Episode Transcript
[00:00:00] Speaker A: Foreign.
[00:00:08] Speaker B: You know that feeling when you're talking to somebody and you know instantly, wow, they know what the hell they're talking about. And this totally makes sense. That was my conversation with Mike Buckbee of Noah Toa talking about AI SEO and how we all should be thinking about getting our brands recommended and mentioned to our potential customers in LLMs like ChatGPT and Gemini. And Claude, Mike has some insights that I've never heard before and I've been pretty deep down this rabbit hole for the last few months and he blew my mind. Things that we should be thinking about both from a website and brand perspective, but also from a distribution perspective. He totally flipped my kind of 8020 rule of content distribution on its head. I can't wait for you to check out this episode with Mike Buckbee of Noah Toa talking about how to get on our brands represented and mentioned in LLMs like ChatGPT, Claude and Gemini so that you can get more customers on autopilot from Inbound Marketing. Hope you enjoy.
I think I want to start by kind of talking about why you saw the need for this, right? Like why does Noah Toa exist? What did you see kind of going on that you say I need to build a tool for this?
[00:01:22] Speaker A: Well, you and I are in very similar communities, so we, I think also have a lot of SaaS co founder friends. And it really started, it was almost a joke, which is myself and the co founder and a couple other people were talking about like, hey, let's just throw our company names in chat GPT and see what comes up. And this was like, you know, basically almost a year ago and stuff was wrong and stuff was weird and lots of issues.
But also, you know, I'm a developer, a lot of, you know, developer co founders and it's hard to overstate just how much AI has taken over the, the developer space.
And you know, if you think about this as different verticals adopt, you know, technologies at different rates, it has just been wildfire in the developer space and it has had a massive impact on the types of Google searches people do, where so much of it has just disappeared.
And you know, putting all that together, you know, how we started was literally, you know, I wrote a script that hit OpenAI's API and spit out a CSV and then I put it into a Google sheet and you know, I sent it around to a few co founder friends for the company we were running at the time. And then, you know, a few of the friends and everyone's sort of like, that's interesting. But doesn't really matter.
And then a few people who had sent out post purchase surveys, which is like, hey, just how'd you hear about us? You know, like just interested to know, you know, how we can reach people better kind of stuff. They started getting, hey, I heard you about you on Chat GPT.
And so then all of those founders started coming back to me like, hey Mike, can I get an update on a spreadsheet? And like, and you know, we saw that too. So in our own behavior, we just saw a really big shift. And I think a frustration with talking about SEO in general is that a lot of the terms are not really nailed down. And so there's.
[00:03:18] Speaker B: Sorry, what do you mean by that? Sure.
[00:03:20] Speaker A: So what is a query? You know, like there's a lot of people talking about like, oh, Google gets so many queries a day versus chatgpt. And sort of implicit in that is we can ignore Chad GPT, we can ignore these new things because the majority of everything is still happening on Google.
But you know, if you really dig into it, what you find is like the number one query on Google is literally the word YouTube.
That's not a query. I mean that's, that's a navigation. Like I'm trying to get to YouTube. I already know what I'm doing, you know, and half or more of Google's queries are that.
[00:03:59] Speaker B: And that's not brand, brand words like castos or brand words.
[00:04:04] Speaker A: The SEO term is navigational search. That like I'm looking for Salesforce login. That is not a decision making query. I'm not researching, I'm not doing anything like that. And, but that's sort of how it's taken in the mix of things. And so, you know, it, it's just, it's not exactly, you know, apples to apples making these comparisons. And so there's a lot of that where, you know, there' only a small sliver of, you know, research being done, which is like, hey, I have this problem, I'm trying to figure it out, I'm trying to do this thing. And that is where I feel like the AI tools Excel, which is like, hey, you know, my, my website's being attacked, I don't know what to do. What are some tools for this? And it spits out, oh, there's intrusion detection, you know, network monitoring, web application firewalls. Well, I'm having this issue.
Well, here's the one that's probably best for you. Okay, great. What are my options in here? And then it's been about a couple and you know, there's a whole, like, search buyer's journey that's happening in AI that is being removed from traditional search for a lot of search quality reasons. And, you know, if people are making business decisions, they're. They're turning to AI tools a lot more than they are Google. So.
[00:05:22] Speaker B: I, I don't know if you listened to the Marketing against the Grain podcast, uh, from the, the folks at HubSpot. I listen to like two podcasts these days. One's a history podcast and the other is this one because they. All they talk about is like, AI for marketing. Yeah, it's. With AI, it, you know, it has shifted quite a bit. And it's like, that's kind of solidly where my kind of passion is these days, is like, hey, this is, you know, as important as electricity, as Bezos says. Like, how. How can I make sure I'm on the right path here? And they had the kind of internal SEO person from HubSpot.
[00:05:58] Speaker A: Yeah.
[00:05:59] Speaker B: Which, like, that's a whole nother topic talking about AI SEO a couple weeks ago.
And yeah, like, a lot of it is behavioral changes that folks are seeing is like, hey, I mean, I have our Google search console pulled up right now.
Year over year, we are down 20%, 25% in just impressions and like clicks is the same. So. So it's like it's all going somewhere else.
[00:06:28] Speaker A: But we see you mentioned HubSpot. So HubSpot, you know, was really the leader, especially for like B2B SaaS, which is where my mind usually goes, you know, in SEO and content for a long time. And they have, I think, done a great job at shifting over to sort of the new playbook and the new ways of thinking about this.
And here's an example of sort of what they've done.
They used to rank incredibly highly for the keyword resignation letter, you know, which is, I'm trying to leave a job. I want to try to not burn my bridges too much. I want to write this. And so my intent is I would like a couple paragraphs of text I can put my name on and my boss's name, email it. And, you know, they ranked number one for that. Millions of searches a month.
And you landed on their page and they had the classic super SEO optimized thing of it, which is a lot more text than you actually needed to describe this. Like, so they had people talking about like, oh, here's an expert at, you know, you know, you need to be. And saying the same thing eight different ways to like a third grader because that's what they were trying to teach Google and it still wasn't very helpful. And then you'd click on a page and then they had a progressive form for you to actually try to get the PDF template and stuff and they, and you'd go through all of that and you know what happened was Google had an AI overview that was for resignation letter that would just pre write one right there in the AI overview when you search for it. And then it had a tab at the bottom that was like, do you want to just launch this as a Google Doc or do you want to launch this into Gmail directly and we'll just send this.
Yeah, that's, that's a whole new SEO threat in that one. There's no, all the normal SEO tricks and competition, there's no content strategy that's going to beat that. There's no like, oh my, my title, my meta description should be tweaked a little. I need more internal links. No, you just need a different strategy. And to their credit, I think HubSpot has decided we're going with a different strategy. And so they're doing a lot more content that is no one could write but them, which is, hey, we're HubSpot. Here is how we approach these problems. Here is how we generate content that's useful, multiple people in our teams. It's a lot more branded, it's a lot more A term I use a lot is space versus faceless content, which is, you know, if you cannot say like when you read the words, you cannot picture the face of the person who wrote it. It's probably going to be done by an AI better and you know, you need at least some of that perspective. And so I think that's, that's a big change which is away from these very high volume, low intent keywords because, you know, HubSpot, what do they really do for resignation letters? They're not a resignation letter service. They don't do anything with that. But it was just like vaguely professional work stuff.
So, you know, you're not going to fight over that. You're going to come down funnel. You want to really fight for the determination type of, you know, searches about like who is the best in the category. I'm trying to solve this. What are some players in there? What are the right categories? And at Noa Toa, which is a service for tracking a lot of this stuff, we have seen some really dramatic changes in like Chat GPT that if you look at the early versions of Chad GPT and you ask like, hey, what's the best email CRM service which HubSpot would be in? You would originally get a list of like 10 companies and now what you get is a list of, hey, here's this company, here's the key features, here's they are best for this category.
And what we're seeing our customers that are having the most success do is reflect that back. And so, you know, Castos say, hey, we are the best podcasting platform for X.
Here's our key features, here's what's going to happen. And because a lot of times those things are wrong or they're the consensus of what the Internet has decided that you're best for, even if that's not strategically where you want to go.
And that's a little different thinking in terms of SEO. That's not a normal sort of SEO task, that's more like a product manager task kind of stuff. And we see, we see a lot of that, which is that the traditional SEO, you know, metrics of like, I'm trying to get these high intent keywords, I'm going to write these posts, it's going to be this traffic stuff and you can sort of outsource it.
I think that becomes a lot harder. So.
[00:10:53] Speaker B: And when you say like, I think this is to the meat of like, what do we do with this? Like when you say we reflect that back. Like, the way I think about all of this is there are questions that people are asking, like actual questions like what's the best CRM for a medium sized business? Or what's the best podcast host for WordPress? You want ChatGPT to know that you are the answer to that?
[00:11:14] Speaker A: Yeah.
[00:11:15] Speaker B: As opposed to what's the equivalent, like podcast hosting platform? Like, that's kind of the keyword we think we should rank for in Google. The things we would do to rank for each of those in Google or in ChatGPT are like quite different from a strategic input perspective, right?
[00:11:32] Speaker A: Well, I think yes and no.
I think I used to be like an in house SEO at a big cybersecurity company and I worked with big agencies and did multi language global SEO stuff for a long time. And the strategy coming into Castos, you'd be like, okay, well what are the things that are really leading into this? Like, you know, because people don't just go like, I'm going to make a podcast, you know, maybe it's something else. Like, you know that you have like stats pages around podcasts and those, you know, lead into this, or you Know, something that's not exactly looking for the service, but like, hey, here's our list of best microphones for podcasters, you know, and a lot of those queries are not things that you're going to rank for anymore because they're, they're too high level. The AI is going to do better. Google will have an AIO for it. So you need something that you know is lower in the funnel. And, you know, we're all, no one has an infinite marketing budget. We're all trying to get the best return for our dollar. And so you need to make a decision like, do I write the article, here's the best, you know, microphones, or do I write something that says, like, hey, Castos is unique in X, Y and Z ways, you know, and work on doing distribution for that. So I think the biggest thing is that content strategy shift, especially in B2B SaaS.
And I think from there, you know, my general view is that the search world is fracturing and that what AI has really done is cracked Google's monopoly. And what we're going to see is AI is a feature and everything, and that's a real different world and you need to, to think about that differently. So.
[00:13:15] Speaker B: Yeah, yeah, got it, got it. So it's almost like a, yeah, positioning manifesto type piece that you would publish.
You know, LLMs would kind of crawl and understand that to position you this way.
The other thing that I think about is like, as you mentioned, like, search is fracturing, the sources for ranking is changing too, right? Like, used to be your website was like the most important thing. And like, fuck Reddit and Quora and all these places, like, I don't really care. I want to optimize everything about my site. It's almost the opposite now to where, like Reddit probably is the most important place to get your brand.
That's kind of what I would say.
[00:13:53] Speaker A: In general, what you're describing is something we call siteless SEO, which is you pretend your site doesn't exist. What are you doing in the world that's going to get you, you set?
And you know, what really comes out of this is there was a technical SEO term which was the duplicate content penalty, which is that for all states, you don't want to have the same content on your site that's on someone else's site because you sort of split the value of that. And in the AI world, the number one ranking factor that we found is repetition. You just want to take the same message and repeat it and repeat it and repeat it. And you need to be strategic, strategic about where you distribute that. So here's an example. So you mentioned Reddit chat, GPT, OpenAI, Sam Altman used to be on the board of Reddit. They pay them million dollars of years for access, million dollars, multiple millions of dollars a year for access to the Reddit data feed to train their systems.
You know, same with Microsoft.
You know, OpenAI has a multi billion dollar relationship with Microsoft. So they can access to LinkedIn. It is 100% straightforward to say that they build off of those two data sources. So it makes total sense to rank in there Anthropic, which is a different system, which is, they make Claude, they're owned by Jeff Bezos. It's a different ecosystem. There's a massive laws right now.
[00:15:14] Speaker B: Whoa. Claude is owned by Jeff Bezos, one.
[00:15:17] Speaker A: Of the major investors. So.
[00:15:18] Speaker B: Wow, didn't know that. Okay, one of the major investors.
[00:15:21] Speaker A: And I'll get to another thing that I think you'll think is interesting. But you know, they're essentially barred from training on Reddit data. They're kind, there's a big lawsuit about, they're kind of scraping it anyway. But you know, if you're trying to train into that, not so useful. So there's some very clear, you know, lines about this and there's also a lot of non site, non search scrapers out there that are pulling information off the web from your site. You should still have a site, you should still do your best for it. But that, you know, it's very clear. Like googlebot scrapes my site, there's a, they're a search index. They're now going to have answers returned. People are going to click on it and come to my site.
There's a massive service called Common Crawl, which is, you know, a crawled version of the Internet that anyone can download via BitTorrent is used as the basis of a lot of these systems because they've been around for a long time and like, why wouldn't you use them? You know, if you're, if you're anthropic and you're, you've been tasked with like, okay, go scrape the Internet, the entire Internet.
[00:16:33] Speaker B: So that's a good starting point at least, right?
[00:16:35] Speaker A: Yeah. Like, why not start with 10 terabytes of common crawl data but there's no search engine attached to that. But it's very important you're in there because almost everyone is like, oh yeah, we're part of, we've used common crawl data. So. And there's a bunch of those that do the sort of specialty kind of work. So it's something we've started doing is like helping sites test whether or not these bots like Common Crawl can get through, whether or not these other like, you know, services can get through.
And it's a, it's a real different world again. It's much more fractured. Instead of worrying about Googlebot, we actually have, I think 24, we're going to launch another one today. 25 different bots that we check can access your site properly.
[00:17:18] Speaker B: Just to one last point on that topic, how important is this LLMs txt file on your site? We have robots txt just for folks who aren't familiar with it. Maybe tell me because I'm sure I don't understand it properly. Let's start with what it is and is it important?
[00:17:37] Speaker A: Yeah, let's start with Robots txt. So Robots txt is this text file you go to any site, you put robots txt on it, you can see it, it's publicly available and it's really a polite asking like hey, the website's asking, hey bots, come index the site. Don't index these pages. Definitely index these pages.
And we have a search study that's going to be coming out soon that actually goes into like here's the top like 4,000 sites from a bunch of different countries and how they're excluding different bots. And we certainly find, you know, massive differences in different areas.
Like Reuters is a great one. Reuters is a news agency. They're very against AI. They're very against AIs doing anything with their site and their robots. Txt has a bunch of exclusions that say like hey GPT bot like chatgpt get out of here.
LLMs txt I feel like was the wish of an SEO who's like I wish I could tell these LLMs like what to scrape and what to do and we're going to reformat our data so it's really easy for them to do. But to date no LLM producer trainer has actually said like oh yeah, we love this LLMs txt, we're going for it. That's what we use.
Kind of weird.
[00:18:54] Speaker B: And so what's your take on like I'm a whatever small kind of medium sized business do I do this? Is there a downside to it?
[00:19:01] Speaker A: I don't think there's a downside but I think it's at this point it is very low on the list of activities I recommend. And in part all of these services are if you look at the actual behavior of them, they are aggressively scraping your site in the HTML of it right now.
They are not concerned about taking that HTML and turning it into some text strings that they can do some work with. That's a trivially done thing that we have been able to do for 20 years without any issues.
So the LLMs txt just doesn't, you know, doesn't seem to move the needle is what I would say.
Again, I don't think there's any harm in it, but I don't think it's, it's vital it would be my take.
[00:19:45] Speaker B: So yeah, just for like a data point, we, we did it. There's a WordPress plugin that does it and gives you a really quite good one in like two seconds. And so like, yeah, I was the same way. I was like, probably not super important.
We did it in like two minutes and it's fine. Like it's pretty good. Like it doesn't, like we didn't do the whole, like I'm going to scrape the whole site and put it in chatgpt and like it just does it. And whatever the logic they have behind the scenes, like it is pretty decent.
[00:20:13] Speaker A: Yeah, I mean, it's sort of like a site map, you know, it's sort of like a site map for an LLM.
So here's something I would recommend instead for those couple minutes like, which is you go to ChatGPT or Claude or whatever service you like and you type in, hey, you know, here's my company, tell me about my company, tell me about the strengths and weaknesses, tell me about, you know, common objections and then check if you have written content that matches those on your site. Because, you know, again, there's a shift in what SEO actually means. What we found is that, and a customer told us this, which is they said Chat GPT is now our most popular and least well trained representative.
And people all day, every day are asking questions of Chat GPT that aren't really like SEO questions, they're just like, like, hey, you know, does. They're basically right, well here's like, does Casto support video podcasts? Like, you know, yes, you know that, yeah, go ask. Yeah, but. And it's, and it's these sort of questions which are like qualifying questions that are very low volume. And you know, in other contexts, you know, I talked again, a lot of B2B SaaS. The question's a lot more like, hey, is this SaaS? Are they HIPAA compliant? Because if they're not, they're out of the running entirely for Our procurement system, we're not even going to bother looking at them. And so you can imagine there's just an unending number of these questions for any business, like, do you have, you know, a location in California, do you do XYZ and this other feature as well. And, you know, getting that sort of general business knowledge to be plainly stated on your site and put in so it can be picked up, I think is a positive of this, is that I think that's good things to do anyway. Like, I think it's good to state more of those features plainly, and I think it's good to, you know, prioritize that in a lot of your social media, you know, we're talking about putting stuff on LinkedIn and Reddit to say, like, hey, you know, we do have a video feature.
Here's how it works, here's our thinking about it, here's how we put it in, and that's good all around. And I think that takes some of the sting out of this, which is like, it's not this whole other new thing to do that, you know, may or may not work. The behaviors and things you need to do as a marketer are a lot of the same sort of things. You just need to shift somewhat what you're doing in there.
And, you know, those are positives as well.
[00:22:46] Speaker B: Yeah, yeah, it is like, almost a little more clear. Right? Like, I don't have to try to futz with gaming the algorithm. I just like kind of say what I do. Put it on the website, put it a couple different places. Like you're saying, being really clear and explicit about it, checking to see that that actually is recognized, and then probably just keep doing that more and more.
Yeah, I have a question about the kind of ask ChatGPT, and then I want to tell you kind of how we're thinking about it and have you kind of shoot holes in it. It.
So I asked chat gbt discast of support video podcast. It said, Yes, I use ChatGPT all the time. So my question is, how much of that response is just in my account because it knows everything about me. So if you asked, would it be different potentially?
[00:23:35] Speaker A: You know, okay, there. There is a memory feature which is, you know, the way I view it is from time to time, it's a summarized set of all the past questions that you've asked. And that's how a lot of the AI systems work now because they have limited what's called a context window, which is just like, how many past questions. So they actually do this Technique of like summarizing your past stuff, trying to pull things out and in chat gbt, you can actually say like, hey, remember this for next time. Forget this kind of stuff.
[00:24:06] Speaker B: Yeah, yeah.
[00:24:07] Speaker A: So there is, there is part of that and you know, to be fair, there is a part of that that happens with Google as well, which is that if you search for like, hey, what's in the McDonald's near me? And you do that in Virginia beach versus Boston, you get a very different answer.
[00:24:21] Speaker B: Yeah.
[00:24:23] Speaker A: So there's a lot of other pieces of this where we do already have personalized search results in a lot of ways. So having a separate system like Noah Toa, that's, that doesn't have a memory can be useful in some cases.
But. But you know, for the most part, my personal take is that a lot of this is still pretty primitive compared to where it's going, especially with like Google. Google has a ton of information about all this because everyone uses Gmail and Docs and all these things.
[00:24:53] Speaker B: Yeah.
[00:24:54] Speaker A: So, you know, there's much more footprint to build a sort of Persona around in order to return this stuff.
[00:25:02] Speaker B: So I want to come back to that because like, I want to kind of think ahead about like what AI, Like a little bit more of like a macro level, like what's AI going to kind of do to us all? I want to talk about that because I really value your perspective on it. But to kind of take what we've talked about so far and this could be like a three hour episode, but take what we talked about so far and have some action items for folks. I think first get what you do on your website, which at casters I would say we've done that. We have an enormous amount of content. I think for the size company we are, because SEO and content has been our main thing so far.
Maybe make sure that's really clear to answer the questions that someone would ask in an LLM, not just for like keyword query based stuff for Google.
So redoing some content and maybe making new content cool, I would think of that not as a blog post but like as a page. Because like that might not make sense in like a blog post, but like this should just be a page and interlinking and all that kind of good stuff.
I would then basically do the same on Reddit and LinkedIn.
I would probably spin each of those posts to be kind of slightly different format or whatever. Kastos has its own subreddit, so I would just post most of it there and then LinkedIn as like an article on the company page, because those are public. Is that kind of 80, 20 of it?
[00:26:22] Speaker A: I think that's the, I think the 20 of it is creating the content and maybe the 80 is the distribution, you know, cool. Where, you know, so much of this is again, the repetition where I, you know, you have a feature, you want to describe it, you want to put it on, you know, the LinkedIn company page, you want to put it on your personal page, you want to have that described in a review. Like, and it's not, not in like the structured data way, but in just like, oh, here's another way for us to talk about this. This is an interesting use someone had of it. And you're just going to repeat this message over and over again in lots and lots of different ways that I think, you know, drive home, oh, we have this feature. It's good for these reasons.
And that's a little different. I mean, it's sort of the equivalent of like we were talking about, you know, the example everyone hates is the recipes. The recipes with like, oh, here's my. Yeah, yeah, I was a graduate student and made apple pies, you know, my.
[00:27:20] Speaker B: Grandma, blah, blah, blah. Yeah, yeah.
[00:27:22] Speaker A: Well, I mean, all that is there for SEO in the same way, you know, instead of that, you know, I think, you know, there's a LinkedIn post that's like, hey, you know, this stuff is good for these reasons. And you're just trying to find twists and ways to say that over and over and over again, which is a marketing thing, you know, a repetition of these messages.
[00:27:42] Speaker B: Yeah, yeah.
[00:27:44] Speaker A: So to, to.
[00:27:45] Speaker B: We, we did, sorry, YouTube as well, right? Like, yeah, gotta, gotta be on YouTube. Right? So maybe that's kind of the Three Horsemen is like LinkedIn, Reddit and YouTube. Every piece of content should be in those three places in some kind of way.
[00:27:58] Speaker A: Well, I mean, here's, here's why it's tough and why I say search is fracturing.
You know, it's not generally considered to be a leader right now, but like Meta's AI system.
So Meta, you know, owns WhatsApp.
WhatsApp has literally a Meta AI button on the home screen now.
Billion people use it. Do you know what it's saying about your company? Like, you don't think of WhatsApp as search. And you know, even Instagram, like, Instagram's product managers have come out and said, like, we are really rethinking search for Instagram where it used to be. You know, search for tags, you search for items. We're to do A lot more like solution kind of searches, like broader kind of stuff. And people have been doing that for a long time anyway, like, you know, we were talking beforehand and said like, oh, like my wife and I went to Boston for our anniversary closer to where you live. And you know, she didn't go to Google to look up stuff, she went to Instagram, like restaurants and restaurants in Boston. Yeah, and those are, those are commercial queries because we went and spent money there.
And you know, there's a lot of that happening already. And this is what I'm talking about. Like the fracturing of search is like AI enables this to search just to be a feature that's everywhere and we're just going to see more and more and more of that. We had briefly mentioned Jeff Bezos and Anthropic and all of that.
If you go to Amazon right now, there is a Rufus shopping assistant that's in there which uses Anthropic Claude behind the scenes to the extent that if you go to it and you ask like, hey, can you write a hello world application for me in Python? It will write a little one for you right there in the chatbot on Amazon site and then it will try to sell you like an intro to Python but book. And so this is an interesting thing where like maybe you don't care about Claude, you know, but you're, you're an E commerce seller. You have a feed of stuff that's going all the way out to, to Amazon and you know, where you used to be able to type, you know, products into that search bar at the top of Amazon. Now you can type in, hey, what's the best 10 for like cold weather conditions? If I'm trying to do this stuff and it will give you a reasonable answer. And you used to have to go to Google for that fracturing.
Seeing more search as a feature is just how I, how I view all this stuff. So you know, it's a, it's a weird time because we're in this transition, but I just see more and more and more of that happening.
[00:30:28] Speaker B: So yeah, you know, I heard, I lied earlier. I listened to three, I listened to three podcasts, but I just got turned onto this new one. It's called Unsolicited Feedback.
[00:30:41] Speaker A: Okay.
[00:30:42] Speaker B: It's the X Hub, Brian Balfour and they were talking in a recent episode about how it's maybe becoming slightly more clear like the positioning of each of these tools, right? Like Claude Anthropic, like for coding. ChatGPT, kind of a consumer facing app.
Gemini, really having Interesting, like context exposure to like how you work and then meta like for kind of social and commerce and that and that like they are increasingly trying to like put their stake in the ground to get the context and the exposure to you know, sources as like their defensibility kind of play. Right. Because otherwise like if you just assume, hey look, we're all just going to crawl the Internet, how is one going to differentiate from the other? But it's like the depth of context and exposure that they have, that could be the defensibility that helps them win in a market.
[00:31:39] Speaker A: It.
[00:31:39] Speaker B: Yeah, I don't know. So that you agree.
[00:31:44] Speaker A: I, I think, I think all, everything you said is accurate and transient. I was how I would describe it.
And you know, Claude is really good for coding right now because they have a Claude code CLI tool.
Literally this week, yesterday, the day before Google announced their Gemini CLI tool that does a real similar thing. OpenAI 2 weeks ago launched a similar thing and you know, on the dev side have certainly seen that these are very modular systems. You know that in the background, you know, we're using Cursor to do development, which is a customized AI first, you know, development environment. And it literally is a drop down. You say like, oh well I want to switch over and use Claude, you know, 3.7 because it's pretty good. Oh, Gemini 2.5 Pro came out, I'm going to switch to that. And literally it's this drop down switching cost.
Yeah, the switching cost is click, click. So you know, that's not, that's not a lot.
I think where things are more interesting is maybe the underlying kind of stuff which is, you know, anthropic I think has released to my mind one of the best pieces of research around, like what sectors are really using AI and for what they're using it for. It is right now a lot of dev techie stuff which you know, you can take for a grain of salt that if you're doing, you know, if you're selling dresses on, you know, custom made dresses on Etsy, that's a whole other business that's not as impacted right now but you know, it's moving towards it and you know, the areas they found that are the most effective are dev and then marketing and then healthcare and then you know, moving down the things and there's a lot of use for not exactly consulting, like therapy therapy kind of stuff, but is like a sounding board and generationally, the younger you are, the more likely you're to do this. And I see people doing this which is, you know, I have a slightly awkward social thing. Like I knew someone and they were trying to figure out like, hey, we invited these people to, you know, our thing. They did an rsvp.
So they asked Chad GPT, like hey, is it okay to send them a reminder? Is that like a socially acceptable thing to do? Like, like. And I was like, oh yeah, that's fine. And then like, all right, write me a polite, you know, thing for this and Chat GPT writes a polite thing and then they get a response back then, oh, it's totally fine. And that con. That's not a search query, that's not research really. It's this sort of feedback loop. And I definitely see a lot of people who, and you know, I have a lot of like pharmacists in my family, you know, for whatever reason don't know how that happened, but they're often in a position where they need to politely tell someone they were very, very wrong about something. Like, hey, you prescribed the wrong thing here. It could have killed this person if we didn't catch this. And they need to send that firmly, professionally and in a non anger way. And so they write it up as like you're a moron. And then they send it to Chat GPT and like professionalize this and it's. And it. And it does. And then they use that and to think that these systems that people are relying on for these very high stakes social things and advice and then they aren't going to turn around and use the same system to be like, hey, what pair of running shoes should I buy this summer? You know, like, yeah, it's, it's a different situation and I think we're just going to see more and more of that companionship and like, like aid.
And it's just going to bleed together with all, with what we think of right now as search is this thing over there. It's just going to be part of it.
[00:35:26] Speaker B: So yeah, yeah, yeah.
[00:35:28] Speaker A: Here's a question, Craig. Have you used any of the deep research tools?
[00:35:32] Speaker B: I use O3 and deep research all the time. Yeah.
[00:35:36] Speaker A: Okay. Yeah, because I mean those are very interesting to me and in particular like Google has a new AI mode which is like a deep research, like which is we're talking about different ways of thinking about things.
That is very much I have an AI. It is going to go out and look at 20 or 100 websites. It is going to read all those in real time, figure out based upon the memory that it has about me what I'm trying to do the goals I just gave it, evaluate each of those and then give me like a summary report. And there's some very high stakes things in there for marketers. You know, we mentioned the bots earlier and googlebot, Google now has a Google extended bot. And if you are blocking it because it's not the normal Google bot, and maybe you're using a WordPress security plugin or you check the bot Cloudflare or something, if you're blocking that, you will not show up in those results because it's using that user agent to make all those requests. Yeah, and that's a whole new world.
No one is looking at the site. You know, it's just bots. It's a site for bots at that point and it's a little scary. So.
[00:36:47] Speaker B: Yeah, yeah, yeah, it's interesting. In deep research, I do notice that the same few sites come up all the time or like within a question it'll go and ping a site several times.
So to your point, like, I think it probably has limited breadth compared to some other things that we do.
Okay, I want to ask like getting to the end of the kind of like, how are each of these platforms and models, like, where are they going like for, for us, like it's not too bold to say like, as a society, like, like I want to say like, where is this going? What are you worried about? And like, what do you think we're not thinking about, like, as, as like entrepreneurs and makers.
[00:37:35] Speaker A: I think the biggest thing that's hard for me to still wrap my mind around is the sort of implicit understanding of what Google is and does because Google is still the behemoth in the space. And for two decades there was sort of this implicit deal with the open web of like, we will crawl your site and then that's fine and we'll send traffic to it. And that has really changed, you know, and I think Google looks at the AI tooling stuff as a chance to make the open web search look a lot more like their other search products, to look more like local search, which is a lot more pay to play, their E commerce search, which is a lot more pay to play. And that, you know, from a Google product manager's mind, they're losing a ton of revenue by just sending this traffic out for free. And they've been slowly, you know, boiling the frog, you know, to use an analogy, for years. But this is, this is the big change where, you know, what we haven't seen in the AI search space yet is really direct advertising. Like there's been some brand stuff you can get into some of the AI search stuff by doing like basically opening up like what platforms you'll let your Google Ads be on. But there hasn't been like, oh, like, yeah, we're gonna bid 10 bucks for every time someone asks what's the best brand of pet food for my husky. You know. Yeah, like that. That hasn't happened yet and that's still a huge open question to me, like how much this could become pay to play in different systems.
[00:39:09] Speaker B: So yeah, yeah, I'll add one thing to that and then I'll add my, my answer to that question.
Because it's funny, like hosting a podcast. I don't get to talk about what I think.
I guess that's being a good host.
I agree. I think, I think everything will become pay to play as organic content is reduced to like the value of it is reduced because like it's so easy to create a shit ton of it. That is really good, especially on the written side. So I think like written content reduced, like the value of it reduced immensely. Distribution again, like you said, being the 80% of it, like just getting it in the right places.
I think like organic social toast, like aside from YouTube probably, if you consider YouTube a social channel, I think like you see it everywhere. X LinkedIn like reaches way down because the volume of shit is so high that they just have so much more stuff there that like you're pretty good post. Just can't compete just on a volume perspective. So like that's my take and I've heard other like pretty smart social people say that like it's just going to go to a pay to play if you for sure want distribution.
I think, I think the thing I worry about the most isn't like it's not that some vibe coder is going to build Castos in a weekend and will lose business because of that. It's not that podcasting is going to become irrelevant because I actually think we're pretty well positioned against this like artificial content surgeon. You know, podcasting and YouTube probably pretty resistant.
The thing I worry about is like 30% unemployment.
[00:40:49] Speaker A: Yeah.
[00:40:50] Speaker B: And I think like, I think it's as likely as not in the next year.
[00:40:57] Speaker A: I, I have a different take, which is cool. So like, let's just take marketing SEO. There are a lot of people that are very concerned, but I think the fracturing of it makes there be a ton more work. You know, like SEOs think about this as like, oh, well now I don't need to do this because it's saying, but it's evolving. So. And, you know, I've laid out in general how some of this stuff is evolving, but I do think there's a lot more to be done in just sort of making sure that the right message is out there on the right platforms in the right ways. And if you need to worry about five platforms instead of one, that's just more work. And if you need to worry about like, oh, the repetition of this message in a. In a meaningful, interesting way, a bunch of social stuff, that's more work as well. And so I think that's maybe a template for how this gets distributed into lots of other areas.
Because on the creation side, what I think AI does is the super form of like a grammar check or a spell check. Like, we don't have as many copy editors, like, you know, per person as we used to, but, you know, the average level rises. So if you need to stand out in business and do stuff, you still have to apply the human element to. So I feel like the. The floor comes up, but I don't know if the ceiling does, if that makes any sense, that you still need to have, you know, the human orchestrating it. The. You know, and that. That's really my take, which is that, like, it's considered the default that everything is going to be spelled right and the grammar is right in any document you write, because you can just throw the basically free tools at it, you know, to get that. That's not. That's not like a special thing.
Yeah, it's going to be a coherent, vague piece of writing that's going to hit literally the average. But it's going to take a human to really tweak it, to make it different, to make it interesting and to raise the level of it.
[00:42:50] Speaker B: I guess the question to play devil's advocate there is if we just admit that we're writing for LLMs now, does it have to be exceptional or is the repetition and the distribution of it more important than a single of piece. Piece being amazing? Like, does a single amazing piece, does it have more weight with an LLM rather than you saying a bunch of different times in a really structured way that machines can understand?
[00:43:14] Speaker A: So here's an example. You know, I'm trying to live this. You know, part of the reason I said yes to this podcast is I think, you know, I'm trying to build an audience that literally hears my voice and that, you know, I talk to people about this stuff.
But one of the things we did was we created a really Great piece of content earlier this year called the Pay Rank Report, which was started with a question of what is the impact of AI skills on SEO salaries? That was, that was the starting place. And what I found by looking through. And so I used a bunch of AI tools for this, like deep research, a bunch of AI analysis, a bunch of AI dev stuff. And what I did was I pulled job listings directly from corporate websites, not from LinkedIn or indeed, because that really pollutes the data.
And what I found was on average there's about a $15,000, which is about, you know, 15 to 20% rise in salaries if the job posting had AI skills on it. So if it was a SEO analyst and then an SEO analyst plus, you know, some chat GPT or do other stuff, that's worth 15,000 more to that company. And that was not exactly what I was expecting. But you know, it's an, it's an interesting sort of situation and I found that. And so that's where the study started. And then it sort of broke out to be like, oh, well, if you're an SEO, what are the other skills you need to learn? Like should you add in video, should you add in social? So should you add an email? And there's a different, you know, boost or negative, you know, for all of those video, for whatever reason was actually quite negative. So I don't know what that was.
[00:44:55] Speaker B: Okay.
[00:44:55] Speaker A: Yeah, but, and you know, there's a lot, there's a lot of caveats in that data, you know, like what is, is it a video person that knows a little SEO or is it an SEO person that knows a little video? Like those are two different things, you know, so, but it's interesting to look through the data anyway and you know, that's, that's literally where it's at right now.
[00:45:15] Speaker B: Yeah.
[00:45:16] Speaker A: So, and there was a trend to be less juniors, more senior people, people more like orchestrators, conductor kind of people, more people doing lots of different skills and things. And you know, that's, that's both sides of it that, you know, I don't know if it's fewer people making more money, more junior people, or those positions don't exist and people just jump in at a higher level because they can, because they're using.
[00:45:42] Speaker B: Yeah, you know. Yeah, yeah. No, I mean, I think, I think just set up at a macro level. I think I do see the middle disappearing. Probably companies, employees, people will be negatively affected and laid off and have to go kind of do whatever and then some will be successful and kind of elevate I don't know, it just seems like AI is driving the wedge pretty solidly. Yeah. Interesting. So, okay, I want to wrap up folks who want to check out Noah Toa and later, like, kind of who, who, who would it be best for? And like, what should they look to do there?
[00:46:20] Speaker A: We have a free offering. You go to noah toa.com, you enter your website. It will tell you like, hey, here's the bots that you're blocking. You're not. That's just a great place to start. You know, they can't reach your site. You're not going to be able. Anything you write isn't going to be influenced by.
So start there and then you can check a few things in chat, GPT and then we branch out to the other services and stuff. We also have this thing called the Biscuit Framework, which is not specific to us. It's just formal version of basically our talk today, which is, hey, here's how you should think about some of these things. Here's how we recommend shifting your content strategy. Here's some, you know, different aspects of this that you should consider moving forward. That's free. There's not even a registration for it. It's just a giant blog post.
[00:47:04] Speaker B: Awesome.
[00:47:05] Speaker A: So please check those things out. I'm on LinkedIn. I'm happy to chat with people. They can also just email me mike@noahtoa.com so awesome.
[00:47:14] Speaker B: Awesome, buddy. I appreciate it. This is super helpful for me and I'm sure it will be for, for everybody too. Thank you for coming on.
[00:47:19] Speaker A: Well, thank you for having me. This was a delay. It was great catching up. So awesome.
[00:47:23] Speaker B: Awesome. Cheers, buddy.
[00:47:24] Speaker A: By.
