Episode Transcript
[00:00:01] Speaker A: Hello, I'm Jeremy Rivera, host of the Unscripted SEO Podcast. I'm here with Michael Buckbee. Why don't you give yourself a quick introduction and then we'll get started?
[00:00:10] Speaker B: Sure. Well, first of all, thanks so much for having me. This is awesome. I really appreciate it.
I think I do actually have a title of, like, you know, co founder of Noah Toa, and we do, you know, work with AI search, trying to, you know, track ranks in the different AI search stuff. But I think, really, I've spoken to a ton of SEOs in the last year, and I think I really feel more like an SEO therapist where it's just kind of weird that, you know, as someone who's a little bit of an outsider to agency SEO and we can talk about anything, but I think I have a much more positive outlook on, like, what SEO looks like and what's happening with it than maybe the industry as a whole.
[00:00:50] Speaker A: I have to admit, my last conversation with both Garrett and Morty Oberstein had kind of vaguely funereal tones at certain points in the conversation or just kind of move along to get along, like, who knows? But I actually. I am an optimist at heart, and I keep pressing on, and I see these things and I see these changes.
I am behind the times when it comes to adapting to all of the new changes in AI SEO.
I come from a product management background, working with Raven tools on rank tracking. So I'm curious to see what your approach has been, how you're accessing that data. Is it. Is. Are these AI SEO tools consistently delivering brand mentions in a way that's really actually trackable? I'm curious, what is your approach to ranking in AI search?
[00:01:57] Speaker B: I'm happy to talk about that. I would like to just jump, like, one step before this because you said you're not terribly familiar.
[00:02:03] Speaker A: Yeah.
[00:02:04] Speaker B: And I think, first of all, nobody is, myself included, even though I spend all day, every day with this. I think we're all trying to figure this out. So I don't think you or anybody should feel the slightest bit bad about not knowing all this stuff. And even if you knew it all today, it's changing so fast that three months from now, everything you knew is kind of thrown out the window.
I think the next piece that often comes up in this is.
And I think that even in this conversation and what I talk to most people about is there's this remove, which is AI Search is over there. It's like this new thing that's coming. Maybe. Maybe it's not and my perspective is a lot more that it's here now. And it's here now because of Google, that, you know, Google has said in there. And I think I'm also less trusting of everything Google says than maybe a lot of other people.
[00:02:55] Speaker A: No, I'm definitely. It's not a pinch of salt, it's a pound of salt.
[00:03:01] Speaker B: Yeah.
[00:03:01] Speaker A: Like seeing, okay, if they said that exact phrase, what are the most weasel nosed ways that that could apply to a very specific scenario and be absolutely correct? You know, I read a series called the Dresden Files and he's a wizard always dealing with fairies. And they only speak the truth, but that doesn't mean that they can't trap you eternally into slavery and bondage. So you gotta be very careful with the precise wording of things.
[00:03:27] Speaker B: Absolutely. I'm gonna say in, in the gradation of like what really needs to be truthful and what doesn't. I think the quarterly calls to investors that Google does is one of the places where that truth needs to be really crisp.
Because if not, you know, you have these investor lawsuits and all sorts of other issues.
And in their, you know, quarterly calls they've talked about over a billion people have used AI overviews.
And that's mainstream, that's just mainstream. And we're all using those more every day.
Google is also pilot, is now launched. Yeah. And you know, if you look at, you know, what they're saying is like where search is going in web search right now, there's really two surfaces to this and there's a surface that's, you know, Gemini, where you type into a text box and you get a bunch of answers one way and you type into Google's homepage into a little text box and get a bunch of answers other ways. And those are rapidly becoming one box. Like it is weird sort of that they're two boxes now. So, you know, looking even in the very near future, it's going to be something that looks more like AI search in general than you know, 10 blue links style search.
[00:04:48] Speaker A: Do you think that this is a continuation because I mentioned this with Morty, that we've actually had, you know, on our phones, these AI, these assistants assistant programs for a long time. Do you think that that's possibly what Sundar Pachar was saying? The surface of search is changing in 2025. Do you think they're finally going upgrade the, the primary interface which most people now interact with search is through mobile search. You know, they have mobile first indexing, mobile first experience. Do you think they're finally going to be rolling more of an AI experience into that or are we going to see a step back from that because of the lawsuit? You know, as we look at, maybe Chrome is going to be forced away from Google and their monopoly on the whole space being created through these brand partnerships where you have to use Google. Do you think that's possibly a next step or that's just in the question zone?
[00:05:57] Speaker B: I think AI is in the future is going to be like databases that all of us use databases constantly. Every mobile game we play has a half a dozen SQLite databases in it. Every SaaS app we use has a bunch of databases behind it. And my thought is AI is just going to be ubiquitously saturating every application, everything we do, every text area we fill out.
So I don't see that going away.
I look at this like you mentioned, you know, I have a little bit of a business economics background and I look at a lot of this in terms of action and reaction.
If you're not a developer, I think you might find it shocking the degree to which that the AI assisted tooling has just absolutely slaughtered the number of Google searches that developers do by just a massive, massive amount. And it's not even so much a one to one of like, oh, I used to search in Google, now I go to chat GPT or I used to search in Google, now I go to Claude. It is a lot more just built into the tools. Oh, I need to make a for loop in a language I'm not terribly familiar with. I literally type out in my editor make a for loop and it does it. I'm like, oh no, I meant it to have a zero index instead of a one index and it does it. Then I select a chunk of code I don't know and I just say explain this to me and it explains it to me right there.
And you can see something like that. Where that would otherwise be a dozen, two dozen questions I would have asked Google, gone to different sites, found a different answer and tried to parse all that together. And you know, Google is developers. Something else they said was that 25% of the code at Google was written by AI last quarter.
[00:07:48] Speaker A: Yeah, I saw that stat. That's, that's, it's great.
[00:07:52] Speaker B: And so they are definitely using AI tooling themselves internally and I think experiencing the same thing at a very real level. And if I was in their shoes, I would be terrified that this is the leading edge of all these tools because now that I've had that experience with all the coding Tools like I'm writing something up in Google Docs right now and I keep wanting to jump to those things. Something like Canvas that's in, you know, Claude or Canvas that was just released in OpenAI where you have a document and you just tell the document to do stuff like, hey, replace all the headers with, you know, active voice, change these to this other font. Like I just want to tell it to do this stuff instead of going through and like selecting each header and fixing it, like, just let me tell it what to do. That's so much, that makes so much more sense.
And we are just at the very small edge of all of this stuff and we're going to see just more and more of that happening.
And so to your question of do I think Google is going to pull back on this AI search integration? I think the answer is no, I don't think they can. Because in my mind they view this as an existential threat that the chat GPTs and Claude and Meta, all of these other massive companies that compete with them and pay them money and hate it and back and forth. They are in a spot now where there's a real chance where they can split off and do these things and there is so much money involved in that that it's just, you know, it's just insane.
[00:09:31] Speaker A: Yeah, I think, you know, there's a lot of things about scale at Google that most people sitting at home aren't grabbing. The concept that a football sized field full of hard drives stacked on top of each other is just one of their database centers. You know, just the scope and scale and size of the enterprise that is. Google is outside of our, a little outside of most of our grasp for our day to day process. So it is challenging to see from a SEO perspective how to roll with it as well as what is spinning the wheels versus getting in early. So I created an event for one of my clients building links based off of doing, doing a cleanup. And I went to Perplexity and I did a search and I asked, hey, do you know anything about a beach cleanup in January in Sarasota? And it found the Reddit thread that I created and it and the, that had the link and it was also able to pull the client's page and parse information from it. And I was like, wow, I did not realize, like I was just doing it as a joke. I was like, oh, I wonder if this is showing up yet. But no, like within two hours there's this significant change and I went to Gemini, I went to Copilot, they call it now, Microsoft branded Ping chat is now called Copilot. And I was surprised. All, all five of the ones that I checked, GPT, Perplexity, Gemini and Copilot, you know, they all were able to find that brand information, were able to find that event. And there's also like a positive feedback, like a thumbs up. So my inclination is, oh, hey, this is like a positive feedback cycle.
And I'd seen some other SEOs looking at it and saying, okay, there is a positive feedback cycle to people just inputting queries and that adds information that these tools are then resurfacing.
Which if we look back in time to early days of Google, you know, we, we saw them struggle with that positive feedback loop.
So are we going to see an evolution in AI tools of the, you know, when spammers get a hint of success, they'll take it all the way as far as they can and push it as far as they can.
Are we going go into a new wild west stage here with the boundaries of traditional AI or AI and SEO?
[00:12:17] Speaker B: Yeah. So I think the answer is absolutely, scumbags will do scumbaggery. I don't think there's any doubt of that. And I think in a lot of ways what's happening right now with AI search does look like the very early days of Google and Bing, where there's a lot more simple ways of ranking. And something I see a lot of is you just need to literally state the actual words of what you want to show up. Because I think, as you know, SEO has gotten very into signals and these ranking factors and these different pieces.
But there's still a problem with just almost like straightforwardly saying, what do you do? You know, And I think that's a real challenge. So, I mean, there's one more, I think, interesting aspect of like, is this a now thing? Like, how mainstream are these?
Something that comes up a lot is again, just like, is it a now or is it later?
And we talked about like Google. Another big player in this is Amazon. Amazon now is on the homepage of the Amazon website. In the Amazon app, there's something called Rufus. I don't know if you've looked at that at all.
[00:13:33] Speaker A: No, I haven't looked at Rufus at all.
[00:13:36] Speaker B: So Rufus is their buying assistant and it is based on Claude, the AI service. Claude from Anthropic, because, like, way behind all of this, Jeff Bezos is throwing money around and he threw money to Claude. And so now you have always been able to do product search in Amazon, where I Want this Sony lens, I go and I put the model in and there's a bunch of sellers and reviews and pieces.
What you've never really been able to do is go to that top box in Amazon and type in, hey, I want to do this. And with the lens, I'm going to use it for calls like this. I want to use it essentially as a webcam. What's a good one that matches this body of the camera I have? You can't type that in there. No one would want to do that. You can do that with Rufus because it's backed by Claude and also the Amazon index, you can ask those type of questions. And so this is one of those places where I think we're starting to see a peeling back of some of Google's ubiquitousness. Where there are these jobs to be done is a framework to think about a lot of this stuff. But like, hey, figuring out which Sony lens I need to buy, that's something I can do inside Amazon. Now. I couldn't before and I had spoken about the ubiquitous, like the saturation of AI tooling. It's still clawed under the hood inside Rufus. So you can ask it, hey, write me a hello World in Python. And it will write you a hello World in Python inside of Google's assistant tool on the Amazon. Sorry, said Amazon's assistant tool on the Amazon homepage. It'll then like, pitch you some Python books as well. But you can ask it. You can ask it anything. You could ask it for travel advice to New Orleans, you know, if you wanted to. And it. And it has this certain degree of information and it's backed by a certain continually updated data set. And that's a new weird thing. That's a new weird thing for SEO.
Yeah. And so, and I think along with that, there's sort of a question of can you influence these? And, like, how does that work?
Obviously, I get asked this a lot, so this is actually the first time I'm actually sort of debuting this. But we call it the Biscuit framework. Okay.
[00:15:53] Speaker A: And so Biskit Framework, let's go.
[00:15:56] Speaker B: Biscuit. Yeah. So this is trying to step by step have myself or an SEO, like you're talking to a client. This is to walk them through the levels of awareness and capabilities. And, you know, I talked about being positive, something to think about. Like right now there is such a link between SEO and Google. Like, it's just everything. And in the future, I think there's going to be a lot more need for a lot more strategic thinking, because that's not just going to be the case where suddenly you have to think about Amazon, Rufus. Suddenly you have to think about how do we also work in ChatGPT? We have to monitor that. How do we work in Perplexity? How do we work in Claude, how do we work in like? And let's just pause there. Apple and Meta haven't even really released their stuff yet, you know, and there's.
[00:16:51] Speaker A: Two, two more players. So when we go from one monolithic, like I remember when people stopped asking how do I rank in Yahoo?
[00:17:06] Speaker B: Yeah.
[00:17:06] Speaker A: You know, like yeah, Ask Jeeves isn't.
[00:17:09] Speaker B: On anyone's, you know, audit list.
[00:17:11] Speaker A: Yeah. When I started like, you know, AltaVista, Hotbot, Jeeves, like there were legitimately other search engines.
So SEO wasn't just Google, but it narrow, it certainly narrowed down and then it was almost just, you know, there's also Bing, which was Bing and Yahoo, but it was really like an afterthought of like oh well, you know, they've got better webmaster tools. But you do the same thing for Google that you do for Bing. So really focus on it now.
[00:17:42] Speaker B: Yeah. And so this framework is also trying to express a lot of my frustrations because I've seen a lot of articles like hey, here's how you rank in the LLMs. And most of those articles seem to be warmed over. They're all directionally positive things to do, but I don't think they actually matter for ranking in ChatGPT.
[00:18:02] Speaker A: Okay.
[00:18:03] Speaker B: Things like work on your page speed, work on your knowledge graphs, like all of these like sort of pat answers for stuff you should be doing anyway, so through the framework. So it's an acronym. Of course it's an acronym. So BE is blocking bots. And this is something again, like imagine you're an SEO, you're working for a site. Just go to the robots txt right now and check do they have an exclusion that's disallowing, you know, ChatGPT bot to index the site. And there's a lot that were out there because there was a really knee jerk reaction about like, oh, they're gonna steal my content.
And you know, my take is that there's a lot of different kind of content in the world and that's like a Stephen King problem, that's not a me writing marketing copy problem. Like it's not really possible for my marketing copy to be stolen. Like please take it, you know, and.
[00:18:57] Speaker A: I want to see my marketing copy on as many sites. In fact, part of your job is to get ripped off. It's like, oh, there's 15 sites over here that also do marketing. Take my text as long as it somehow leads back to me.
[00:19:11] Speaker B: Yeah. Like even this biscuit framework, we're going to put a post up about this and, you know, you can read it all and go through it. Please steal it and write your own article about it. Yeah. So the action is you just want to make sure these bots can, you know, index your site, can crawl your site to get the information. Now, this doesn't mean you have to let everything out there. If you, you know, spent months building a very report you release once a year. That's a PDF, that's gated. Yeah. Lock that off. Don't let that be part of it. There's still funnels. They're still trying to get people into things. But I think, you know, even more so than we've experienced with SEO, we need to get more information into the LLMs because they base their decisions off more different information.
The other piece of this is that, you know, if you have something like Cloudflare or if you have something that's like, you know, there's a lot of WordPress plugins that do security bot detection.
[00:20:04] Speaker A: Yeah.
[00:20:05] Speaker B: You may be inadvertently blocking these even if your robots Txt is open. And I've seen that too.
So there's some different ways you can check that. We have one built into our service, but you can still check that manually.
So once you've sort of, you know, assured yourself that they can get in, the next part is really sort of indexing in the same way that you think about Google Search Console, like, is, you know, Google indexing my site. Are there issues? And one of the easiest things to do is just ask it, like, hey, who is my company name? Like who? And we talked earlier about, like, a lot of times we don't actually state what it is we do. You know, we're the, the Iowa Corn Farmers association, and we just assume that everybody knows exactly what that is, you know.
[00:20:56] Speaker A: Yeah, I definitely run into that. I had a, a client and they had pictures of restaurants and they're like, oh, what's your restaurant like? Oh, no, no, we, we do signage for corporations and we swap out all of their brand elements. I'm like, you know, you never actually said that on anything on your site. You said, we've got the best customer service. We'll turn around what you need right away. Professional, like, professional what? You know, we literally went through, we copied, copied and pasted the text into a document. I said, okay, tell me what this business does and they couldn't based off of the text, like, oh, we never actually said what we did.
[00:21:35] Speaker B: So.
[00:21:36] Speaker A: So it's as simple as that to start.
[00:21:40] Speaker B: Well, again, we're starting as a consultant. You don't want to be caught flat footed. And something like, I talk to a lot of people who come to me in a little bit of a panic because, hey, my, the CMO of the company I'm working for, they just asked me how come when I ask in ChatGPT it doesn't say what we do, it says the wrong thing. It says some other very minor competitor that happened to actually state what they did on their website six months ago. Like it says that instead of our thing.
And so a lot of this is just, you know, trying to come up to speed on this. And so just ask in all the different search services, hey, what's our name? What do we do? What are we best known for? And keep track of that. Like that's not craziness to do. And I think some of this gets into two misconceptions people have, one of which is that these services are really static and you've kind of already spoken to that with like how quickly they get updated. And I think there's a real blurring together of two different concepts, which is AI models versus AI search services.
An AI model is trained up to a knowledge cutoff and it doesn't know anything before that.
[00:22:56] Speaker A: Correct.
[00:22:57] Speaker B: But you know, like you said, there's additional information that's coming in. A lot of these services call out to the web, they call it to Bing very often, get results back, show you those. And even something you can try is you can ask ChatGPT, hey, what's the date today? Like it will tell you the correct date. And it does that because it is calling out to some system service that says what the current time is. But it's certainly you can't impart the knowledge to it of what is the day today into a database of information about it. So these things change a lot and they change both in terms of the data that's updated in them as well as what we would think of as algorithm updates, the weights that are changed. And if you look at just ChatGPT, between ChatGPT 3.5 and 4, there's actually been like 13 different updates where they've either updated their data knowledge cutoff or they've updated their weights and how they interpret the information.
And that's before you even get to. Oh yeah, if you ask something and it decides like I Don't know what this is, and it calls out to Bing. So there's a tremendous amount of change that's happening here. That again, you know, anytime there's an algorithm update, everyone in the SEO community is very interested because this means impactful things for our clients, for our companies.
Yeah, so that's the I and then S is for sentiment beyond just knowing about your company. You want the AIs to think positively about you. And this isn't in, like, the Terminator sort of way, but.
So here, here's again, some examples.
Ask, is my company ethical to ChatGPT? Ask, you know, is my company responsible? Does my company treat people well?
And the reason you want to do this is that AI search mashes up all of these different sentiments into like a big gooey pile in a way that, you know, asking, hey, what's Wells Fargo bank, you know about? In traditional search, you get sponsored ads, you get Wells Fargo site because it looks navigational, you get, you know, maybe something else. What you get out of chat GPT is, hey, they've had a lot of issues. They've done a lot of weird, different things. And they mentioned this, which is like, hey, they had an issue where like 500,000 credit card accounts were created without customer consent. You know, and that's a difficult thing. Like, if you're a brand manager for them, how do you deal with that? Do you want to be in that situation? Does that color, like, hey, what bank should I use? Should I use Navy Federal, which, you know, ChatGPT thinks very highly of, or should I use Wells Fargo?
Like, these are real, actual things that are happening. So you want to track the sentiment and work on that. And then after that, competitive ranking, actually, and we talked a little bit about AI overviews.
I think there is also a strategy change where so much of what we used to do as SEOs was really about top of the funnel, because top of the funnel had very high search volume.
And it was a way to do very classical, sort of like, I'm going to make this skyscraper technique article that's like 10% better than the last person's article and cover a little bit more information. I'm going to get a thousand people a month into this article, and then maybe 50 of them are going to sign up for my newsletter. Then I'm going to. And I think the AI services are so much better at answering these type of questions that I don't think that's where SEO should be putting their emphasis in terms of content strategy.
[00:26:47] Speaker A: So it's more like BofU, like at the bottom of the funnel. Comparative, like you versus your competitor. It'd be better for AI to ingest your spin on the situation. Your comparison where you put yourself, like, try to be somewhat, you know, parallel, but listing more of your advantages, not just saying, oh, they're a dirty dog, like, yeah, but having that article as something that they would ingest and add to the mix. If that didn't exist, then you're rolling the dice as far as what random third party review is going to be out there and whether they like your platform better or not, it's right.
[00:27:35] Speaker B: So like here's. So this is where I see the divergence of you only have so much money, you only have so much time. Like every company has these limitations. So where are you going to put the emphasis and where are you going to put the effort? And you know, if you're a bank, you know, are you going to write articles about compound interest? Are you going to write articles about the different types of savings Accounts and Roth IRAs or is chat, GPT and AI overviews just going to answer all of those questions much better than you can with like handy buttons for like. I didn't understand that. Explain it to me more simply. You know, I, I totally understand that. Give me the better, you know, definition of that.
Do you want to put your efforts trying to beat out the, you know, I think about it in terms of like search quality. Like that's something SEOs talk about. Like, did we actually answer the search intent of what the person was looking for? Did we actually, you know, satisfy what they're doing? And the truth is nobody wants to search for anything. They just want the answer.
Nobody wants to go on a search journey. They just want the answer. And the AIs are much better at doing this. And I think that is just a losing proposition that you as a person with a single site are going to be able to write an article that's as functional as this interactive, you know, very easy to use system that says like, here's, here's your answer. I can answer follow up questions. It's just right here, you know, so don't fight that battle. Fight the battle for like, hey, what bank has the best, you know, Roth IRA plan? I can sign up for what financial institution, you know, best money market account. And try to put your emphasis there as well as on, I think, more strategic opportunities. I'm not actually that familiar with banking. I'm much more from the SaaS world.
[00:29:29] Speaker A: Yeah.
[00:29:30] Speaker B: But you know, to start having more, less explicit Just search content and more content that you're developing and writing that's actually trying to reach people. And I think that's something we've gotten away from in the search space.
[00:29:47] Speaker A: Yeah, it's almost more we used to like as an SEO, it'd always be like, oh, hey, you know, oh, this quarter of your search is branded. Let's ignore that and let's look at what your piece is unbranded or oh, hey, you've got 75% as you're branded, you need to expand your unbranded query space and create content that's going to target outside of that. It feels like that paradigm has shifted because as you're saying and pointing out like it's not completely true yet in all regular organic search, that you can't create generic, you know, topical industry information. But it seems like more of the future is more about mixing that effort with a brand focus and mixing that effort with, you know, expanding those branded queries as well as other activities elsewhere to boost those signals.
[00:30:44] Speaker B: Yeah. And if you take it as like that, it's not, it's not just search content, but it's content in general. But it's a content strategy, not just an organic search content strategy.
[00:30:57] Speaker A: Interesting. You're challenging my worldview, Michael.
[00:31:02] Speaker B: Did you see Rand Fishkin's Spark tour report about like the breakdown of the different search types and like, what's happening in there?
[00:31:11] Speaker A: I've seen some, you know, for somebody that like left SEO, like, he hasn't really left SEO, like he's still cropping up and doing some, I think some very fascinating and very much needed breakdown of information he had touched on. I did read his piece about influencing LLMs in terms of asking them what are the citations and sources that you use to determine that this was a top rated restaurant to go to and then use reverse engineering and saying, okay, well this is either what I need to target or sites and resources similar to this, which turns SEO back into more of a marketing, a bigger marketing role, digital PR connection and thinking about not just like, you know, not so much just going in semrush, looking at keyword volumes, putting out content and gaining links. Instead, it's more about, okay, well, where can I get placed? That's going to be positive for my brand. That's going to be, you know, influential.
It's definitely, you know, how is it going to improve the understanding of the entity that I am optimizing? So it's almost entity optimization. Whether that's schema data directly, whether that's content on the page or third party references. Like I made a post about this on Social last week that I'm starting to think about SEO. As you know, the E is for entity search, entity optimization and you're optimizing your site to be the best understood entity in all of the ways possible. And that is the optimization process. Rather than taking one tool and saying this is what you know and that's Google, like, this is what they like. So optimization means that. No, it's, it's optimizing your entity to be, like you said, accessible, understandable and well liked. I guess that sentiment. So what's the. Is it, is, is it biscuits, like British biscuits or is it, is it a cake?
[00:33:22] Speaker B: Just biscuit. So CU is for, for competition. And so this is what we actually track in Noah Toa. We actually track like, hey, if you're like, what's the best email marketing platform, what's the best web application firewall, what's the best restaurant in Seattle?
So we track that and we track that across the generations of the different search models, the AI models within that. So within ChatGPT we track like, here's what that looked like in ChatGPT 3.5, here's what it looked like in 4.0 and here's what it looks like now in search. And we do that actually not really caring about the citations and not really caring about the websites because we know so much of the traffic. Isn't that like great, you showed up as a citation in AI and an AI overview. Did you actually get any traffic from that or did they just get their answer and they went on their way? They probably just got their answer and went on their way. And we see a lot of that with how people use chat GPT. There's certainly some traffic referrals from it, but there's a lot more of the discovery happening in the AI search world that then gets flipped around and you do a navigational search in Google.
[00:34:41] Speaker A: So it's, it's again about reflection of brand and visibility, but not one to one. It's like it's like on social media now of there's so much dark social like yeah, that TikTok is not meant for links. Like you're watching a TikTok, you're not going to click a blue link and go to that site right now. But you might remember, oh, Haberdashery is a company in New York. What's Haberdashery New York? Oh, and then that's why there's so many more. Navigate, navigate Navigational branded searches.
[00:35:14] Speaker B: Absolutely, yeah. And so in terms of really trying to like, dissect this, a real interesting thing when you look at how what the outputs look like of doing the same query across the different models is they're getting increasingly sophisticated. Like, so what's the best email marketing, you know, service? You ask that to 3.5, it gives you like a short list. Like, here's 10 names, you ask that in four is a summary. Under each of those names, you ask that in the latest, it actually breaks out, hey, here's the key features of each of these. Here's who is most likely best for small business, best for large enterprises, best for legal people.
And by splitting out that, we're able to then reflect that back in what we write. So it's not enough to just write, hey, we're the best for this. We need to actually reflect that back and say, okay, here's our key features. Put that back. Who is this best for? And to write the content and pages for that. And why I'm so positive on this for SEOs is that is just hugely valuable for both the very narrow case of organic search as well as for all the other marketing functions that are happening. Having that page that really speaks directly to each of the niches, each of the verticals you're operating in, that's a huge. That's huge for the sales team, that's huge for people writing newsletters, that's huge for people browsing your site that didn't come in via search for all of those things, that's a positive. That's a win for us, you know, and it happens that this also helps to educate, you know, the search engines.
The other piece of this. So moving up to now, the U is there's unique distribution for each of these.
So ChatGPT, they and OpenAI, they have a tremendously complex and weird relationship with Microsoft.
[00:37:11] Speaker A: Yeah.
[00:37:12] Speaker B: And I don't even want to get into it. I honestly don't even really understand it. But literally at the tune of like billions of dollars, like, they are team Microsoft, and part of that means that they have preferential access to better data from certain sources than others. Microsoft also owns LinkedIn. Microsoft also owns GitHub, and they have access to those. So if you're trying to rank in ChatGPT and it's not displaying your brand right, do you actually have a LinkedIn page? And this is something that I've come up with multiple times. Like, sasses, they're like, yeah, we do. And you look at it, it hasn't Been updated. No, you don't. You don't have a presence there. Like, maybe you have a tiny flag in the ground. And I've seen this, and even in other areas, I was actually helping someone who. They had like a ski mountain resort and E commerce shop for ski equipment.
[00:38:08] Speaker A: Yeah.
[00:38:08] Speaker B: And their competitor, who is across town, has like 500 followers on LinkedIn, which is not huge. That's not an insurmountable kind of thing. They have 500 followers and it's fairly well updated, which I think just means they paid the extra $10 in buffer to add LinkedIn as a channel. So when they publish something, it also shows up on LinkedIn.
[00:38:31] Speaker A: Yeah.
[00:38:32] Speaker B: And they absolutely were getting crushed by this company because they had that. The other thing, this competitor company that was doing much better than them in AI search was they had a subreddit, and it was about the area. It was about, like, regional skiing. And it happened to be their brand name was the subreddit name, but it was about regional skiing. And so it talked about, here's the weather today, here's the other stuff, and we're just absolutely crushing it for those things. And so distribution is per AI search as well as, like, your website, because maybe you can get your website going, but you could certainly do these other things. Like, these are not insurmountable obstacles, but they are an expansion of responsibility and an expansion of role and I think an expansion of the value of what SEOs bring, which is thinking about this content in a strategic way. Distributing it, writing it, running those, looking at the results of that.
And all I can see, every time I dig into any one of these is, oh, my gosh, there's so much work to be done.
And.
And that's why I'm so positive on this.
Yeah.
[00:39:49] Speaker A: And so moving along, remind me what the U was.
[00:39:53] Speaker B: The U is unique distribution.
[00:39:55] Speaker A: Unique distribution.
[00:39:57] Speaker B: I would call it just distribution, but that totally destroys my accuracy.
Yeah. And so this next part is intelligence, which is that business intelligence, that we do want to track the metrics of this because we do want to show the value of what we're doing.
And we have, you know, people seeing improvement, and that's what you really want. And that's sort of the answer to the CMOS question of how. How are we doing this? And the thing we've seen most effective because we are rapidly moving into less and less attributable sort of world, is the easiest way to do this is to put it in context of the competitors. Like I mentioned, the two ski resorts. I mean, you have to make a decision when you go there, am I going to this one or am I going to this one? You're probably not going to both.
And so putting it in terms of, oh, we do show up better than our competitor, we are now seen in more places than our competitor. That's something that helps ground a lot of the information in terms. And it's more, it's honestly more a way for the SEO person who's trying to advocate for these campaigns to be able to say, look at what they're doing. We should invest more in LinkedIn in order to get these kind of results. We should invest more, you know, on doing Reddit. And the last part of this, the T is for truthfulness.
And there is a theme of like SEO expansion. Just consider this. This is actually something a CMO said to me, which is that ChatGPT is now our most popular and least well trained representative of the company.
[00:41:35] Speaker A: I like that. I'm going to steal that quote and put it on a picture.
[00:41:38] Speaker B: Absolutely.
And it's the truth because they were not developing content that went out there that explained all of, hey, do you do business in California? That's a money question. Like, if the answer is no and I'm in California, I'm not going to use you for my services. Like, does your product have an integration with mailchimp? If it doesn't, I can't use you.
These are very important things to convey.
[00:42:09] Speaker A: Boolean.
So Mark Cook, he put out a distribution of the framework of these types of queries that all queries fall into. And of course my brain was like, oh well, how many in an article, which intent should I optimize for? But if we back up and maybe we look at those intents to kind of look at if this is done in traditional search these type of, you know, Boolean. Yes. No instructional walkthrough like what of those are left if GPT does the information, like the super informational stuff, what's at the bottom of the funnel that I need to address in my marketing effort? Like I need to have something out there that says, yes, I optimize flanges or I don't optimize flanges or I my poop scooping service provides for apartment complexes or it doesn't provide for apartment complexes.
[00:43:14] Speaker B: So well, how I've been asking people to do this is with FAQs, just like up your FAQ game and make it very clear. Here's the question, here's the answer. And think of this as, you know, it's AI Content bait. And you really are trying to train these services to answer these questions for you because the services are very confident. You know, ChatGPT, you know, it may say in some cases I don't have enough information for that. It might search for things, but it has to have something to search for. And certainly it is very confident even when it's incorrect, which is something as humans we are just not great at dealing with. So people are asking these questions of these services and you want it to give the right information. So part of what we're doing with no ETOA as well is tracking these questions and then seeing are the answers correct, you know, is the pricing the right, what it should be, you know, all of these different kind of things.
Yeah, and so that's, that's the Biscuit model, which is Biscuit framework, whatever you want to call it. But it's really just a series of steps that, trying to be very reasonable, not upend everything you're doing, but you know, to take positive, you know, steps to improve SEO as far as your business goes, which I know is a big thing that you advocate for as well, I think is really cool. As well as, you know, the profession as a whole. And so that's sort of why I'm positive on it and why I think, you know, more people should be looking into this now rather than later.
[00:44:52] Speaker A: I love it. Anytime I have a useful acronym I can, I can get a lot more done to wrap my, my squishy brain matter around a concept and it actually does make sense.
[00:45:10] Speaker B: Well, and the nice thing so much of. I'm sorry, I didn't mean to speak over you, Jeremy. I was gonna say so much of this is really about trying to manage up that, you know, as a, as a client, as a. Sorry, as a consultant, you have to manage your clients. As an in house, you have to manage your executive team and any piece in here you can pull out and it's a net positive for the marketing group, it's a net positive for the company.
[00:45:36] Speaker A: Yeah.
[00:45:37] Speaker B: It's not something that detracts in any way. It's not something that you can do and mess up and fall on your face in an embarrassing way because these are just very clear communication items and you know, easy enough to, on that front to propose this. It's very risk free in that way and it heads off so much of the, yeah, how come we don't listen this? And the questions of like, yeah, go put us in chat GPT. Like go put us in the search console for chat GPT. Like these are the things that people ask me. Like, I'm like, that's not really a thing.
You have to, you have to do these steps that implicitly sort of get you in there. And I think those things are coming. I think there's going to be a lot more ad products on this stuff. I think there's going to be a lot more tooling. But right now this is, I think, the best practices.
[00:46:30] Speaker A: I, I think at some points we've always said, what would Google look like if we could pay for it, you know, as a paid service to give us the answers? And I think we're actually seeing that that is what these AI tools, their models really are. They want subscriptions to fuel it and not be, not have the ad revenue basis as its starting foundation. So it's kind of interesting to see the, the way that, that influences the, the continued evolution of these tools. But I do have one, I want to bookmark that and come back to a question that I, I haven't seen addressed is hallucinations. LLM hallucinations. And in, you know, in the basic idea of, you know, using GPT itself as a, as a research tool is flawed because it makes stuff up like it's literally like a liar, you know, like, but you told it to lie to you, like comfortingly, like based off of what it knows, lie to you as effectively as possible.
So it's interesting to me, you know, with the integration of these search elements, do we see as great as, as a highest, highest. The same amount of propensity towards hallucination? Like, you know, in GPT4, I asked for articles to tell me about this specific event that happened written by New York Times authors and it made up a list. Half of the authors didn't actually exist and the quarter of the authors that did exist didn't write those articles. And the remaining quarter of that quarter that wasn't the right article or meant literally the opposite thing with the introduction of LLM powered search is to what degree does do hallucinations still exist in your experience? And you know, you're playing around with this a lot, like, how often does that occur? How big of a problem is that that they're trying to address with it? Or in what way have they changed how the results are served to minimize those hallucinations?
[00:48:47] Speaker B: Sure.
A couple different items here. One of which is you're making incredibly good points about this, especially in terms of there's a lot of ways in which AI search is not great at search, especially compared to Google and some of that is just, it's a lot newer, you know, Google's 20 plus years of use now versus, you know, AI search, which, and this is going to sound crazy because I think it's already been so impactful on the SEO industry. It's an accident. It's an accident that it's kind of good at search in a lot of ways that people weren't prepared for because it wasn't written. No one sat down and said, like, I'm going to spend a bunch of money to make this LLM thing. It's going to be great at beating Google. They didn't make a company around that.
And it does tons of other crazy stuff that, you know. And we've, we've spoken about it as a database really, where it's a database of a bunch of information, but you can't take a database and then ask it to also make a picture of a flower sitting on the moon. You know, you can't take a database and also ask it, like, write me, you know, a new vampire story and like Bram Stoker, you know, would, you know, that's. It's a new, different thing that happens to also be good at search. And what we're seeing now is as it goes from a model to more and more AI search services, we're seeing more and more of those issues being beaten out. A good example is, you know, we've all joked about this, like, you make an image of a human in any of the services and, oh, they have six fingers.
That's really a problem from like a year ago. Any of the services now almost always have the correct number of fingers. And what did they do? They literally went into like the game tools and they took pictures of lots of hands and lots of different orientations and got more data and more information. And now those models are really good at making hands. There's still issues sometimes, but it's better. And we're seeing that with AI search as well in that in a lot of ways, ChatGPT and Claude and these other places that still have billions of dollars have not even really tried to compete with Google yet in a lot of ways because they're doing other stuff. They're doing code generation, they're doing all these other things.
So that's part of it. The other part is from a very abstract way. There's two different kinds of hallucinations. There's hallucinations of ignorance, which is, oh, we don't have this data, so we made something up. And then there's hallucinations of reasoning which is that we did have the data and we said, yeah, it's a good ide to mix together flour and ketchup to make a cake. Like a real different thing.
From an AI search perspective, the hallucinations are based on ignorance, which is, oh, we don't have your pricing information, so we made something up. Oh, we didn't have your feature list. We made something up.
So that's addressable. Like, we, we can deal with that for our clients.
We can't deal with the reasoning stuff, which is getting better and better and better and cheaper and cheaper and cheaper.
So it is the early, early days of this stuff. And so I am, I don't argue with anybody when they say there's issues. There are issues. There are lots of things. But there is so much promise, I think, both for the services themselves, for us as SEOs, and for, you know, people to have better search experiences and new tools and ways of doing things.
[00:52:25] Speaker A: So where can people find you? I know you're going to put out this biscuit framework. I'd love to see that all come out. So where can people find you? And tell me just a little bit more. I want to know what your service actually looks like. What is it that you intend to do when somebody gets to your tool?
[00:52:48] Speaker B: So service is called Noah Toa, kind of like Krakatoa, but for knowledge.
And you can go to knowatoa.com k n o w o t o a dot com. I'm sure you'll be very nice and put it into the.
[00:53:01] Speaker A: Absolutely. Yeah. You'll get tons of links. Don't worry.
[00:53:05] Speaker B: We are indexed on Google. We're not saying give up all your things. You just type in Noah Toa. You'll find our site and right now it is open to use. You can go in, you can type your site in. And what we will do is we will extract the current traditional SEO keyword terms you rank for. We then put those through a whole process to try to figure out, take out the brand terms, take out the terms that are informational, that will probably be answered directly by the AIs, and try to really focus on those commercial terms that are going to make or lose you money.
And those are the terms, like what is the best xyz, what is the best category in here? Who's the leader? What are the factors we should be looking at? And like, who is hitting those factors? We'll generate all those questions for you. And then on the free plan, no credit card.
You get the top 10 of those out of chatgpt so this goes across those different services and you can see where you rank right now, where you stand, which is most people don't know, you know and most people, someone is going to ask you sometime soon. Like I haven't seen a lot about this Chat GPT and I typed our company name in and didn't really come up. I typed in something and it's not there. Like are people coming in? And this is a way to get there. The paid plans then crank up, you can ask more questions, you can put in your own questions about, you know, is this the right information? As well as currently we cover ChatGPT, we also cover Gemini. Gemini from Google is what is used as the basis, because Google said so, of AI overviews. So doing well in there is good for AI overviews. We do also cover Perplexity, which you mentioned, which is an AI hybrid search engine. They have a Sonar LLM model and we cover Meta's models which you can get to. If you had a Meta AI, you have a chatgpt like experience for free and people are using it. And I could talk another hour about like all the things I think Apple and Meta are going to do but you know, I think in the next year we're going to see WhatsApp just like built in with a Chat GPT competitor of like you just ask anything you want and it just tells you the answer from, you know, their LLM, stuff like that. So yeah, so I would recommend. If you're listening to this, please check it out. It is a new service. We've had some troubles both lots of people using it and the other pieces, the AI services are really fragile and so we've had a lot of. Oh, Claude's down today. Well, that'll make answering questions on Claude a lot harder. So yeah, but I think we're in good shape so please try it out. And I'm on most of the different services and happy to talk to pretty much anybody really big on Blue sky right now. That's been a lot of fun being on there. That's where we've been chatting.
[00:56:05] Speaker A: Thank you so much for your time, Michael. And I'm going go have some biscuits with my clients after this.
[00:56:11] Speaker B: Awesome. Thanks so much, Jeremy. This has been a treat. Sorry. Bye.
