uh, if you're an SEO and I say, uh, can you make my SEO better, is that like, you know, improving, improving your facial features? How— what does that mean? What's an SEO?

Uh, great. Uh, so SEO stands for search engine optimization. Generally speaking, depending on where you are in the world, that might be a different search engine. So what is a search engine? Google Search is a search engine. Bing, the default search engine on a Microsoft browser. If you are from South Korea, that'd be Naver. If you're from Russia, you probably know that as Yandex. And then if you're in the Czech Republic, I think it's now Sesnam. So again, it's a little bit different depending on where you are in the world, but generally speaking, in the U.S. markets, people are talking about Google. Sorry, Bing. That's just how it is.

And if you're from the '90s, it's Ask Jeeves, right? Yes.

Yeah. So I know this type of SEO. I just didn't know that you could actually have a title that is SEO. Is it search engine optimizer? Or are you blank?

I've always found that weird because that, like, in the SEO world, like, the job titles are— they, they're vastly— they range, right? Like, because there's, there's no, like, one particular title that I've ever seen. Because, I mean, I did it for, I don't know, a little bit under a decade at— in the real estate world. But it was like, even searching for that, it was like you could get a wide, wide range of like what some— because a lot of places didn't know what that— like, it's not that they didn't know what like search was, but they didn't know how to classify the work that you would do to achieve success, right? Is that a good way of putting it?

Uh, yeah, I would agree. And I would say like, you'll have other roles and then there'll be like jobs and responsibilities includes SEO. And you're like, sometimes people just don't even know that, hey, you just posted a paid search position and that person's main duties are managing ad budgets and doing audience research and running experiments. And then out of nowhere they'll be like, oh yeah, and then we expect you to know a little bit about SEO, not knowing that those are very different worlds that have similarities. But it's, again, you can get down to pretty deep rabbit holes there in terms of weird nomenclature within SEO. And just how the general public likes to use terminology. So yeah, I hear you there. The titles are whack. Just like I used to work at an agency, like agency titles are crazy. Like you join in, you're entry level, but you're a manager. And then it's like, how many reports do you have? Well, we work in a cross-team matrix and therefore have zero reports, but I'm a manager, right?

That's right.

So is it safe to say you've heard the term, can I— can you SEO this for me? Like probably hundreds of times, right?

I think that's the fun thing about that, right? Like for me, when I hear that in that context, it's, can you make this better? Right? And then now the question is like, okay, then what are you expecting? It's kind of like asking a designer, can you make this pop? And you're like, what do you mean? Like, do you know?

Like, I always love getting the questions of, hey, I have this new webpage, I want it to be number 1 on Google. Oh sure, not a problem, just let me—

Yeah, yeah. And it was always the question of, in what in particular? I always tell the story, um, when I was doing this for, uh, like individual real estate agents, right? Like we would have a consult, right? And it wasn't— it was like design and it was like building them a website, but it was also their their social media and like web presence, right? And, and, and, and sweet, sweet woman, an older lady who, uh, you know, obviously like the SEO and like the tech world wasn't her thing, right? So it's not— it was, it was definitely not a knock against her. And I don't, I don't say this to like poke fun, but she, the one day she handed me this printed out list of comma-separated just words. And I was like, okay, you know, and during, during this talk, and I was like, okay, what is this? Oh, these are the things I want to, I want to be number one for when people look me up. And I was like, not reading the list right off the bat when she handed it to me, I was like, okay, that's, it's an ambitious goal, but you know, I'm like, why not? And it was like, because we were in the Chicagoland area, it was the Chicago White Sox, Chicago Bulls.

Nice.

I was like, I'm like, wait, wait, are you talking about like when someone Googles something, you, you want to be number one for, for these terms? And she's like, yeah. And I'm like, in my mind, like, I, I literally went into one of those like moments where you like, like it feels like the machine just freezes, you know? And it's like there's no, the button's not working. And I was just like, I just kept staring and I was like, I, I don't even know how to approach this. I'm like, this is like, you know, but—

It's so difficult.

It gives you this kind of empathy because then you realize this isn't everybody's bag, right? This isn't everyone's game. And they, you know, as much as like they're in their world, like selling, buying and selling homes, like there's this other part of it that they need. But, you know, it's almost like when, you know, currently, like in, in just an ecosystem quality where it's like Now that, you know, AI has come up and like agents and all that, it's like there's a, there's a whole nother world like adjacent that's gonna come crashing together.

Yeah.

That if you, you have to like, your expectations and like what you think it is is very, very off at the very, very beginning, right? And it's like working into that system and trying for it to make sense. Like for that individual, like for that agent, like after having the conversations and stuff, it became, she almost was like, I am so like, I had no idea like this was like, yeah, of course I can't rank for Chicago Bulls. Um, you know, but it was like that transition to that is, it's like hitting a brick wall at first, you know? So it's, it is, it does get a little crazy at first, especially as things change, right? As rapidly as they are. So no, but there's a, there's a, I have so much respect for people that like continuously, like in the SEO world that like you, Victor, that do it day to day now where it's like mine, I don't necessarily touch it as often as I'd like to anymore, but. In this world at this speed right now, it is something just remarkable to try to keep up with. So kudos to that for sure.

Wow.

But let me get, um, let me jump right into some questions here because there were some things that I really wanted to, you know, ask you and kind of get your take on. And like one of the very first ones, and I know I, I had shared some, uh, ideas with you, you know, previously, I just reordered a couple of them. But like for, and, and, you know, coincidentally too, it was really, um, I just saw this morning that I think, uh, Dharmesh, the CEO of HubSpot here, just posted a pretty interesting thing as well where he kind of called it— wasn't SEO, it was— he changed the acronym a little bit because now it's more like agent optimized versus who you're optimizing for has significantly changed in certain aspects, right? So he has a really good, if you haven't seen it just yet on LinkedIn, Um, he has a really, really good post about that as well. But how do you, Victor, kind of, uh, differentiate between like traditional SEO and then what we can call, you know, at this point until somebody comes up with a cooler name, I guess, like AI SEO?

Uh, yeah, I differentiate it as, uh, you know, the human need is the same, right? Like you have a problem, you're looking for a solution. The tool is shifting in usage, right? Maybe you used to be an avid Google searcher and now you're starting to lean on these, uh, LLMs or agents, whatever that progression might be, more. And the— there is a lot of overlap from a technical standpoint, right? For search engine optimization, if you're doing technical SEO, you kind of learn about informational retrieval, you learn about how search engines are built, how they have to find documents on a web, store them, right? And then like when people are retrieving that, how that's spat out from the index through retrieval. And then you think about all that that's being— that's happening in Google Search, right? At least that's what I'll mainly be talking about. And then you think about these LLMs and how they also have to find all this stuff on the web, right? How they have to store all that, how they— when you send in a prompt to an LLM, would also have to retrieve information. So a lot of the technologies there are actually overlapping. And as, you know, website owners or business owners that create content on the web, you have to deal with both sides right now. It's not like a straight out, this is what bots get, this is what search engines get. Because we don't have a final say yet as whether Google will be able to, you know, win over this to like OpenAI or to like Anthropic or Mistral, like anywhere around the world, or I don't know, DeepSeek, right? Like there's competition everywhere.

Mm-hmm.

But that underlying technology is similar and the human need is also very similar. So the nuance in between there is what we kind of try to break apart every day in our day-to-day as things are rapidly evolving. But going back to like, how do I stay sane when it feels like everything's always like insanely fastly evolving? It's the human behaviors, things aren't really changing. And the technical foundations of how that data is still manipulated, but in a different way of manipulation, that is actually still the same thing. Right? And so you can build off of these foundations that you do know to be true, and then you can kind of like poke at it, right? In the olden days, uh, that old lady with that CSV file with a bunch of words and commas, right? You would put then the, uh, meta keywords field, right? That's like SEO in the early days, and just throw as much of that you can, right? And, uh, you might even add those keywords in like white text and white background. Keyword density checks, right?

Right. Font size 1. The date.

Font size 1, like it's tiny. Put it on the very bottom of the page. So, and then the joke is now like, those are things that people are adding into like their research papers or something and trying to catch like reviewers off guard or like you're sending in your resume, right? And then people are like adding all these like old school SEO hacks where search algorithms have penalties assigned when they're able to catch these, uh, gamification kind of, uh, tactics, right? That we would as SEOs classify as black hat SEO, which is things against the terms of services of a search engine and their guidelines, versus what's considered white hat, which is, oh, uh, makes things that humans want and then make it very accessible to search engine crawlers or bots to understand that content. Right now, that stuff doesn't really exist for these LLM crawlers. You're really seeing a revival of a lot of these black hat tactics coming out because you're like, oh, I can trick an LLM into saying that I am this person. Whereas it's just not grounded in truth, right? Whereas in a search engine, they went through a lot in being able to verify facts. And so sometimes they also have to balance out perspectives as well, right? And so those are all fun things where you'll see some parallels in search and some differences in AI results. Right.

Have you seen, like, I, I have you having done this for so long, are you seeing, uh, like exact same trends or are people getting smarter with this new iteration? Like SEO 15 years ago, you did X. Are people doing a similar thing now? Like just trying to repurpose that for the AI era or are they augmenting it a bit? I know it's a very general question, maybe not specific enough.

I would say all the really experienced folks are just having a field day going like, I can't believe this now works when I try to feed this to an LLM. There's a lot of private discussions going around. There's also just discussions of folks going like, oh, this doesn't work anymore. I guess they finally figured it out. Yeah. To note, it's not a matter of if you should do it. It's a matter of like, when will they figure out like, that's something people are trying to game, right?

Yeah.

So, it's kind of like, if you work in like web security, right, there's always going to be some exploit. And there's always going to be some patch. It's just a matter of when that'll happen. And similar things are basically happening right now with agentic search, if you want to call it that, where if you're able to feed or poison the data that they're getting via these, you know, keyword or data stuffing methods, then, uh, you can bias those results and answers.

Right, right. I'm thinking like specifically about like, was it Cloudflare that recently told like, like the scraping for AI is like, no, no, no, you know, you have to pay for this? Is that like the middleman saying kind of like catching on quicker than they did before to kind of, uh, govern this?

So that's a problem, right? Uh, so some context for the viewers here is these LLM crawlers allegedly are, uh, you know, using open source information on the web, but there is no, uh, way where, uh, the normal sites are kind of compensated when their content is accessed, right? They're pretty much cutting off a lot of these websites that rely on traffic, maybe for ad revenue, or maybe they rely on traffic to understand whether conversions happen at someplace. And because it's so new, that's a lot of attribution is kind of breaking, a lot of credits or revenue streams are breaking, especially for news publishers. And so that's why that's a huge concern for parts of the web. And Cloudflare's proposal is, well, then let's make it so that we can charge these AI crawlers whenever they visit and leave that as an option. So the website or the World Wide Web's whole ecosystem, which is often fueled by ads, doesn't break, right?

Right.

Like, we need to publish, we need to have human content on the web. Otherwise, the web will all just be AI-generated content and nothing new or novel is going to come out of it anymore. They wanted to make it so that there's a monetary incentive for humans to continue to make good things to feed these models.

I was going to say, that brings up an interesting question and I don't know how convoluted this question is going to be, but do you see it getting to a point where people are now using AI to help create meta descriptions? to help populate stuff. Do you foresee a world where AI basically is training itself to not perform as well with SEO? Because sometimes, like, sometimes AI can make mistakes, sometimes it can make things up, you know? Like, sometimes, like, the em dash is like a huge no. Reading a document, if there's like 15 em dashes, somebody definitely used AI to generate it. Like, nobody ever uses an em dash. Sorry, em dashes. But It's just one of those things where like you see a lot more of this context coming up, whether it be in web copy, whether it be in meta descriptions and things like that. The people that are using this AI to create those things, do you think that the AI that is searching that, it's basically like reading its own novel at that point that it's making and then making decisions based off of that? Do you think we're going to get to a point where like a lot of this stuff is just I guess basically like whose line is it? Like it's all made up, the points mean nothing.

Like it's almost like Dead Internet Theory where there is no— everything being made right now is all just—

Exactly. Like, I mean, do you foresee some people going to the extreme with AI to try and optimize their site, but in reality killing their own site at the end of the day because they've not done as much human oversight?

Uh, so For something to show up, right? If you think about it from a very human way, it needs to be remarkable enough or deserving to show up, right?

Mm-hmm.

And then if you're saying like, hey, uh, all of these businesses are jumping onto the AI train and then they're just leaving to the AI to make them rank number 1 for searching for Chicago, but you really don't have the foundations for people to want to prefer seeing your business when they're searching for Chicago, it's just not sustainable. Like everyone else has the same thing, right?

Mm-hmm.

And so one of my theories is how do we avoid that catastrophic feature— future where everyone has the same tools, everyone's throwing out the same thing and nobody's actually accomplishing anything, but they feel like they're doing something well because These tools are telling them they are. Well, you still have data that says whether or not these things are working well or not. And if it's not working well, then, you know, people will claim that it does X, but the results will speak for themselves. The other thing here is we're in a transition period. So we are still the humans in the loop, right? We are the ones catching those hallucinations and errors you're seeing from AI-generated things. You have a level of trust you have with the business entities that do choose to use, uh, AI generation either as, you know, fully or assisted or some sort of mixture or none at all. It's kind of like that, uh, you know, is this an organic label or like a, you know, you've got grades of things you look for in content now. Um, it's, it's kind of like, what, what was it, Velvet Sundown or something, that AI-generated like, uh, Spotify band? And it's context for, again, people that are viewing this, like, you can look it up on Spotify. It's like, it's got like a million or something streams at this point.

Mm-hmm.

People are like, oh yeah, I don't mind this AI artist. Um, I was like, They'll listen to that. It's so— some people will be fine with it, some people won't be. And ultimately, as long as the people consuming all this content isn't more bots and it's actual human beings, then I'm okay with like this transition. Because imagine if you're creating all this AI-generated stuff and then it's like only going to AI bots on the web, and then at the very end, like no humans ever consume any of this. Then it's like a very weird future we'd be in, right? Like where it's like we're not— we're no longer optimizing for any human beings at all. It's like we only care about what the bots think. I think that would be the true dystopian that you're describing.

We try to please our overlords. That's why I always try to start every episode with when the AI takes over and they go through and they comb through these old episodes. I am completely on board. Thank you for taking over and I appreciate everything that you're doing, AI. Thank you. But I wanted to pull a little bit on a thread that you were talking about a little while back. That was the cutting out, the Cloudflare being worried about AI cutting things out. What are— and I'll use Google as the example. And again, sorry, Bing, as you say, and DuckDuckGo. But what are some of the more significant changes that you've observed with respect to the SERPs and the algorithm? When I say SERPs, for those listening, the search engine result pages. What are some of the significant changes that you're seeing thus far? Again, things are progressing and there's something else I'll bring up a little bit later that as far as search browsers and stuff is that might change things as well. But currently right now, if we're just talking about, let's say, a Google search result, what are some of the influential changes that you're already seeing that we can attribute to AI with The results themselves?

Yeah, so that's a great question. Um, so search engine results pages have changed a lot, uh, and it's kind of ironic, uh, because Google Search, at least the home page, used to be a single search bar, right? Now you've got some Discover, or if you're on your phone, it's got some of— if you're still logged in, it has suggestions for you based off of all these other things you looked at. That's some machine learning behaviors there on suggestions. So already that homepage has some AI in it, but you probably don't know about it, right? Now let's go into the actual search. So the Discover stuff is fascinating because Google likes to call it, you know, we're searching for you before you even know you need to search for that, right? Based off your history. Now the next part that's interesting is that's having a huge impact is, for me, definitely AI overviews. So this is when you get this little box that pulls together results for what you're searching for and creates a summary of it and formats it in some way and then lists a bunch of citations of where it got all those sources from so that you Unfortunately, what we're seeing is people aren't really clicking on where they're citing to as much as if you had ranked number 1 for it. And sometimes those citations, because they're jumbled together between different sites, uh, are inaccurate or misattribute things or associate your, you know, answer from your site the wrong way, right? So there's some—

I was seeing numbers like, is it— and is this— I don't know, like, tell me if I'm off on this, but like upwards of like 35% of like organic view or organic like clicks are getting like lost because of the, um, I call it rank zero. Is it still considered rank zero? Like, I don't know, for people listening and watching, like when like universal search like first became a thing, right? Like you would have these like things that showed up above position 1 and they would call it rank zero. Do they still even do that, call it that at this point?

Uh, I think As that is one, um, description of it, uh, I would say position zero was more of a vernacular inside SEO circles where it's more about a different search feature called featured snippets, which is, uh, a summary pulled from a single website. Um, and back then a lot of folks were already kind of like either upset about it or very excited about it. depending on where you were already ranking, because some folks, while they were searching on Google, would stop going to your website because that summary was good enough for them to not have to visit your site, right?

Mm-hmm.

And you can almost see AI Overviews as kind of an evolution of that featured snippet in that it's taking a similar position. It's leading to you know, a little bit less clicks than you would expect if you are in position 1 without a featured snippet showing up in that. And, uh, that's why people like to call it position 0, because it's not the traditional blue link that you see that is on the top of the page. It is a new box that is on top that's taking over people's attention, and it's a separate algorithm within Google Search, and there's separate ways to optimize for. And that is the same for AI overviews and for featured snippets.

Have you messed around with the— oh, Nick, I apologize. You continue on, my friend.

No, no, go ahead. Dennis always has good questions and he stops me from rambling, so I always like— go ahead.

Okay, yeah, I, I am— we're talking about what I feel like it might be antiquated at some point in time soon, like these, uh, these search results pages. I like— I'm starting using the Dia browser lately, and you know, you don't even go to a search page necessarily. You type in a question in the address bar, you know, the most clicked place And then it just goes straight to an AI response. Have you messed around with that? And do you feel like this is the future of interacting with the web?

So I also have a theory here where, uh, the intents in Google Search have been broken down into a lot of different content types that are good fits to answer a person's needs. So, uh, Nick was just talking about like universal search. Well, what is that? Well, Google organizes the world's information, and that's not just like web pages and the text that's in there. It's also the images, the videos, uh, the PDFs, and what have you, right? Uh, and then it also does finer, uh, segmentations there, like your flight information, hotels, the news, scholarly reports, and all that. And so what you're seeing, instead of that showing up on a single search engine's results page with all those put together, almost like glued together on a board, and then you're like, hey, this is the page 1, and then there's all these different kinds of mediums to answer your question, be it an AI answer or a highlight from Reddit, which is now happening more often, or a YouTube video that is still owned by Google, which they like to now do that a lot more than they used to, right? Instead of that experience where you have to go through all that, the agent-led one is kind of more like I asked the question in a very natural way, and then it goes through a list and synthesizes all those different sources that I just talked about. But in kind of a more curated format where the thing is one answer versus multiple answers pieced together like a magazine cutout kind of situation, right? And so I think that's the new user experience that a lot of people prefer because gosh, you have to click a lot in a traditional search, right? You have to really filter through all those things to get to what you want.

For sure.

Whereas with an agent, you're like, oh, that's roughly wrong. It's, oh, was I not specific enough? Let me give you more details, right? And then you can watch that all just happen versus you going on and thinking like, okay, what do I need to type out or research or go into these sites? And hopefully I have the right sense of solving the problem I'm trying to solve in that site. Maybe that one site can solve it all, or maybe I need a combination of research, right?

Mm-hmm.

That's where you kind of also talk about deep research from these agents where it's pulling from hundreds of websites. Just to synthesize an answer for you that is more comprehensive and spares you from all the clicking around and thought process.

Actually, OpenAI, one of the notes that I put for myself to bring up what we're talking about is OpenAI is launching their own browser as well, which people are saying Google just woke up to a nightmare. The headline at Google just woke up to a nightmare. So like those type of things, do you see like a shift in people using those more traditionally than you would like a Google like search result? Because like I even see Google trying to do things as well where it's like they're trying to bring in more. And I don't know if this is like to combat like what could be eventually like the new thing is to use like a, you know, an agent browser versus a regular browser. that's how they're even going to classify them. But I've seen like Google's now indexing, I believe, Instagram, like Instagram posts, and they're pulling in all of these other sources. It almost seems like if you have an agent browser like that, that's kind of like bringing together all of these different things to create one answer. Is it almost like, I don't know, is Google able to combat with that unless they do their own type of agent? Browser? Is that like where the future is going?

I mean, I want to extend this a little bit to Victor. Like, like, for example, Diav, once again, that I've been using lately, actually using your, uh, your own— anything you're logged into, it's logged into. So it has access to all those things, which, you know, further curates that feed for you. I, I— it's so powerful. I can't imagine Google not having to reply in some sense to something like that.

It's almost like that MCP type of functionality within itself, right? Where it's got connections to other agents and other tools. That's— yeah.

Yeah.

That seems like it's going to be the way it shifts then. Would it— would that be like a bad assumption to think that that's going to be the next? Because, you know, you have like iterations, right? You have— okay, you go from Yahoo, uh, link directories, right, into like what would then be considered like more of a traditional search. And then does this feel like Or will this be just the next iteration that humans move to?

I think it's a matter of trust at this point. And I think the feature y'all are really talking about, yes, you're talking about NCP and all that, but these agents can have this feature that I want to say OpenAI coined called memory. Right, where your— the past context influences your future output.

Mm-hmm.

And that does exist to a much, uh, a limited extent in Google Search, right? But it doesn't feel as magical. Um, so if you have, let's say, like your Gmail connected, sometimes you— there are some searches that can pull on like, you know, when your flight is next, right? So some of that is there. And so if you were to ask me like, hey, is this where things are moving? It's like, yeah, it's been moving that way for quite a while actually. This is technically not new to me. It's just a matter of like how pervasive will it be? And then why has it not become more pervasive? And it's got to do with privacy. It's got to do with advertising and how do you deal with advertising in these situations?

Mm-hmm.

It's got to do with, um, uh, you know, whether or not people really want this because privacy is, you know, really important for some. For others, we just kind of accept every cookie we get on any website not knowing what it really means, and we click on every legal agreement. I mean, we all do that, right? Like, just— I don't know what my new Apple update is. I just click it and I just go on, right? Like, it's It's, uh, I'm okay, right? I didn't sign away my life and everything about myself. Nobody can create a bot of myself and replicate me, right? I'm all good. No conspiracies here, right?

Yeah, let me, let me switch personas for, um, well, not, not switch personas, but kind of shift it just a little bit. We've been talking about more of the traditional user with a lot of this, whether it be like, um, the agentic searches or just like your traditional Google search. But let's kind of move kind of into the realm real quick of just content creation itself, right? Because there's a whole different— I don't want to say a whole different, but there is a different type of person here that is also going to be affected by these kind of changes. I'm just curious, like, how do you see like the content creation strategies, uh, you know, kind of change as far, you know, as far as these things kind of iterate? Like, where do you see content creators having to adjust in this world?

So I think you need perspective, right? So in an AI answer of any sort, usually it comes with some opinion. It doesn't And it sometimes will try to balance that opinion as well. And so when your opinion is unique and grounded by some sort of fact that is true that only you as the content creator has, then that becomes a lot more authentic and citable, right? And I think there's a lot of talk in our space right now about brands being kind of the way to cut through all this AI-generated noise where having taste, having curation, having a brand that speaks to something that stays consistent, that promises things and delivers. You're now creating content not just for keyword sake or for what people are searching for sake. You're doing that in the context of what your brand stands for or what are the values you stand for or some of the expertise that you have That you can shine on in kind of a very hyperlocal or hyper-specific example that you can provide.

Mm-hmm.

And that is not like AI slop in any way. You can use AI to assist that process. And I think that's the key thing here. It's, you're not going against AI here, you're working with AI at this point because we're all being asked to do more with less. And really the most valuable thing you have as a human right now is this intuition of things that matter to the people you serve. That's through all the interactions you have with your coworkers, with your customers, whatever it might be that AI does not have the data ingestion yet for all these interactions yet.

Would this be safe to assume too then that this is going to be the point in time where structured data or like data tagging, if you will. Like, I, I usually just sloppily say that it's just tagging data, but like, you know, structured data, you know, semantic data. Is this gonna be like one of those pivotal moments where those that either didn't do it or, you know, like, because I, I'll admit, like, and again, this was a few years back, but like even in the traditional like search results world, right, there was, there was tagging you could do. There was like the way you could structure your data with, um, God, what was that? Wasn't, um, I, I'm blanking on, uh, JSON-LD. No, there was the, um, it was, it was the structured, uh, semantic— it was, um, the markup. But like, there's like structured markup, like for example, like if you could tag a recipe a certain way and it would show up in the search results as a recipe. I think there was even like a really, really, really quick foray into like authors of things where your little picture could show up and then people—

Yeah, it was like the schemas. The schemas. Yeah.

But they abused it like really quick, so they got rid of it. But like, it feels like right now— and correct me if I'm wrong— but like, to me, like, this is going to be one of those pivotal moments where those that didn't do it or didn't like take it as seriously, it's going to be so much more important now because of the way things are going and the way things are kind of shifting. Because that's— I don't know, is there any other real way? I mean, I know it'll scan data, but like telling it— like you said, there's, there's tactics and things that you could use previously where like Traditional search would be like, yeah, we've caught that. Nice try. Thanks, buddy. Where it's like, it's not happening yet. So the way the data is being structured right now seems very, very important in these stages right now. Is that a fair assessment?

I would say if you didn't do it in the past, you're screwed. I'm kidding. No, you're fine. It's too late. It's all over.

Shut it down. We're done.

These machines right now, they have knowledge cutoff dates, right? And so you can't ask, or you can some recent things where now they also have web search as a tool. But again, it's not like you're— it's the end of the world if you don't structure your data. The difference here is right now most of them Except for the traditionally search engine crawlers. They're visiting your site, they're pulling the HTML, they're not rendering any of the JavaScript, right? And that's what they see. So there is a subset of inaccuracies in your web content if a lot of it's all like JavaScript rendered. And you're like, okay, am I screwed if I do that? Well, the good news is a lot of the web likes to steal your own content. too, right? Or they'll repurpose it or talk about it. So going back to the whole discussion about like, oh, Cloudflare has got this solution. Well, they can chart— like, if I own this site and I implement that, one loophole that these LLMs might have is someone else might have created a carbon copy of my site, and that site doesn't care. And they're just like, you know what, I'll play arbitrage here. I'll charge you 1 cent less. for that same exact content. Like, you can see these loopholes coming through, right? Like, it's—

Yeah.

So same thing here with like this data stuff. You're going like, oh, I didn't do this, I missed the train. Well, like, how do I make up for it? It's like, well, maybe I'm not in a bad position. Other people have already talked about some of the things I should have included, but they did that for me. Uh, and this brings up other issues, which is they did it for me, but they did it wrong. And, uh, they are saying that my product costs $2,000 more than it actually does. And these LLMs are therefore not suggesting me because they believe we're an exp— uh, more expensive product than we actually are, right? Those are the new challenges we have.

That's a good point.

And you can do that by trying to fix your own stack today, right? And kind of reach out to the people like, hey, we've done an update. And then there's like, you know, the marketing way where you like do a big announcement, everyone realizes, oh, I've been putting out really old outdated stuff that's not helpful, and they'll update their things that way. And then there's, you know, kind of the more nefarious methods that I've kind of been suggesting, where it's like, hey, uh, there are a lot of places on the web you can publish content, right? You can use those as avenues to influence, uh, these LLMs on a different answer because it's a It's more of a consensus machine where if you're able to feed it the same answer in multiple places, it's more likely to suggest that as a next token predictor, that as the answer, than if you were like, oh, I only have this most up-to-date information on our website, which by the way is outside of your knowledge cutoff date. So I really hope you use the web search tool so that people get the right answer. Are you going to leave that to chance if you're an SEO? No, you're going to try to fix that. You're going to realize that that is a problem and limitation and you're going to look at the technical foundation of the things and go like, what are some solutions we can implement today?

Excellent. Those are really good points. Yeah, I could see people taking advantage of that very quickly. Now with respect to For me, with a lot of this, especially when we were talking about earlier with the reduction in clicks, right? Or the reduction in visits and the reduction in what you could like kind of summarize as metrics. If I'm one of these people that has to create like, I remember back in the day, right? Like, you know, some of the very more basic, you know, reports, especially if you were at a company that, you know, was just investing in search optimization or Oh, and just to give a shout out, I wanted to make sure, uh, the thing Dharmesh was talking about, he referred to it as search agent optimization, so SAO instead of SEO. So I thought that was pretty interesting. But, um, but in the scenario I'm bringing up, like, you know, you work for a company that's trying to invest in SEO, and a lot of the reports that you would run at the— I'm talking like years ago, because now it almost seems like there's a lot of arguments can be made that that's not really a good indicator of quality or success. But there's the traffic reports, there's the this and that, right? And I'm not talking about what the backup is on the Dan Ryan in the city. I'm talking about actual things that were, if these numbers are going down, people that don't understand the reasoning behind it and the shift start to sweat. Do you know what I mean? So I guess in an AI world like this or in an AI-dominated search landscape? Like, how do we— how are we going to measure SEO success? Or how do we shift the way we're measuring SEO success?

Uh, that's a booming industry right now, right, where people are offering AI measurement tools. Um, and currently the approach is to track various prompts that you would be able to put into these solutions and then measure either through visibility or citations of your website. And then you could also retag certain sites that then show up that you, if you're in a larger company, might consider either owned or earned media, which is a very agency way to say like, can you influence you know, those other sites too. And then you're like, you're actually part of, you know, what we kind of control on the narrative on the web. So what has changed? Well, currently there is no volume metric to any of these tools. And so there's a lot of estimates being thrown everywhere. And because the underlying technology between kind of, uh, a user's query and the content that comes out is the same, which is embeddings. So a search engine not only matches the words that you type in, it also matches to like the rough meaning of a web page and that question you throw in. Same thing with LLMs, right? And so, uh, those are similar Uh, in the way kind of we talk about topic clusters in SEO, where a topic cluster is, hey, let's go on a very broad topic that people that are interested in what I do may search about. Let's look at kind of that journey while they investigate this topic. May I, as a business or a brand or whatever, may offer things that are tangential or related or help solve some sort of problem they have in that topic. And then what are the related searches they do and where do I fit in that picture? What are the terms I should go after? What are the content I could go after that would fill that gap? And then where are the points where it makes sense for me to capture that traffic to some sort of mechanism as a business I can capture, whether it's leads, whether it's a phone call, a form filled out, an ebook they capture and leave an email, and then I nurture with emails and all that, right? All that stuff you're used to on the web and all that, now you're condensing that and then you're going like, oh, and then we have an equivalent for these LLMs and what we're tracking. We don't have the volume metric anymore, but the behavior is the same and they're staying on the GPT or the LLM or the chat experience. We have no analytics on that part as well. We're relying on, again, that visibility and that citation. We're relying on the answer being relatively positive because we can track some of the inputs that go into these answers. And then we're hoping that the end result is they go to our homepage, they call us or some way. And right now the best method to do that sort of attribution is like, how'd you hear about us? Right? And it's like, oh, did you use like, you know, an LLM or ChatGPT to find us throughout the way? Right? Like that's our best hack right now as we are lacking those volume or analytics in those 2 places. And that's kind of how you're drawing that correlation to that lost traffic that you just had all in search. It's not that, you know, the traffic is gone and people don't need to buy anything anymore. It's just that you're not seeing it because your analytics doesn't cover it. And the intent is still there. The behavior has changed, but the tracking is not there yet. Gotcha. Okay.

Yeah, because that's, that's gotta, that's gotta like play havoc on those whose like main responsibilities are to measure these things and are to bring those reports and stuff. Because I know like depending on the, um, how, uh, not in-depth, that's probably not the right word, but like depending on the place you work for, how invested they are in what's going on with all of this, with results, with search, and like How, you know, AI is playing a role in this. It can be probably a big headache trying to have to explain some of this stuff when it's like, yeah, no, we just want to sell XYZ widgets and we, you know, 90,000 people come to the site a month. Now you're telling me, you know, only 60,000 come, you know, fix it. And it's like, but you're not, you know, so it's, yeah, I could see where this becomes, and I mean, again, like with everything, time will heal all wounds, but like. Right now, I, I do empathize with like people that are kind of put in that position right now.

Yeah.

Speaking of with time, like where do you, and I love, and especially with AI with this question, because like it changes daily. So like depending on the timeframe I give, it could be like 1,000 different answers all at once. But like in the next year or 2, let's say that, cuz then that gives you what, 365 answers times 2 almost. Um, where do you, where do you see this go? Like where do you see it going? Like where do you see AI's impact on SEO? going from this point on in the next year or so.

And we will call you on this later on.

Yeah, exactly. I mean, like, we're gonna, we're gonna take this clip and then we're gonna say, look everyone, Victor was wrong, you know. And but we're gonna hold it, we're gonna hold it, right? We're gonna hold it and we're just gonna wait for the right moment.

Set a calendar invite for a year from now.

Yeah, it's just gonna be random messages from us. Hey, remember, remember the thing you said, right? Just like video clips all day long. But But no, I mean, like, as a best estimate, because like I said, I mean, next week we could be talking about a whole different thing that's happened or a whole different way of doing something. You know, like, look at how fast MCP servers just, you know, hit the ground and completely changed the way people were talking about things. And that was in a matter of like a week or so, right? So the best you can, and we won't hold you to it, I promise.

Uh, like next 2 years, it's kind of like already happening. And there's just so much more interest in AI-generated content and workflows, right? How to do it right. You're gonna see a lot of job descriptions kind of change and titles and names change a lot in the next 2 years. It's gonna be assumed that you need some sort of AI knowledge for any kind of search job going in the next 2 years. And that's why there's a lot of worry and concern because, yeah, some folks don't really update their knowledge on, uh, what's changing, right? So you're just not going to be able to pass off and say like, I'm an SEO. It's like, well, do you know any of this AI stuff? It's like, nope. It's like, oh, good luck. Um, and yeah, and definitely don't ask an, uh, LLM going like, how, how do I pass this interview on this? Because it's not going to have up-to-date information on that. Uh, so yeah, uh, you can— so again, you're going to see a lot more, uh, collaboration between SEOs and like content generation and how to find that balance. You're gonna see job descriptions change. You're going to, you're gonna see renewed interest on kind of the more technical side of SEOs, which is log file analysis, understanding, you know, we like to say what you eat is what you are, right? That's a really common Idiom. Well, for bots, it's like what they're gonna ingest is what they are gonna believe your site is about, right? And so looking at those crawl logs, seeing if you're you know technically sound or if there are issues, and you definitely want to fix those so that the bot has a good representation of what your business entity is and is actually accurate, right? There's going to be renewed interest there. And yeah, I think you're going to expect all these. traditional third-party tools, be it Ahrefs, Semrush, and all that. They're all just gonna throw a bunch of AI features out there to solve that booming industry of how do we measure the impact now that traffic is down and there's so much interest in this growth in AI. They're gonna fill that out with a bunch of features.

How does one— oh, go ahead, AJ, sorry.

No, I was gonna say I'm gonna call it because you mentioned a good point about the technical SEO, especially when it comes to implementation of the technical SEO as well. Like, that's where I can see that I'm like, most people are probably gonna be like, oh, I can just do this, right? And just make some changes. And you could really cripple a site if you do some wrong stuff. So it's one of those things where like, that's where I can, in my head, that's where I see like the intersection of before, like people will be like, I don't wanna go talk to the developers. They're just like sitting in the dark basement. They're not gonna listen to me. And it's like, one of these things where like if you're coming to them with technical things, like I'm wondering what the working relationship, if it's going to cause more stress or if it's going to mend that working relationship that happens. Like, you know, we always have that developers meet marketers, fights always happen. Like, do you foresee like development teams coming closer together and being more cross-organizational working knowledge-wise? Or do you still see this separate silos of work where The SEO people just stick to SEO, don't really do web-based knowledge and the web-based people know that. So you give them your plans, then they figure out how they're gonna implement it. Or do you see more triads working together to fix these issues?

Yeah, I think SEOs become the technical people if they're in a marketing org and then they become a creative person if they're in a product org, right? And so let's talk like crawlers visiting, right? Like if you're doing a CDN and you have like— you have bot detection and you want to prevent denial of service attacks, right? And your CDN does that for you. And somehow it accidentally blocks these LLM bots, what's gonna happen to your visibility, right? And you're gonna have to have that discussion with your developers, or you're gonna need to have that relationship with your developers in the first place to even know that is happening, right? And so I see that getting closer in that regards because now it's not just about marketing does marketing things and then asks for dev to make it look some way. Now it's like, no, you have to be very particular. Like, Hey, did you know that these are the listed IP ranges from these LLMs? Like, can we put an allowlist or whitelist on these IPs so that we always show up? Can we not serve like JavaScript bot challenges because we know they're bots anyways, right? It's stuff like that. And then they also understand why that's important. And that's if we don't even show up for these bots, we don't show up for these LLM-generated answers. Uh, and that's where the hallucination kicks back in.

Mm-hmm.

If these web search tools from these LLMs can't access your site, it's not gonna say it can't reach the site. Half the times they'll actually just hallucinate and go like, oh yeah, I visited a site and, uh, this is what it says. And it's actually, you know, an excerpt from all these other sites. And then like, if you really dig, dig into that, you realize like, wow, like sure, it'll try to defer to your site, but when it can't, that's when a lot of that funny information, misinformation happens.

Right.

I highly recommend for any developer just to go on to their LLMs, go look at the network requests of what pops in, and you'll see all sorts of funny things. They'll search for queries, they'll tell you the titles and descriptions and snippets that they're pulling for, for that RAG, and then to summarize and generate things. It's all there. For a marketer, That's like, oh, that's like developer stuff. When SEO is like, oh, this is our stuff. And then, you know, like developers like, oh, that's like kind of marketing but also kind of technical. Like there's like, so SEOs are in that funny middle ground right now. And, uh, hopefully we'll be building a lot more bridges for, uh, both teams.

Yeah, no, no, yeah, that's cool.

Now you talked about, uh, I want to ask you, what are some of the, um, like staying, like, because you said like not having that knowledge then puts you At a certain point, there's not going to be able to be an SEO that says that they don't know AI tooling. Is there basically any resources that you recommend for people to start to try to jump into that if they haven't already?

That is a great question.

I'm lazy, so I would I think you mentioned there was one, isn't there like a, there was a newsletter I think you mentioned to me once. There was a, what is it, SEO FOMO, right?

SEO FOMO is from Aleyda, yeah. But, you know, if you're starting out from like scratch, scratch, maybe also from Aleyda Solis, Awesome SEO, you would probably go with learningseo.io. I honestly think you should start with actual problems first instead of just focusing on education. Right now you are just, you know, 1 or 2 experiments away from being an expert in this stuff. Like, actually get your hands dirty. Like, actually go type in some prompts, look at what the answers are, see if you can change it, see if you can publish pages and have funny things come out. Like, if you are actively doing these things and seeing how these things change, You're already ahead of most folks that aren't trying these things. You can read up all you want, but that's all gonna be old stuff. And there's nothing better than—

Like an hour, right?

So yeah, yeah. And that's why there's no, like, I would not recommend signing up for any courses, right, on like, this is how you do this stuff. I would say like, if I were to really recommend folks, on like what should you do, uh, you should focus on the fundamental documents on like crawling, indexing, and ranking. Like that stuff is still relevant for, uh, AI answers. And I would say like any human behavior related things in marketing, uh, you can chase the algorithm, but that will just help you beat your competitors.

Mm-hmm.

But if you chase after how human behaviors will evolve, That is what the algorithms are trying to address, right? So if you wanna be ahead of the algorithm, you have to really understand human behavior. Now, the complexity here really is how is this new technology changing human behavior? And if you're able to answer that better and create a good experience there, that I believe is kind of the ultimate question we all need to answer to really win and be ahead of the algorithms and also be a good business, right? How do you actually show up in those instances? So that's the most white-hot answer I can give you as to like how to stay ahead is study your customers and anticipate. Again, going back to the dumb way right now is I ask them, how'd you find out about us? Did you have any difficulties? And then just start by being less wrong, right? So instead of focusing on this is the way to do it, Patch out all the points of friction that they're experiencing in that journey and then figuring out the root cause behind them inside those LLMs or outside in Google Search with AI overviews, wherever it may be. And you'll just be in a better position and you'll have much more relevant experience as well.

Excellent. Thank you.

Great.

We are running up on time, so I wanted to— is there anything you wanted to give a shout out to, Victor, or anything like that?

No, I'm good. Everyone just stay, you know, stay happy, healthy, be kind to the folks around you. You know, there's a lot of change in the world right now. And yeah, I think as long as you're, you know, grounded in your core beliefs and the people who care about and you're able to take care of them, I think we're all very fortunate to be able to to do so and just spread that energy around. And I think that's ultimately what I look forward to. Even with all this AI change happening, I am a techno optimist, if you will.

Love it. No, thank you. Yeah, we appreciate you coming on and talking because, I mean, like I said, this is one of my favorite topics and I think it's like, obviously it's very timely, of course, and probably will be timely for the next, who knows, months, if not years coming up. So No, we really do appreciate you coming on. I wanted to thank you for taking the time out. And for all you rock stars out there that were watching and listening, again, don't forget, please hit like, tell everybody, share us with all of your friends and family and strangers. Just walk up to strangers in the street and just start telling them about us.

Please, please.

It never hurts. And yeah, so, you know, thank you from on behalf of all of us here. Thank you very much for watching and listening. And I'll leave you with this. I mean, this is why it's very, very important to start tagging your data and to start making sure that the data that you're creating is relevant and good and pure because it only took Ultron 5 minutes to get put on the internet to find out that humanity was the problem and then tried to eliminate it. So really get out there and start tagging that data.

Dang, it's true.

Dang, look at you, look at you with the Marvel references. I love it. I mean, it always works, you know what I mean? There's certain things that you can make references with that just always work. This is one of them. Love it. All right, y'all, see you very later.