The Digital Mind: Have We Outsourced Human Intelligence? | Jeremy Tang

00:00:00

targeting whites space. It's largely uncontested. >> What's your playbook? How do you take over all of their traffic? >> Cuz their traffic from an organic perspective is not even where the main game is. You build a platform called CMAX, a programmatic SEO platform. You see about 60 million visitors through innovative AI processes. You're right at the forefront of AI. >> The number of queries that Google sees almost doubles every 7 days. 15% of the searches they see every day, they've

00:00:30

never seen in their history. Brand new searches that have just come out of nowhere. Cuz a lot of people have assumptions about this is the only kind of people that I target and not realize that potentially there are adjacent opportunities that you never bothered exploring. >> You make it seem so simple like look at all the things that no one else has answered. >> Content momentum equals organic performance. It's a really a volume. more you put out there, the better you're going to perform on search

00:00:59

engines that will then translate impressions and then click and secondary to that is >> what is the first step I should take to take advantage of the shift that's happening. >> I think first is don't get distracted by >> this episode is brought to you by Wick Studio. Here at the agency podcast, we're building a community and we would love for you guys to be a part of it. So, we would love to hear from you. What are you enjoying the most? What would you like to see more of? And what do you

00:01:32

think might be missing? Drop a comment. Make sure you subscribe. And now, on with the show. The reason I've got you back today is I want to unpack how the internet is changing. And I know from your perspective with what you do day-to-day with your agency is you not only had been in the SEO game for about two decades, you've built a platform called CMAX, which is a programmatic SEO platform, but what I'm finding really interesting is that your systems that you're plugged into and how you're

00:02:00

looking at AI, you see about 60 million visitors through innovative AI processes and you're right at the forefront of AI and where the technology is going. So for someone listening to this right now thinking to themselves, where is the internet? Like where are we truly at right now? What's going on? And how can we get up to speed with how we should be thinking about platforms like Google, Open Claw, ChatgBT. Jeremy, I want to understand where are we right now in the timeline of AI and how should business owners be thinking

00:02:33

about their relationship with AI and the internet? to kind of understand AI. I think it's important to understand there's really three different flavors of AI. And the AI that we're most exposed to, the most familiar with is really, I guess, the the beginning stages of where AI is supposed to be. And if I was to use technical terms, and I'll try to put in a more layman's term, right now, most people's exposure to AI through chat, CPT, etc. uh is called uh probabilistic AI. It's a or I would term

00:03:10

it correlation AI. So what it is is they're engines that go out and find relationships or rather observe relationships and then based on those relationships they spit out answers that seem to kind of like be in line with those. >> And when you say relationships you're talking about data on the internet. >> Yeah. just basically correlations like I see this word is usually paired with this other word so therefore that's what I'm going to serve up because it's likely going to be the combination

00:03:40

you're looking for. So that's pretty much how uh most AI that people are experienced with works. And then on the other spectrum of that is really what's called deterministic AI which is less about correlation but more about actual rules. You know if you do this then such and such. It's it's a lot more definitive and I would even argue to say that's really not some people use deterministic AI but I would actually call it more rules-based engines cuz typically a integral facet of AI is this kind of

00:04:18

correlation this randomness aspect of it which is why it's able to make connections that are typically hard for other systems to do or even people to do. So you got these two extremes, probabilistic which is very much correlation based and then you got deterministic which is rulesbased. So what we're seeing right now is we're still in this kind of correlation phase and in order to scale we really need to get to deterministic and I think there's like there's a lot of technologies out

00:04:48

there that are getting there but a interim step between that is what I call causal AI which is where we need instead of just relying on correlations etc. What we need to do is we need a mechanism through which we can run experiments to test those correlations to ensure that actually there is a causal link between what we're observing. You know a classic example of what a correlation is. Say for example when ice cream sales goes up drownings go up as well. So a standard AI would go, oh, it might draw the conclusion,

00:05:29

okay, we got to watch those ice cream cells. Like, you know, it's causing drownings. Whereas, of course, you need to kind of drill down a little bit more on that in order to interrogate it, test hypotheses. And then based on that, and once you've tested it, then you can get to deterministic AI, which is rulesbased, where you've actually found real insights that you can repeat. And the reason for that is uh the what is holding back the scalability of AI is the hallucinations. I mean it I think it's underappreciated.

00:06:04

It has very low consequences and yeah maybe it gets a word here or a fact you know off not a lot of consequences but when you start doing more you know serious transaction maybe financial transactions or you're making important decisions or uh or you're making specific recommendations for clients that becomes quite problematic and there's a lot of data out there you know some people range between 2 to 10% is a typical hallucination rate but the problem is is that let's say it's 5%. An

00:06:36

average and we use a like an everyday analogy. Imagine you know a 5% error that means one in every 20 times it's off. Imagine driving a car and great if you're say traveling at driving like 5 km an hour and one in in 20 times your brake doesn't work. Okay, maybe you can you can get away with it. It's okay. But imagine you're a Formula 1 driver and you're trying to go really fast. Imagine one in every 20 times you use the brakes. It doesn't work. That's a big problem. And I think that is where

00:07:11

AI needs to head or where it is heading rather and where business owners need to be very aware of the current limitations of the scalability of conventional AI tools and need to progress quickly into what I call causal AI i.e. being able to run experiments, test hypothesis, and then get the data before building rules-based engines, which is pretty much traditional software. And I think that's why like systems like Claude, etc. are kind of like a an indication of that trend now >> because that's what it's doing

00:07:49

essentially. It's got that correlation aspect to it but it's also able to use uh deterministic infrastructure like maybe using at a basic level Excel to run calculations etc. So I think this is to kind of navigate this I think it's first of all really important to understand these kind of three tiers and the surprising thing is actually this structure is pretty much how human civilization has evolved for the last 3,000 years since like Egyptian time in Mesopotamia. The process the threepart

00:08:23

process that I'm talking about is actually the the scientific method. It's pretty much how as humans evolve. They've observed something, right? Some phenomenon. Then they apply the scientific method to it. And then they go, okay, well, why is this happening? Why is X happening? You know, what are the inputs Y that's causing X? And then finally at the end of it, then once they've interrogated it and they have proof that it's repeatable, then it becomes scientific law and then it gets

00:08:54

used. So I think if you use it at that level, you'll understand where it's heading and I guess as business owners where we also need to head if we're going to scale AI, which is I think the most uh the biggest pressing issue at the moment businesses. >> Something you've been talking about recently regarding this and I think this is going to help the audience understand the lay of the land regarding >> who's controlling what. You called Google the master index. >> Yeah.

00:09:26

>> What does this actually mean? Like how should people think about Google in relation to what you're just talking about now and and the future of their business? >> Absolutely. Um, when we chatted originally last year, I already had a quite a strong inkling that AI search and traditional search engine optimization were there was like at least an 80% overlap. But since we spoke and particularly there have been two dates that come to mind. There's July last year and September around September 10th last

00:10:05

year. They're two critical dates which made us realize that in fact it's not a 80% overlap. It's actually 100% overlap. >> Can you explain what you mean exactly? >> Exactly. >> By the overlap >> 100%. >> It was never clear about okay well AI particularly let's when we talk about AI let's talk about chatd they've got a huge market share. when people are searching on it, no one really had full clarity about where it was getting its information. It was

00:10:36

incredibly ambiguous. There was theories, but there was no proof around those theories. >> I forgot about this. So, so at the very early point in the market when ChatGpt was being exposed as this amazing platform, >> people you're saying at that point in time had no idea where it was getting its insights from. Well, what it was doing, you'll me like at the start what was happening. I'm not sure if you remember when you searched a, you know, a fact or information, it will give you

00:11:02

a warning like, oh, please note our information is only up to date up till 2021 or something like that. It's basically its index and it and they quickly realized that in fact that was a bit of a weakness because people wanted more current information. So what did they move to was a process called rag ra a which is retrieval augmented generation and what that meant is that they would instead of trying to be the information source for the queries it would pull information elsewhere and use its AI to

00:11:37

summarize the insights from that process. And during this period up until July when they transitioned from be their own search engine so to speak with their own index to using external data sources um up until July like no one really understood where it came from. A lot of theories was Bing because Microsoft is a you know a shareholder of um open AI and you know and it makes sense as such. But what was really interesting, it was reported across the industry in July that there was a huge spike in

00:12:15

impressions across websites around the world. Like a significant spike impressions, but not a corresponding spike in traffic as a result of those impressions, like very dramatic. and further investigations in the industry such as okay, I'm only going to uh show content that only Google can index and therefore and and see if it surfaces up in a GPT search. Test like that confirmed that GPT from July 2025 started to use Google for its information retrieval and that's why it was such a increase

00:12:56

and the reason why it was such a large increase is based on the underlying mechanism by which AI search works not just GPT but pretty much all AI by engines pretty much and it's a process called the query fan out. So what happens is that when someone puts in a question in AI, so for example, I'm looking for running shoes for beginners. What it will do is it will interpret that question and then split it out into a lot of subqueries, maybe 9 to 12 sub queries. It's called the fan out. And

00:13:37

these are effectively long tail or longer tail searches that attack that same question through multiple angles. And what GPT and other AI engines do is then they search on the data source which is Google find the results out of all that summarize the information and then give you the answer. And this is the underlying reason why AI answers seem so rich and so good is because they don't just take that qu. It's not like I search for running shoes for beginners and then it just simply goes and

00:14:16

searches on Google or Bing and then gets the results and then summarizes it. It doesn't do that. It does way more than that which is why the results are a lot richer. And so instead of say running shoes for beginners might be running shoes for beginners with flat feet, running shoes for beginners for the gym and treadmill. A lot of different kind of variables. You can imagine how that works. And that's really the underlying mechanism and that explains why there was such a pronounced increase in

00:14:48

impressions that were being seen by websites around the world without a lot of corresponding clicks because it was purely for information retrieval purposes. Now the other date that was really interesting is that Google saw this huge spike. You can imagine now one question actually translates to at least 10 to 12 sub queries. So it's it's amplified significantly. So they try to shut it down. So se around September 10 what they did is they changed their rules around what level of results

00:15:22

would they serve when someone >> So you're talking about Google's >> Google um making a conscious decision to not give the data to these other platforms. >> Well, it's not necessarily not give the data but make it harder because previously in Google, you could have searched for any term and you could see up to 100 results at once. >> Yeah. >> And this is what a lot of >> Just like when you go on Google yourself and you type in >> cute photos of puppies, it would give

00:15:50

you a 100. >> Yeah. You could get up to 100 results. >> Yeah. And then Google realized hang on a second this is getting absolutely ridiculous. All these particularly GPT was you know manipulating this fact and able to scrape a lot of data from its results. It restricted it to from 100 to 10. So a dramatic decrease. And at the exact same time that that policy change was implemented all the GPT results started to change as well. at the exact same time, which then leads us to conclude that in fact GPT

00:16:33

is essentially doing Google searches for you. That's essentially what AI is. And Perplexity to a lesser degree, they're obviously not as big, but if we focus on like GPT as a proxy for AI, that's essentially what's going on. when you search on GPT, you search for something, it goes does the query fan out and does those fans out in Google and essentially it's summarizing the results for you. And so that's where the overlap became not maybe 80% but pretty much 100%. So it's quite significant in insight.

00:17:13

There's no there's no real difference between the discipline of SEO, search engine optimization, and AI optimization because it's pretty much the one data source, and it's unlikely to change in the foreseeable future. If you want to grow your agency, you need more than just a website builder. You need Vick Studio, the all-in-one platform that's built to help you scale. Design smarter in a hyperflexible, responsive editor, cherrypicking whatever clients need from pre-built UI

00:17:50

elements and templates. Deliver robust backends with built-in e-commerce solutions for every industry, letting self-maintaining infrastructure just run itself. Stop reinventing the wheel with every project and instead create a sharable design system with reusable assets, apps, and components. Clients growing too fast? No sweat. Wick Studio AI powered CMS lets you turn a single layout into hundreds of additional dynamic pages. It's true. More clients no longer has to mean more chaos. Wix Studio gives you one centralized view of

00:18:26

every project and makes real-time collaborations seamless. Do your agency's best work at scale. Build your next project on Wix Studio. When you think about this as a concept, what what does that mean to someone like yourself who's working with some of the biggest businesses >> around the world and their relationship together with Google and AI and how they show up in search terms? when you're looking at this and what Google's doing, what do you think? >> Um, this is not investment advice. I think

00:18:57

Google's in a I think Google's in a very privileged position in the uh AI value chain particularly in relation to search information etc. And I think given that it's really important well I guess from a marketing perspective first of all for businesses is to not listen to the rhetoric that's out there about this separate discipline of uh answer engine optimization genative engine optimization AI optimization etc that they are you know one and the same thing but to I guess realize is that it's all

00:19:43

just an interface to the same underlying data. So, I think it's um it's incredibly important that people don't get distracted by any information that contradicts that in terms of okay, well, there's a very separate strategy to have a greater presence on AI versus on Google. It isn't. It's actually pretty much the same thing. Maybe with a slight tweak on how they summarize, but it's essentially the same thing and it's it's definitive. All the data points to that.

00:20:16

>> When you look at these different AI apps, you're essentially looking at different rappers for the same system. >> Yeah. Essentially, yeah. when it comes to search like when you're trying to in obviously there are uh you know like um in terms of data analysis all that not so relevant but in terms of where information retrieval is needed Google is in the best position I mean I think it's really telling that uh you know GPT started using Google for its information retrieval considering its close

00:20:47

relationship with Microsoft and it would have made sense to use Bing But for whatever reasons that we shall never know, um, it hasn't translated, they've decided to use Google instead. And I think it's kind of like it's it's telling about, uh, I guess the I guess the perceived quality of the information coming out of Google. And also it just its level of access is unmatched. I mean, they had a uh I can't remember exactly when, maybe it's a couple months ago, but you know, Google had a um

00:21:25

signed an agreement with Apple because they Apple have stepped out of the AI race and said, "Look, we're not going to build our own AI anymore." And they decided to partner with Gemini. And now you've got one system that has visibility over pretty much almost every mobile device except the ones some out of China, but the majority like in terms of Android and iOS and you can just see all the data out there and it's all kind of converging into this one point which is Google. So, it's really important to

00:22:00

keep an eye on what's going on there to really know where AI search is heading is really Google will dictate a lot of that. >> When you think of these different apps, you're saying a part of it or a large portion of it is informationational retrieval. >> Yeah, it's critical. >> What else is a platform if it's not just that one component? I think it's layers of so the ways that these kind of different apps can differentiate themselves is in relation to the specific area of information that

00:22:37

they retrieve. This is where I think where the opportunities are rather than generalist apps that kind of help you find sneakers etc. But you see um along very specific verticals. So there are a lot of AI startups that are kind of in the medical space or in the legal space, maybe the accounting. So very verticalized. So where there's very specific information, that's where um these kind of businesses will be more decoupled from I guess the Google >> ecosystem and be able to I guess succeed

00:23:15

independently of what Google does. And um I think that's where the the opportunities are in getting proprietary first party data and being able to access it. But also there's the extra logic around how you process that information. As I mentioned, most of these um systems are built on finding correlations between information. But the way that these systems are evolving is they're building a causal AI layer which is experimenting to see you know cause and effect and then >> and then translate that into kind of

00:23:52

rules-based engines. >> How do you how do you then see the difference between Gemini Chat GBT and these other different platforms like Claude? Do you do you see much of a distinctive difference between what Gemini are doing together with Google and Chat GBT for example? Um, I think each of them have their nuances and Claude appears to be more in terms of as a business or a personal productivity tool rather than necessarily a great place to find, you know, information on X and Y. I mean, perplexity seems to be more in that

00:24:28

position in terms of being able to get real-time data and uh and synthesize that. Do you think that if Gemini has, >> you know, if Gemini is Google, does then it not have a structural advantage over platforms like ChatBT? >> It's interesting. You would think so. I I believe they're still working on it. um like from personal experience as well as external benchmarks, Gemini has not tended to perform that well from a just standard query style functions but I have a feeling that it's just a

00:25:07

matter of time like they can play the waiting game they basically control all the data so at the moment I believe you know tools like GPT perplexity etc are better able to synthesize the information and provide >> the insights out of it. But I don't see that being the case. >> Well, that makes sense to me because if Google's making a play here for the long game, it means that they're going to outlast a lot of these startups by, you know, remaining as the default master index like you're saying.

00:25:43

>> Yeah. I mean, the whole the caption is, you know, like when Google sneezes, AI catches a cold. like it's like I think Reddit's a really good example of um you know just how linked you know uh all this is prior to uh the September incident where Google started shutting down kind of like you know the results you could see Reddit was considered almost a a proxy for AI because it's heavily cited >> like uh whenever you searched on Google lo uh sorry searched on chate BT Lo and

00:26:17

behold, the citations would be largely around Reddit. So that's a and to such a degree that people were saying AEO GEO is pretty much Reddit SEO. If you can optimize your presence on Reddit, then you're good. And everyone started to assume or infer that okay well GPT and these engines must have a intrinsic bias towards Reddit >> and as a result of that you know I think uh Reddit's valuation went up >> and you know I I can only imagine that if I were trying to build chatbt or claude right

00:26:58

now if we were dependent upon Google and Google are trying to block us out of the platform you would be thinking about what is the alternative pathway. >> Absolutely. >> Yeah. >> Absolutely. >> Is that creating friction in Silicon Valley with these different companies right now? >> 100% it would. And and to the point like my Reddit point was like when September occurred and the results shifted from 100 to 10, Reddit dropped traffic dramatically. They lost $3 billion in

00:27:25

market cap and they haven't recovered >> because of that. because of that and a lot of people's and the only reason people originally and this is where the the the challenge with I think I mentioned a bug bear I had with the SEO industry which is now the AI industry is like it's very superstitious a very superstitious uh industry they say oh because Redd's showing up here okay therefore you know GPT must love Reddit for some reason and what we uncovered because of that after that September

00:27:56

update is that no, it has nothing to do with some sort of, you know, intrinsic preference. It's simply because Reddit appeared in the query fan out. >> That's >> Can you explain the query fan out? Like what what exactly you mean by that? >> Yeah, it's essentially translating a query that we have, a question that we have in the AI and translating that into a lot of sub questions. >> Okay. Can I use an example here and maybe you can just correct me if I'm going in the wrong direction. So, if I

00:28:26

were to say, you know, pre-AI dominance, if I were to go on Google and search a car, >> I might just type in cars for sale. >> Yep. Possibly >> as an example. Y >> now people are no longer searching like that. They might type something in like, hey, I'm looking to buy a BMW M2 with SA Paulo yellow uh with the CS engine and I need to get it from a dealership near me. Submit. >> Right. Would you say that that difference in prompting has shifted the entire way the internet operates and

00:29:02

would you also say that both of them create two very different service queries? >> Absolutely. So I think there's this two phenomenons we're talking about. So in terms of what you're referring to in terms of that richness of of search behavior um and that's what's called an extension of the longtail of search where people are more comfortable uh searching for information like nowadays nobody just searches for new car when they search for a car you know or new shoes it's a lot more specific and I

00:29:33

think AI has helped with that but probably more so a mobile and voice search because now you just speak it and So, so that's one phenomenon that's kind of like uh extending out what I call the longtail. Like I mentioned previously that, you know, 90% of searches are are long tail. I think that's kind of expanding continuously. I mean, Google released a a um a stat where they said 15% of the searches they see every day, they've never seen in their history. Like brand new searches that have just

00:30:05

come out of nowhere. >> What What exactly do you mean by that? How can I picture that? >> Uh well, all all they're saying is it's a >> out of a 100 queries that I see in any given day, >> yeah, >> 15 of them have never been searched before in the history of Google. >> Like they record >> all the queries that have come through >> and 15% every day are brand new. >> So I want to make sure I'm understanding this right. when I search something on Google,

00:30:37

there's a massive database that is contingent upon this has been asked or inquired about before and then now every day 15% of those queries >> have never been asked before. >> It's not contingent. So there's no reliance on the fact that okay uh it's Google still works the way it is. It's just what they're doing is they're recording all the queries that are coming in. >> 15% of all queries every day now are new >> are completely new. they've never seen

00:31:04

which which speaks to the evolution of how people are searching like it's expanding and typically those searches are more specific much richer than historically like where you know people might have searched you know the start of the internet oh new car on new shoes or something like now it's like the awesome query like a BMW etc >> like uh and if you kind of like extrapolate those numbers out that means the number of queries that Google sees almost doubles every seven days. >> So it's expanding

00:31:39

>> exponentially >> dramatically. Now you add on top of that the concept of the query fan out and to explain that in you know using analogy of say uh what it would do is it would translate that and work out okay well what are the different like dimensions through which I should be searching that would be useful for the individual search for a new car. So, it might be uh new cars for a large family, a small family, large boot um under $20,000 or in my local area or it basically tries to anticipate the likely intent of

00:32:24

the person and find all the information they're likely going to need to help make an informed decision or progress a decision. It's essentially distilling what we would normally do. Like, you know, if you're on a journey to buy a new car, you're not just going to search once. You're going to go, "Oh, okay. Search that. Oh, then I'm going to okay, based on results. Actually, let me let me search a little bit more specific." It's about anticipating that journey

00:32:51

that people take and doing it all at once. And so, that's really the the the query fan out process. Now you tell your clients to map out the exact subarches around these queries uh that the AI will run. What exactly do you mean and how can people take advantage of this change to start I guess uplifting their brands, their companies to to be seen by the AI platforms? >> I think it's full of first of all a recognition of where the opportunity is. >> Where where do you see the opportunity

00:33:22

in all of this noise? I think it's it's it's not where most people are focusing and that's the you know probably the the headline point where everyone is still you know looking at this from a particularly you know AI search from an SEO lens which is let's go after the big terms you know people search for like whatever the the used cars the new shoes etc >> and so just to make a distinction there you're talking about the short key phrases or key They call head terms

00:33:54

which are top of mind, easiest to recall and they >> for me it might be like brand agency identity design, logo design, just those key phrases that might be commonly used. >> Absolutely. And so that's kind of like where a lot of the focus is. But now you've got this phenomenon like before all a before AI it was you know that represented maybe 10% of the market typically less. But what's happened is is with the voice and AI search with the increasing evolution of how people

00:34:25

search. you know that longtail is expanding and now you've got another mechanism in play which is with AI which is essentially 10 to 12xing the longtail because AI doesn't search using head terms it searches using longtail terms where you've got a situation now where in order to be present on search engines and AI engines you actually need to have extensive coverage across Ross these smaller terms. It's no longer enough to just go after and it never was before, but even more so now. Just going

00:35:02

after a couple of, you know, keywords or topics here or there. It's just not a it's not the game anymore. It's all about surface area. So, I think that's really what people need to understand and um and recognize first and foremost is that the game is about lots of content. It's just content momentum is the absolute key to it. >> Okay. So, I want to dive into content momentum in just a second >> just to wrap a close a couple loops here. When I'm thinking about the Google

00:35:37

relationship with Reddit, >> why was it that Reddit was showing up so much when Google was showing those hundred listings? >> Yeah, absolutely. >> It's because Reddit as a discussion forum covers a lot of topics unlike maybe uh you know a website like car sales which is cars. Okay. Reddit covers everything >> cuz I think of Reddit more as the town hall meaning >> exactly >> people are on there just absolutely going on rants. They're talking about pop culture. It's very opinionated. It's

00:36:13

>> I can imagine it like you get 10 million humans together, they're just going to talk about everything. Whereas Google seems to have been attached to businesses trying to position themselves to fight over certain key phrases. This is how I differentiate it in my mind. Well, the reason why Reddit was appearing is uh in spite of all the you know, however someone might assess the you know, the nature of the content, etc. There there's so many subreddits covering pretty much every topic out

00:36:41

there. And as a result of that, Reddit pages were appearing on search results, extensive range of search results. They were appearing in the longtail, all these longtail queries >> because they just have the content subject at scale. >> Yeah. whatever random topic you can think of, I guarantee you someone on Reddit's probably complaining about it. >> So if I collect Smurf dolls, there's probably an entire subreddit with thousand pages of content around how to collect Smurf

00:37:10

>> dolls. Absolutely. And because Reddit was appearing in this fan out process uh and when you appear extensively within a fan out say for example you have you do 10 searches you get all the results and you see a certain brand or an entity or an idea etc surface across them. What it does is increases what's called the citation probability increases the chance that GPT at the end will then surface that idea or that brand. >> Okay. surface area. >> Yeah. So, it's basically Yeah. If I So,

00:37:43

I look at 10 search results and I notice, oh, Reddit's here, Reddit's here. Oh, multiple times, then I'm probably going to trust Reddit as an entity more >> than others. So, it's really the only reason it wasn't because GPT had an ingrained preference for Reddit. It was rather it was just appearing in the fanouts. >> It had the most citations. >> The most Yeah, it was appearing the most time. But then when the results went from 100 to 10 and they kind of like

00:38:14

restricted even more then those that presence almost disappeared or was reduced significantly as a result of that all the citations changed. >> So forget about this idea of Reddit SEO is AI SEO etc. No it's because of the presence on the query fan out and that's what I meant about cause and effect. A lot of people were using correlation to assume that it's uh you know causation on why you know it's appearing on why Reddit was appearing on GPT. >> But we realize now that has nothing to

00:38:51

do with some preference. >> So how have these AI platforms worked around this issue? If if Google's trying to >> act as a gatekeeper from them getting to as much information as they could once upon a time get access to. Y >> where are they going now? Are they going straight to Reddit? They're trying to access their own proprietary data sets. I mean, that's the only way. If if you're a generalist AI platform that's trying to just cater to the masses and their information

00:39:20

needs, etc. I think it's going to be really hard. I I don't see how you can surpass the level of access to data to information in a real-time basis above and beyond what Google's able to do. Okay. >> Unless you verticalize like all these ones have to verticalize in order to >> I see what you're saying. So like one AI platform might go all out into the medical space and have all the medical books, all the doctrines, >> uh all of the peer-reviewed studies. Okay. When you talk about Google, you

00:39:51

talk about how Google spends$ 26 billion dollars a year just to be the default search engine. How are they doing this? And and why is their key strategy focused on owning that position? I think they just understand that, you know, the entity that controls the data or has the best data is the one that's going to win in the long term. The underlying models themselves um I think it's getting harder and harder to differentiate them. What differentiates them is the the training. >> You think about like human beings etc.

00:40:34

like what often diff like obviously we have all else being equal we have you know innate traits and you know basic level of intelligence etc what differentiates one person another is the level of training that they have and the quality of the the data and I think that's where and I'm sure there's other you know uh monopolistic kind of kind of kind of objectives but from an AI perspective as said they've really locked down they've got the they've got the browser they've got the devices

00:41:04

devices both you know iOS and Android um and you know with YouTube all that they've pretty much and they've got data you know um data pipelines from a lot of you know from key social media networks >> plus they have access to everyone's emails we have >> of course >> terabytes of data that we have on our Google drives and we recently read somewhere that they have access to all of that so everything we've ever uploaded >> emailed etc. has all gone through the the Google

00:41:36

system meaning that they have certainly a monopoly over data. >> Absolutely. They are in a privileged position. This the only way to compete is niche down. You have to Yeah, that's it. >> Another data point that you cited was that chat GBT traffic grew 52% but Gemini grew 388%. Does that then mean that Gemini is taking over? I think in the long term for the more general AI space as said not going to comment on the the vertical ones. >> Yeah. um everything points in the direction of

00:42:12

they are in the best position and on top of that they also have the monetization a engine behind that because you know AI is very expensive to run and what's interesting about Google or Alphabet rather is that they're also vertically integrated they also do their own chipsets hardware data centers etc which gives them another advantage over other players in the market. So there's just so many modes that they've established at the foundational infrastructure level all the way to the

00:42:49

data level >> and then also the monetization level that it's I I I haven't seen any player out there that has a more compelling stack that you know just um yeah I'm still grappling with how how it would be possible for >> to me it sounds like they not only own the farms but they own the factory and they own the retail stores and all the sales reps in between. >> Absolutely 100%. And um yeah, scary, but that's just how it's alling. >> So, if that's true and I own a business,

00:43:22

what is the first step I should take to take advantage of the shift that's happening between all these different platforms? >> I think first is don't get distracted by strategies. Go, oh, I'm going to optimize you for chip. I'm going to optimize you for perplexity. I'm going to optimize you for X and Y. I think first is to realize that need to focus on you know on Google on the principles of what it takes to perform on Google because if you do then you're going to be well placed on

00:44:02

the underlying data source that these third party AI engines use but you're also going to be well positioned for obviously um Google's inhouse house AI as well which of course uses its own search engine for data. So I think that's the important thing. I think there's a lot of news and marketing out there that's you know people selling ah we're going to this special strategy here special strategy. I think that's the first thing is to really simplify it and not worry about that. It's really

00:44:32

the game is really how do you improve your presence on on Google. I think that's that's >> it makes sense to me because I I use Google Chrome. Yeah. >> For everything as well. And I'm just thinking about >> the fact that most people I know don't use Internet Explorer anymore. They're either using Safari or they're using Chrome. And if >> Apple are partnering with Google, uh then that means that they're probably going to be able to site >> or citate what happens on Safari. I

00:44:57

don't know. I'm just guessing here. I'm clutching. you're saying as an as an entrepreneur that they should first and foremost be as deeply connected to Google as possible from a source of data so that you can show up on these AI queries. >> Absolutely. focus the strategy on just on pretty much SEO principles, you know, and make sure you understand the questions people asking and there is said there, you know, there are different tools out there that can give you insights into that and they're tools

00:45:24

that everyone has access to to to get visibility over how you're appearing. >> Yeah. >> And it goes back to the same principle that I mentioned before, which is this game of AI is the same as Google. It's just a question and answer game. find out the questions people are asking and give them answers in the form of content. >> How do I find out the questions people are asking? >> Um there are a variety of tools out there. Um if you're using Google Ads, there is a Google Ads keyword tools.

00:45:52

There's a lot of other uh thirdparty tools out there like um SCM Rush, um Uber Suggest, you know, a bunch of different uh tool sets out there, some free, some paid, some have free tiers from which you can, I guess, get an idea of okay, you put in a a head term like maybe it said new shoes, and it'll give you all the different derivations with which people are searching for that. So, there's a lot of um tools out there available that can that can help you mind that. And >> where where's the simplest place people

00:46:24

could go? Let's say someone's starting a new service-based business and they're doing consulting around >> something like they're a personal trainer. >> What platform would you suggest that they go to to find out what people are looking for? >> Okay. I think um yeah, you've caught me out because I only use uh enterprise tools. Uh in terms of I guess it would probably go into free tools and I believe like Uber suggest is a tool that uh can surface keyword level data. Okay.

00:46:57

>> That I believe you can do it for free live data around what people are looking for. >> Yeah. I I I can't really comment to the level of freshness of the data but it definitely can give you >> a better appreciation of the variety of terms people use around a given topic area. I think that's the critical thing. And then at an uh at more of an enterprise level, you would say SEM Rush. >> You could use like SEM Rush. You've got the Google Ads keyword data tool. If

00:47:21

you're advertising with Google Ads, you automatic have access to that. That's a that's probably the highest quality data you can get. And then there are a slew of other kind of tools that you can do like ah refs is another popular tool. So I would say Google Ads keyword tool, SCM rice, ah refs. Could you interface with Gemini and have that spit out the answer of hey people on our platform are looking up this or do you think they would gatekeep their own >> information? The challenge with like

00:47:46

getting onto Gemini. It's it goes down to hallucinations. >> Okay. >> And but that's actually a really valid point. I think to give you I think it's good enough to start the journey. The critical thing with like say what I call a longtail strategy which is go after don't go after the obvious terms that everyone's going after but go after like those more specific terms that tackle specific problems specific context etc. And I think it if you go after that really the strategy is not to obsess too

00:48:19

much about if I am I picking exactly the right keyword is that the best one I said but rather just get into the discipline of deploying lots of content targeting all those smaller terms and then off the back of that then you get a feedback loop of okay these ones seem to perform better than others therefore maybe I should drill down on this rather than uh spending too much time on making sure you're targeting the perfect keyword. So I think you actually you made a very valid point >> using Gemini GPT etc. I think will give

00:48:50

you a good enough base like you search. Okay. Well, I sell these kind of products or services. Can you please give me an idea in my market? How what are all the different ways people might search for this? And then and then you'll list it out. And then it's simply okay, that makes sense to me. I should write content about it and and do it at volume. >> Okay. Now, I want to talk about queries a little further here. And you've talked about how I guess voice to text. Yeah. How would

00:49:24

you how would you how would I use the right terminology of what that is? >> Search. It's just called voice search. >> Okay. So voice search is something that is tracking or trending upwardly. >> Um for example talking to Siri or Alexa with voice to text or even if you're interfacing with chatbt. I talk to chatbt. Um, how is this changing the way we interact with the internet and how search queries pull data together? >> Yeah, it's it's richer. Like we've got

00:49:51

we're adding a lot more context to our queries and um, you know, it's it's enabling people to get, you know, closer to what they need. But the critical thing is it's gone from maybe two, three keyword search queries to six, seven, eight, much richer. So effectively how people are naturally looking for information is far more specific nowadays than it used to be. >> And yet most marketers are still very much in this mindset or SEO professionals in this mindset. They're they're searching very simplistically on

00:50:27

just these >> two three keyword terms and it's just doesn't reflect how humans are naturally. forget about AI, all that, but it's just humans aren't, you know, uh uh doing less and less of that. >> Yeah. If we go a little meta here for a second and I use a concept with you. When you think about how someone might be using a platform, let's say it's just chatbt and they're trying to solve a problem. They're not yet looking for a solution outside the platform. They're trying to

00:50:57

interface with the platform. Yeah, >> you might be able to give me an example here. >> Before they ask Chad GBT, hey, can you refer some agencies to me that might be able to solve this problem? I'm imagining that they're articulating the problem first. They're spitballing what might be the best possible solution and then when it gets down to some kind of distilled outcome, then they would say, "Hey, based on this solution that you've given me, who can help me with this?"

00:51:21

>> So then you would have a very rich >> absolutely >> uh uh what would you call that? the path that you're following is essentially the idea of the query fan out like essentially you're starting at a certain level like I want to hire a new agency and then you're going through iterations of that as information comes about and you know maybe drilling down on certain dimensions actually no I I should clarify that actually I want to you know AC does this like logo design but

00:51:50

actually logo design for a for an accounting firm actually no maybe a you know accounting firm a that specializes in whatever. >> I've got an example here. So, something maybe a bit more complicated. >> There's a brand we're working with. I'm not going to name names, but >> um you know, they're looking at developing a campaign to become the talk of Australia. >> And I'm imagining that when their internal team were going back and forth, they would have ideated

00:52:19

together with AI what are all the different opportunities or possibilities or methods by which we could get the market's attention. >> Sure. And I'm imagining hypothetically if they were going back and forth through voice to text on a conversation with chatbt, >> it'd be a very rich query that then they could go, "All right, now we've come to some kind of conclusion. Uh could you recommend some agencies that could help us solve this problem?" >> Y >> would then you have a massive longtail

00:52:47

search. How would that then turn into a search if it went from big conversation to solving problem? Yeah. >> Through to recommending a company. Well, what it would do is then it would um it would essentially distill all the context that you put into, you know, the journey. Okay. Well, based on this journey, I've realized that uh the criteria that are important to this business is X, Y, and Z, you know, in terms of this agency. So what I'm going to do is therefore I'm going to search

00:53:18

on Google for agencies with these attributes that have done work with a in retail or in pharmaceutical or etc. all the different dimensions that have been surfaced in that and potentially dimensions that haven't been surfaced like as I said uh you know B because it's very much a correlation engine it might say I've seen other queries like this around and generally people are also interested in >> these kind of dimensions as well and then it will go out and find the information and then that's how it will

00:53:53

distill the end answer that you get is based on the interpretation of that that context Okay. So then what parts of the traditional playbook of SEO are still relevant and what parts are now becoming more and more irrelevant? >> I think it's well the part that's still relevant is you can't appear on search and by virtue of that on AI if you don't talk about something you know you have to you have to have something. So you're saying I got to have a blog. I've got to have

00:54:25

>> not necessarily blog. I said I'm I'm probably I've got a very specific opinion about blogs. I think the waste of time but >> but I have to have written >> you have to have content. You have to have written content. >> Yeah. >> Um obviously AI is getting better at interpreting images, videos, etc. But it still goes down to text. So it's very important. So textual content is a foundational element to have presence >> for right now. >> For right now. and and I I cannot see

00:54:50

how this would lose its relevance in the future. So that is really the foundation of SEO. It's about understand the questions people are asking and then give them content that answers that question. So I think that's the bit that is equally relevant and in terms of >> so let me just make a distinction here. So you're saying if I don't have any textual uh content on my website or on the internet somewhere that I'm consistently producing essentially my business is invisible to AI.

00:55:21

>> It's invisible to Google and AI. So first is be visible to Google and because it's invisible to Google it will all it will definitely be invisible AI like you cannot I think it's like the Maslo's hierarchy like um a lot of people are so worried about appearing on AI they forgot that you've got you need a ticket to be even in the game like you need to be on Google first before you can get on the engines >> so it's really critical to understand and that ticket

00:55:51

>> is at its basic there are nuance is is having content that addresses the queries that people are asking. >> Okay, >> that's the critical thing. >> Now, if that's textual based right now, do you imagine that in the short term or long term, it will then translate across to video and visual content? >> I believe it will, but at its intrinsic level, it will still be textual because it's what these engines do is simply interpret that and turn it into text.

00:56:22

Okay, >> that's essentially it, right? So, >> and I know that the easier you make it for a system to understand >> what you're talking about, the easier it will be for them to give you credit for it and therefore rank for it. >> Okay. >> So, by all means, yes, it's definitely I mean already, you know, Google's able to transcribe like all the videos out there, all the YouTube videos already have an automatic transcript on it, etc. It's already there. But as I said, it

00:56:53

goes down to text. And that's where the AI engines process text. So make it easy. Not saying don't do video, don't do I think that's really important. That's engagement. Yeah. So I'm just purely talking from a uh organic traffic perspective, but I said it shouldn't be just that needs to be also when they arrive to engage them etc. So there needs to be an outcome from the engagement but at the most basic level you need content to get in the game. And the kind of your other part of your

00:57:25

question which was what parts of SEO maybe are redundant or less relevant as I mentioned before are the strategies that focus excessively on trying to nail those big big queries those big terms. they're out there simply because um you know at a more foundational level it's the most competitive and you know everyone's going after it. So that's at a most simplistic level particularly if you're a a upstart or you're a challenger brand. It's incredibly hard to it's not impossible but it's

00:58:02

incredibly hard and it's questionable the return on effort that you'll get. And then that's from a market perspective, but also on where search is going with with voice search, AI search, etc. It's all shifting towards the long tail. Anyway, and the big thing is it's still untapped. Like it's I mean I did a I did a uh presentation recently with a with a client and um they sell used cars and I you know I did we did the analysis a stat um like car sales for example you know obviously biggest site in Australia

00:58:38

selling cars. What percentage of the potential organic search traffic do you think they capture at the moment? >> If I was thinking about car sales.com in Asia Pacific >> in Australia, >> how how much percentage of internet traffic do they own for car sales? >> Organic search traffic. So like people search on Google, let's say. >> Uh I'm just going to take a pun here and say like 36%. >> Yeah, it's 2.3%. >> What? So it's >> where's the rest of it?

00:59:12

>> Well, that's the thing a lot of people forget like you know you get used cars as a term is search like you know is SCM rush 18,000 times a month. Total searches related to used cars in Australia is about 18 million. That's the head term >> and then 18 million spread across over 270,000 keywords. That's why I'm saying that's the long tail like there are particularly say in >> So you're saying that the 2% is just the headlines. >> It's those most obvious terms.

00:59:42

>> Yeah. >> That they are capturing >> but the other 98% is just every other piece of >> like every other little derivation you can think of that they're not capturing. It's just >> that's just out there. And it just shows it's a and I think that's the biggest misconception people have is that um it's it's really is an untapped part of the the search market. >> Play with me here. Hypothetically, you're going to launch the competitor to

01:00:07

car sales.com. >> Yeah. >> What's your playbook? How do you take over all of their traffic? >> We don't even take over their traffic because their traffic from an organic perspective is not even where the main game is. >> But if you were to chase the other 98%, what would be your playbook? You would use it's actually quite a straightforward process. You use the keyword research tools to identify what are the keywords out there. What are all the keywords that people use or terms

01:00:34

use to search for what you sell? Okay. What you can go the next step further and says you know what isn't car sales or other guys ranking for which is a lot. And just build content targeting that exact term. That's it. And you're essentially targeting whites space. It's largely uncontested because nobody else thinks that way about going after all these little terms rather than spending all your time on the one. And >> you make it seem so simple like look at all the things that no one else is

01:01:10

answering and answer that question. >> But I think it could be done at a even a at a smaller scope level. I mean as said we do deployments like in the thousands tens hundreds of thousands of piece of cond which I get you know it's could be quite daunting but I think it's just the it's just a I think what I'd love people to take is that it's the principle of don't do what everyone else is doing you know you can't win that way okay >> and you can start small

01:01:38

>> and all these tools are available where you could uncover 30 50 of these terms that are very relevant but maybe lower volume, less sexy to go after, but your probability of success with that is a lot high because the competition drops. >> So, let's say I create a piece of content in the car space. Let's say we write an article about the best alternative >> uh aftermarket parts for a new BMW. Hypothetically, that's the that's the piece of content. >> Where should I host it? Is it on my

01:02:11

website? Is it on social media platforms? to get the SEO traffic, what where should I plug it in? >> So, it definitely has to be on your website and um and in order for you to get credit for it. Uh in terms of social media, etc., I think there's a very loose relationship between, you know, social media presence and engagement and organic rankings, but I think it should be done anyway for amplification purposes. >> Okay? So put it on my website and then put it on social media linking back to

01:02:44

the article on the website. >> That won't necessarily help from an organic perspective, but I think from a marketing it makes sense >> to do that. But the critical thing is for for yourself is to have that, but also make sure you have a critical mass. I think that's where mass >> a critical mass of content like if you just do one a month, this is not really >> it's like a drip in a bucket. >> Yeah. Yeah. Exactly. And I and and I believe social media kind of follows the

01:03:10

same >> kind of like if you if you think of it like social media if you just post like you know if you just post once like you're on your Instagram like if you post once a week you're >> forget it you're toast. >> And I think um I think people have to use the same lens with as much as they're they they see it's a given. Of course I'm on Insta. of course I have to post every day multiple times a day etc. I think they have to just use the same lens for their organic content as well

01:03:42

that you need to push content out there >> to show that you are serious about answering people's questions and as you get that momentum at that critical mass then then you will start getting credit for it. Where a lot of people give up is because they didn't have that critical momentum. Which is why if you want to take it seriously, you want to try this out, I'd say probably at least 50 pieces of content around a given topic area. Uh if not more. The more the better, actually.

01:04:11

>> So when you think about critical mass for a company that's sub 5 mil, >> yeah, >> how many pieces of material should they be trying to create for their website per month? It's as much as possible pretty much. But I would say >> at at a minimum to get started, >> I think when we work with businesses of that size, uh, we aim for about probably at least maybe 30 to 50 every single month. >> Okay. So, one piece of content to two pieces of content per day.

01:04:45

>> Someone could arguably do this with AI on their own platform themselves. Absolutely. But of course, um the big mistake people make is they just publish straight away. There needs to be obviously oversight to make sure everything is in order, the quality of the content is, you know, is aligned with brand etc. But absolutely. >> So what are those some of those things that they need to audit before they hit publish? >> Well, I think it's because of the 2 to 10% hallucination factor depending on

01:05:15

your industry. Um you know, you have to be very careful about what you say. if it's finance, all those kind of I think it's just basic hygiene. Just all I'm asking people is just read before you publish. >> Just read it. >> Yeah. >> I mean, I've seen we're often brought in and we go, "Okay." And just look at, okay, someone didn't read this, did they? Like before they published just you can instantly see that, you know, something's off. And I think that's the

01:05:43

problem with AI. It's a very confident hallucinator. It doesn't even know it's wrong. Whereas this where the human layer is so valuable. It can kind of has a we have an intuition on something's been off. >> I've caught the platform before just straight up citing a website and a campaign that didn't exist. I remember I was looking something up. This is maybe like honestly probably around September uh last year. Uh I was looking up historical campaigns around logo concepts and it made up and fabricated

01:06:10

this entire fake source of information >> and then when I queried it about it, it was like, "Oh yeah, I made that up. My bad." Here's a real example. >> Exactly. >> So, it's kind of scary. So, you're saying, okay, when you write the articles, at least >> go through it yourself and read it and make sure it makes sense for you and your business and your industry. >> Absolutely. And um and is said I think it's get that momentum and I think with

01:06:35

that level of momentum the 30 to 50 artics a month you will start to see outcomes from that even if you don't invest in authority building or other kind of uh SEO techniques that are out there the sheer kind of like volume itself and the if we want to get a little bit technical make sure the content gets indexed and And there are tools like search console where you can submit a URL or submit that say okay hey I published this content. So rule number one is you won't get credit for something that's not even recognized by

01:07:10

Google. So you have to make sure that it's linked from your site and you submit it on search console which you can do manually and then that will give you some momentum that will give you confidence that more content equals more rankings which equals more traffic. The last question I have around the content subject matter that we're creating and putting out there into the world is how much of online search traffic is actually AI and bots versus humans? If you had to guess where it is now,

01:07:43

>> what percentage of the internet right now do you think is AI >> following the web versus humans? >> Oh wow, that's a really great question. I think I just used the maths behind it. uh let's say GPT is estimated about 15% of the search volume of Google although that stat I don't know exactly know how verified that is but that's kind of been a cited stat it's 15% of uh the search daily search volume of Google and if we assume that AI well we know that AI does

01:08:17

query fan out so let's say that it's basically 10x the volume of of what of how people are normally interacting with Google, then you could argue that it's 150%. Like essentially the 10 15% is actually you 10x the volume on that. Therefore, it's essentially 150% of the volume of of Google. Does that the maths make sense? >> How is that possible? because it's doing so many more queries per question, it's exploding that out into 10 or 12xing it. >> So a human might search one thing, but a

01:08:57

prompt you give to AI might search 50 things. >> Yeah, exactly. So the actual intrinsic volume on Google related to AI could actually surpass human human search. >> That's pretty crazy to think about. >> Yeah, that's a lot. And that's why you can understand why Google was so worried and it was consuming significant enough resources that within basically two months. So July it happened within two months like come September they had to try to shut it down as much as possible. They acted

01:09:35

very quickly because they saw just how much volume it was consuming on their index. When you think about a company doubling its entire traffic because of one AI platform and its search queries, that's pretty dramatic. What What does this mean for the internet, do you think? If this is a sign of things to come, what are we to expect? I think it's I mean um the general it all comes down to the the user experience and in spite of the horrendous amount of resources that go behind executing a standard query and

01:10:16

the energy that's associated with that at the end of the day people like a nice distilled it's all done for me kind of summary. So I mean it's it's increasingly becoming the interface to the internet this kind of conversational interface and I think that's just you know where where things are heading pretty much. Um there are obviously implications of that you know forget about and we could go on about AI data centers and all that kind of it's quite a you know has a quite a significant

01:10:51

environmental implication but um if users prefer that which they're you know we're seeing the trends in behavior then that's where the industry is going to head. It reminds me of that meme on the internet. I don't know if you've seen it, where it shows a video of someone with a pickaxe and they're crawling through a cave and they're trying to get minerals out of the earth. >> Yeah. >> And it goes into a refinery and then they make it into computer chips and

01:11:20

then someone's downloading Minecraft and then they're sitting there mining on Minecraft and it's the circle of life. I I can see how >> Yeah. Google getting 150% increase in their search volume because of one platform uh being a massive problem because we're not talking about just more search queries. We're talking about data centers. We're talking about Yeah. >> energy. We're talking about manpower, staffing. There's a lot of resources that go behind this.

01:11:48

>> Absolutely. And there's also the other consequence of that from a societal level is the convergence of thinking >> because now >> the convergence of thinking >> well now a lot of people are outsourcing their analysis and insights into AI. So it's all getting routed through a single engine and it's these engines are are having a lot of influence on how people perceive the world. So before like with Google, right, you search and there's a bunch of results and then you

01:12:19

can look through it and kind of start making your own decisions about what makes sense, what doesn't make sense, you know, what I'm going to take or not take. Whereas now you've got an extra layer, a curated layer between you and the underlying information. And these entities are basically controlling that, you know, using algorithms or methodologies that are not obvious. And so, um, the danger is kind of like we're exposing ourselves to potentially a convergence in how we think, which I

01:12:58

think is another interesting implication of all this. >> What exactly do you mean? I think the fact that when more and more people become reliant on a given AI engines or AI engines in general to to even think it means that and we're kind of like referring to one source that means we kind of the the opportunity for multiple perspectives kind of decreases significantly because we're all asking the same entity. you know, the same questions and we're getting the same curated set of answers

01:13:39

rather than previously or currently dare say we might be pulling information from a lot of different areas, you know, and dare say even talking to people as well and pulling that and then synthesizing our own ideas versus relying on these kind of relying on these, you know, these these machines. And I think that's kind of something that will become yeah will become a more serious issue as they become more pervasive. >> There's there's a governance that we h are beginning to adhere to and it's

01:14:14

through this Google >> platform perhaps. >> Absolutely. And this is where all the think Wall Street Journal a lot of like these kind of like journalists are really um uh exploring this kind of this potential danger cuz it's kind of all converging like these oligopies you know these a these mega AI companies that have will have you know an extreme level of influence on you know how how people think. It's uh yeah, it's something something to look at. I think it's very serious.

01:14:49

>> Yeah, Jeremy, I want to get into tactics so those are watching or listening at home can start to apply this themselves and get some form of momentum because I can imagine if Google is the battlefield that all these platforms are fighting over, that's where we need to be right now putting our content. And I can imagine stockpiling good content right now could be helpful if we want to start ranking in the future for AI search queries as they become ever more complex. Now before you were saying that Google

01:15:20

stopped doing 100 searches and now they're doing 10. Is this the same for everyone? >> Yeah, I mean it's uh no one can get 100 results in one go now. It needs to be 10 at a time. So yeah, it's it's relevant to everybody. >> Okay. So if this is relevant to everybody, then that means that we must then fight to get ourselves in the top 10 if we want to get clicked on from Google's perspective. >> And then if we want to get recommended also from AI's perspective, we also need

01:15:50

to rank >> exactly >> top 10. So what does a business need to do right now if they haven't done anything? Yeah. >> To start showing up or have a chance to show up in the top 10 for search queries. >> Yes. Don't try to show up in the top 10 that everyone else is trying to show in the top 10 for. That is the probably the number one piece of advice and you know goes into the idea of the longtail and look at what you sell. find out obviously there's an underlying category

01:16:20

but underneath that there's a range of other ways that people can search for what you sell and definitely can use as per you know you put you using AI can really help with that there are tools that I mentioned previously that could help surface more but just using >> Uber suggest >> Uber suggest rush ah refs Google ads keyword tool that's also available and um and and then yeah use Gemini or GPT to uh get some ideas. >> So the first step if you will is knowing what people are looking for.

01:16:55

>> Yeah. What people are likely looking for around your product or service area and just say can you please you know extrapolate from this because this is my business. This is what I do. These are the kind of people that I sell to. Given this, can you extract for me hundred or a thousand different keyword ideas that people would search >> that would be contextual to me that is unlikely to be targeted by anyone else. >> So let's say we create a list. You're looking for let's say

01:17:34

a 100 different topics >> essentially or 100 different search terms >> exactly >> that aren't in the top 2% that everyone's fighting over. >> Exactly. create a list of a hundred and then what? Pick one of those and then create a piece of intellectual property or subject matter content around that. >> Exactly. Write content for you know 30 to 50 a month. Like that's why generally we recommend try to get maybe a thousand of these and if you know um or maybe start with a hundred.

01:18:07

Let's try to be bit more conservative. And once that's done, obviously make sure that the keywords make sense for your business. There's no point sending getting traffic that's not relevant to you. And then just write content for every single one of them related to that keyword. >> How would you suggest people best do this? Would it be more ideal for them to just voice to chat to uh an AI platform to give it as much context that that individual has around their business for

01:18:34

that answer, for that query? Yeah, I mean essentially you um and there different techniques depending on the AI platform you use, but it can save your context around this is my business and this is the kind of keyword I want to target. Can you give me some content ideas around that would be targeting this specific term for SEO purposes that would be relevant to the people that I um that I'm trying to market to, you know, and experiment with it. said just you and I I guess the other bit is remember that you're targeting an area

01:19:10

that is a lot less competitive than you know the the head terms in which case uh definitely you need to spend time and make sure it's accurate and make sure it represents your brand etc. But we don't need to go to the same level of optimization that you would if you're going after a highly competitive term. So it's about it's a really a volume play at this stage. The critical thing is to get confidence in >> the translation of content momentum equals organic performance.

01:19:48

>> Content momentum equals organic performance. So the more you more you put out there >> the better you're going to perform on search engines. Yeah. >> That will then translate to impressions and then clicks etc. And secondary to that is then the AI aspect to it. >> Right? >> I mentioned before that you know AI is increasing. There are ways to measure it etc. But still uh search is still a huge channel and I think should be the focus of this initiative. But there is a

01:20:17

secondary benefit which comes from AI. And previously um you know I used to think that and the industry still thinks it's really hard to measure the impact that this kind of work is doing you know in AI engines but in fact uh we now know that's incorrect that you can actually see what's happening because of this query fan out that I've been describing if you use search console which is a tool that you get for free for every single website. What you'll what we've seen uh is that for these longtail

01:20:51

targeted pages that impressions will go up against those same pages, >> right? >> But those impressions don't necessarily drive all drive clicks. Some will drive clicks, not all of them. >> Okay? So, if I'm let's say your perspective right now is that the search terms have shrunk from 100 to 10, which means that there's less opportunities and everyone's fighting over smaller territory hypothetically. Y >> and that there's 150% of the internet is search fan out queries from these

01:21:21

platforms not particularly humans. >> So then are we therefore writing this content not for other humans to consume but for AI systems to scan get insights from and correlate its own uh synopsis of its own thinking >> to some level. I guess you could say that that it needs to I guess we're trying to publish as much content out there so that these AI engines will ground. >> But is it humans that are reading this or swarms of AI queries? >> Well, it's if all goes well, it'll eventually be

01:22:03

potentially a human, but definitely, you know, the the gatekeepers is the are the AI engines. Okay, pretty much. So, you have to do for both. How do I write good material for an AI system to scan and to validate that it's a good source of information? >> And this is the beauty of this strategy is longtail. I don't want people to get distracted about how do I tweak this and that around you know specifically for AI and not important. >> Yeah. Just write an article, a piece of content that is

01:22:36

legitimately focusing on the term or the underlying intent of the term like you would be able to assess as a normal human being and doing that at volume will be enough to start showing outcomes. Okay. as a result of that and I think that's where people feel that there's some sort of special formula that needs to be in place to you know get that. >> So there's no formula you're just saying it's about volume >> for it just do the volume read it make sure it makes sense to you that's

01:23:08

important and then publish it but just keep publishing. So how does having large amounts of volumes of published content why is it the critical size or the mass of that the important factor here from our experience what I believe it's talking to is the fact that Google's trying to assess which sites are really investing in giving answers to the users's queries and I believe it's interesting thing. We're seeing this kind of convergence in some of the the social media platform

01:23:47

logic and Google's underlying search logic as well. You know, uh you you an expert at at social media. I mean, maybe I could ask you the question, why why shouldn't someone like if someone just publishes like on Insta once like a week, why is that not as good as doing it? >> Yeah. >> Several times a day. >> My theory to this is what Darren Hardy writes in his book, The Compound Effect. I don't know if you've read this book, and anyone listening to this, I would

01:24:23

encourage anyone read this book. He has this theory that if someone, let's just say we're neighbors with someone, one person ate an apple every day for a year, the person next door ate a double stacked greasy cheeseburger with double cheese on it for an entire year. >> After a week, both neighbors look like they could be the same. >> Yeah. >> Six weeks, not much of a difference. But a year, >> yeah, >> you're going to see a radical outcome of difference. I I think the more

01:24:52

frequented the touch point of a piece of content touches the market, the more a human or the system or the algorithms interface is going to trust that individual. So my guess would be if you're uploading a thousand pieces of content per month >> and then someone else is uploading 10 or 50, >> the algorithm, the AI, the platform, the company is going to trust you >> 10,000 times more than the other company. Why? Because you're intrinsically more valuable. So I think it's

01:25:27

>> it's it's like putting money in a bank. the more money you put in, the more consistently you put it in, the more you get back out, the more interest you're going to recuperate. >> Is it is it more complicated than that or is >> No, I think it's it's it's that's the simple dynamic. I mean, at the end of the day, >> all these engines where the social media engines, search engines are just trying to deliver the best experience for their end user and they're trying to curate

01:25:54

the information or the providers of the information that they think will provide. >> Okay. the best experience. So that's why I think you use a social media lens which a lot of people understand like why you would this every day you know you know show up in the conversation more often than not um to be more relevant and it's the same concept as you know search engine optimization. You just have to keep publishing to show that you are a trusted source of information and on top of that you are

01:26:26

putting out content that isn't really tackled by anyone else. So for example uh why doesn't it work when you publish on in if you're just reposting other people's stuff if I did that a thousand times why would that not work? >> It's because you're not adding any value. You know intrinsic you know net new value. Well, there was a point in time where people were doing that on social media and they were having a lot of success, but then now the platform knows >> exactly

01:26:54

>> it. >> Yeah. Instagram, if you if you do this, if you screenshot someone else's piece of content and then you upload it, >> it knows that there's an original source of that piece of content and you've replicated it. >> Exactly. >> Is the Google platform the same? >> Exactly. Like they have they understand where >> and this is where it's important actually. This is another tactical aspect is where you're syndicating content. If you're syndicating your content to

01:27:19

other websites, the exact same content, and that website you're syndicating to just happens to be more trusted, high authority, then it's possible you could look like a copycat, too. >> Because uh and this is the problem. There's a common problem with a lot of ecom sites where they've got their own unique products and they're and they're maybe selling this through Amazon or through any number of marketplaces and they publish and it looks like oh they look like the first one to get index so

01:27:48

therefore they get credit for it and you look like you're just a copycat. That's that's kind of a niche case like a lot of people are are >> you know kind of like directly republishing all their content but it's really important but I think >> the social media logic is actually quite sound for why you also want to apply that same logic for search engines. >> Yeah. >> Keep it fresh. >> Okay. Now can people do this too quickly? Can they make too many pieces

01:28:15

of content too quickly? And can that create any problems? In terms of how we're implementing say CMAX our system which is a hyperscaled approach we would tend to avoid doing more than 2,000 at once. So maybe that is the limit. Try not to write 2,000 articles in a week. >> Okay. >> So >> why is that the limit? uh because what we want to do is we want to test another mechanism uh called crawl budget uh within that Google has on how much resources it will assign to even look at a page of content

01:29:00

that you publish. So it's not a given that when you publish content that Google will even look at it and it's a function of what's called crawl budget and there's a lot of theories about what it goes behind but in terms of our perspective on essentially how much Google trusts the site it goes into you know to its age it's you know other reputational factors etc and uh it's and the reason why that one or 2,000 pages at the start is it allows us to stress test the level of authority of the site

01:29:30

how budget it has. If we see that when we publish one or 2 thousand pages, it gets indexed like Google adds it to the index very quickly, it means that the budget is good. If it's not, then there's something holding back. Like Google doesn't trust the underlying site itself enough that it's even going to bother looking at your additional content. And then there are mechanisms we then deploy to improve that authority. So, if someone's not had any content go up on their page yet, how

01:30:01

should they roll out over the next, let's say, 12 months? >> Oh, if you're doing 3050, that's very like that's not um and you're just submitting it into Google directly. Uh you will not trigger any major issues like this is quite a safe low volume. I wouldn't let that inhibit people from trying this strategy, the 30 to 50 month. Yeah. Let's say someone wants to ramp it up from there and they want to do hundreds of articles or pieces of content per month. How can someone

01:30:33

actually even put themselves in the position to start doing that? It's extremely challenging which is why we ended up building CMAX because of that exact challenge. Because another variable that then comes into play is when you're deploying lots of content uh you also ideally need to keep it fresh because there's a concept where when Google sees a piece of content it will also know how frequently that content is uh is updated is refreshed and if it isn't done for a period of time uh essentially the algorithm's

01:31:13

level of trust in content starts to degrade. And it's different for depending on the level of competitiveness of the term you're going after after what are other articles that are tackling the same uh kind of like query. For example, uh if you've been the only one targeting whatever obscure car part, etc., then no one else has come in with anything similar, then there's a chance you'll have a little bit more longevity. But if more people come in then they'll be assessed based on then freshness will

01:31:46

become a variable. So you get into this kind of challenge of if you're doing say hundreds or you know if you got even thousands per month then what ends up happening is that it can start to become a liability if you do not then refresh that content. So we recommend probably at least an every 12 months to do a refresh cycle on all your content. And CMAX as a system, the way it's built, it does updates every six to eight weeks. And this is because we're doing it at a, you know, literally

01:32:20

thousands, tens of thousands scale. And that's a really important aspect of what we're seeing the algorithm is evolving to is that that freshness aspect. Again, it's sounding more like social media. >> Yeah. >> It's it's like converging quite a lot with social media principles on freshness. >> So, let's let's say someone's listening to this right now. Yeah. and they're opening their laptop and they're saying, "Okay, Jeremy, how hook me up. How do I

01:32:47

start to pump out a couple hundred blogs a month, a couple hundred pieces of content a month? >> What's the process you would encourage them step by step to go through themselves >> to start creating this process or to start replicating what you're doing? I know it's difficult. I know there's a lot of work involved, but if you were to completely break it down for people to go, all right, step one, sit down, open up this platform, punch this in, steal this, move it over there.

01:33:11

>> Could you walk us through this and create a mud map here of someone how >> how someone could do this for themselves? >> Yeah, I mean it all starts with the what are people actually asking and finding out all those terms as per the tools we mentioned. Get a huge list. Ideally, not a hundred. Maybe even as many as you can get. >> How how many is a good list? Let's give >> to start people off. If we say get six months worth of, you know, momentum, you know, of, you know, doing 30 to 50, you

01:33:43

know, articles per month. >> Yeah. >> I would say let's let's start with a thousand. >> Whoa. >> A thousand keywords. Just keyword research. That's quite easy to get. Okay. >> A lot of these tools spit out lots. like you'll realize that there's a lot of varied ways that people look for. >> So, how do I do that? What would you suggest? Let's say I open my laptop. Where do I go next? >> I said the tools like uh Google ads keyword tool. There's, you know, Uber

01:34:09

suggest, SCM rice, ahfs, >> and then use those platforms to identify a thousand things, >> a thousand ways. Yeah. people that are maybe the a thousand keywords that aren't the obvious ones, not those like two, three keywords like simply maybe a little bit and you'll have to make a judgment call on that. You as the domain expert would know, okay, well, this is a little bit this is how most people would initially look for this and this is probably more, you know, this is would

01:34:38

be defined as longer tail. >> And then what's a long tale typically? Let's say short tail is like two or three words. A longtail is typically what seven to >> it's really hard. I mean typically it's like four four onwards but it can be three. It all depends on the context of the of the >> typically speaking it's longer they are hence longer tail but it's they tend to be longer strings longer query strings >> but it's really about the context of

01:35:06

that search the dimension of it. >> Okay. Um, >> so let's say I get a thousand listed out. Then what? >> Then what you do is then that's your that's the plan for the next, you know, couple of months where you go, okay, well now I have to um work out how I'm going to write a piece of content related to each and this is not blog content as mentioned. I don't believe in blog content because blog content is purely forformational purposes. there must be some sort of commercial intent

01:35:37

behind and this is what you can build into your prompt as said to to kind of streamline this you would kind of let's say GPT maybe you create a custom GPT in which you load a lot of information about your business like all the content about the products you sell the kind of customers as mentioned as much context as you can and then within that context you go okay well I want to write a piece of content that um is relevant to my you know the c any of the customers that I'm targeting and uh

01:36:13

beyond just purelyformational purposes I wanted to kind of tie in together with our value proposition about if you're searching for this this is how we can help so you need to kind of embed that you know within within the copy and um once you do that you know it's quite a streamlined process to then get uh spit it out and then review it manually and then before publish then get ready to you know publish it and just do that. It sounds like a you know bit of a task but yeah you do that a thousand times and uh

01:36:50

as long as make sure that you're choosing you know the keyword choice is quite important but I think as a domain expert you would know that this is relevant or not to your business and that that's probably the most critical lens and you just got to make sure that it's also not too busy. Now, some of these, if you're using tools like SCM Rush, HFS, they have some scoring about how competitive that term is. But what I would recommend because it's not quite obvious how they do that is just search yourself, look at

01:37:19

that and go, okay, well, that looks like look at the people there and says, yeah, actually, I reckon we could do better. We could do better than that. And boom, assess it. Look, you can you can go more sophisticated, build agents to do this assessment, but I think if you're just quite new at the AI game, etc. I think that's the most basic level. It's a little bit of hard yards, but then you'll start to build a process around how you're writing it and then you can build more agentic workflows that will

01:37:50

follow that logic. >> Epic. >> Yeah. >> Okay. You talk about a massive blind spot with free AI users. What is it? >> Oh, yes. Yes. So, uh I mentioned before that one uh misconception that I had previously was that you know it's extremely near impossible to measure the impact of a or measure AI visibility and uh thanks to the query fan out that occurred starting July um we realize actually first of all impression wise we can 100% see whether or not we're increasing our

01:38:33

presence in this query found out based on the increase in what we call zero click impressions or lowclick impressions. We can see that in search console it's all there. You can you can actually uh interrogate the tool. It's all free. The tricky bit about that is of course well what about traffic? Like how do we know that any traffic's coming? And traditionally how traffic has been measured has been in in say Google Analytics it would say look I only want to look at traffic that comes

01:39:00

from GPT or perplexity etc. there's certain labels it looks for and when we used to do that um we go okay well it's not much that's why we never really took AI that seriously because wow the traffic is tiny so great you can get more impression you know you can appear but it's not really driving anything and then what we did was uh we just randomly thought okay well let's let's just turn off the filters let's not let's not use the standard way you'd look for you just

01:39:36

say let's look at all traffic going to CX pages I longtail targeted pages and the reason why it's interesting because CMAX pages typically aren't part of the main navigation they're not an obvious part of the site so it's not something that people just stumble upon it's very deliberate so it's kind of fairly clean data and when we remove the filters we realize wow there's actually a lot more traffic than we thought from other sources but the biggest other source was

01:40:05

direct and said this is really bizarre interesting and we saw that for most of our clients the way that impressions were growing kind of followed like this direct traffic and you thought that's weird why would that occur why would someone just come and visit the like this page makes no sense and then we delved deeper and did our own research and we realized that you you know, for GPT users, uh, 95% plus are on the free plan. And there's a little idiosyncrasy about free users in that when you click on a

01:40:45

link on a free plan is that for some reason, GPD strips out all the information that Google Analytics needs to identify where that traffic is coming from. And by default, if it can't work it out, it will classify it as direct traffic because there yep, it's in this miscellaneous bucket and I'm going to label that as direct, which is why we realize now and all our our all our reporting has changed as a result of it. We now report on all traffic in particular direct traffic going to these longtail ted

01:41:21

pages. So you can in fact measure the number one your appearance your AI visibility in the fan app through impressions in Google search console and you can measure the downstream impact in terms of traffic but make sure you don't just look for GPT referral traffic but also direct traffic to these kind of longtail >> why why does this matter why does it matter that someone's on a free account or a premium account and how does that affect >> I have no idea Yeah. No one's really

01:41:52

understood and that policy could change. >> Yeah. >> Um yes, when we saw this said it makes no sense and uh but we're just look at the data and you know it's really interesting and and the volume is modest relative to organic for a lot of our clients. In some cases, like one client in particular, um, this kind of like services business, it ended up being that that direct SLA AI traffic was ended up eight times higher than the organic. M >> so it's quite a interesting range on why

01:42:34

some are driving more clickthroughs than others but um but the quality of the traffic is also quite high you know coming from AI if you look at like conversion rates etc from it so although it's small it's still I think something to look to something to kind of closely monitor but as I mentioned if you are optimizing for search engine purposes as a flow on impact the AI traffic or visibility and then traffic should come with it. >> Does it matter where the AI traffic comes from? So let's say someone's

01:43:14

viewing on their mobile or a desktop or a country or a region. Does does this play any effect on how people search for things on the internet? I think it's like with I mean we use the same lens as search. Does it matter if if someone's arriving on your site engaging with your content? Uh obviously we can put country filters on top of it. That's quite easy. >> Yeah. But does it rank the the quality or the value of a of a click or you know someone landing there? From what we're

01:43:49

seeing from AI, it tends to be uh a higher quality. The reason being is that um I guess the level of trust people are imparting on AI recommendations appears to be higher um simply because there seems to be more thorough research that goes into the into whatever gets surfaced. The thing that I'm concerned about like when I'm thinking about this is if you look at Google as a platform and these businesses are paying to have themselves presented or promoted on this platform. >> Y >> but the majority of the searches and the

01:44:31

clicks is now becoming AI. >> Ah right got it. >> Why would then people still continue to pay to rank if they're just going to get fabricated traffic not really doing much with them? >> Yeah. Unless then there are businesses also showing up on AI, >> right? Okay. So, I guess the underlying question is, is AI bot traffic influencing Google Ads performance? >> Yeah. >> Yeah. >> Yeah. That's a good way of phrasing it. I think um at the moment there's

01:45:08

very little evidence of it actually from a commercial perspective, you know, because you're paying on a cost per click with a lot of these engagements. >> But if my clicks are bots, what am I paying for? >> Uh well, I think this is this is another kind of like rabbit hole about let's go down about about click fraud, etc. >> Yeah. I think um there's based on how these AI bots work and what they load and what they don't load, the probability that it's these bots are causing fraudulent clicks on

01:45:45

ads. I don't think there's enough evidence out there at the moment. There's definitely a impressions kind of like inflation. So in terms of any media placements etc that are where you pay on an impression basis there could be a an argument for you know some you know questions to be asked around it but from a CPC level I'm not sure but it's absolutely something that should be monitored very closely because this is a ever evolving space so the capability of these particularly with all these agent

01:46:23

workflows where these AI engines are doing more and more for the user to the level of even maybe buying things uh you know like um open claw and all those kind of you know uh different >> almost autonomous systems. Well, let's talk about openclaw for a moment. How has the recent headlines of open claw affected how markets? Let me rephrase this question. H how has the recent headlines of openclaw and its disruption on the market impacted how people are navigating the internet and thinking about how to conduct business

01:47:04

online? >> Yeah, I I think it's quite fringe to be frank. I mean uh I mean there's a variety of data out there but there's still a very small uh I believe like sub 5% number of people who are actually using AI in a meaningful way. So I think it's we're still in a very early stages and you know the open claw I think is very headline Grammy because you know it's people are able to do a lot more with it etc. But it is still a a a small but extremely vocal community. But it

01:47:39

doesn't represent the broader market. the implications of it, what they're doing, uh, I think gives you an indication of what's down the line because essentially, as I said, they're, you know, they're using all these essentially autonomous agents that are, you know, performing functions that systems can't discern from a real human being. So, it'll be very hard to differentiate between a bot doing something versus maybe, you know, a, you know, an actual legitimate human being.

01:48:13

>> Do do you think the game we're in right now when it comes to marketing and clicks and getting attention is now down to the war of who can replicate human function without being caught that it's actually not a human. I it's a weird place to be because I'm now seeing >> these platforms trying to replicate humans >> and it seems that people are trying to replace human function at scale. What happens when you can't make the distinction between AI and a human?

01:48:42

>> I guess you just have to look of if the AI is like uh paying actually uh bring out their credit card but it like with you know manice perplexity etc. They're all like embedding agents on your actual desktop which allow you which are increasing the level of access to the underlying to the actual computer itself and its functions. Um yeah, that line is getting thinner and thinner between what the AI can do and what you can do as to how do marketers navigate that and that's how we've

01:49:18

approached marketing all along is like well okay in spite of all this etc. uh if I'm investing a dollar, can I clearly show that this investment is producing >> $5 or $6? Um because I mean, look, you know, in terms of even before AI, um and we did investigations, we were doing work with, uh what was called Fairfax back then on on um uh you know, display advertising, etc. I mean that that area was already rife with a lot of bot traffic. But at the end of the day, it just came down to well, okay, it's a known evil, but if I

01:50:03

stick to my core metrics and understand the unit economics of business, if it still makes sense, keep doing it. If it doesn't, stop it. And yeah, but it's just a necessary evil that we can't >> avoid. It's just, you know, but yeah, it's it's a it's an interesting point. I hadn't really fully thought that through before, but yeah, it's great. >> So, something something that you teach your clients is to audit their own direct traffic spikes. >> Are you talking about when you notice

01:50:31

that there's a surge in traffic all of a sudden on your website? >> No, it's rather that if you want to have a look at potentially, let's say specifically, I mean, you should be aware of your traffic anyway, but um but when you're trying to look at potentially AI traffic um is just be aware that it might be showing itself not directly as coming from GPT but in fact direct. So often what we do is overlays of looking at impression growth, particularly zero click impression growth or as I said low

01:51:08

click-through rate impression growth and then have a look at the pages that seem to be appearing for those impressions and seeing what is all the traffic going to not just GPT traffic but just realize that AI traffic is hiding under the direct label. Okay. And can people tell the difference between human traffic and AI traffic if you were to analyze someone's website? >> Sorry, that direct traffic is actually humans. It's not AI, right? >> It's not AI. Apologies. I should have

01:51:42

clarified that. So, >> one thing is that um the way the AI work is they download just the HTML of the page. And for the the way that most tracking works is based on loading a piece of JavaScript. So, the direct traffic that I'm talking about is actually human traffic. So you can tell the difference between a human scrolling a page and an AI scrolling a page. >> Well, the way to look at AI interacting with your content is through what's called log files. >> Log files. >> Server log files. Server log files. So

01:52:12

essentially >> on the server, most people don't even know how to access it. But essentially there these files that get accumulated where every visitor that comes to the site gets recorded in this, you know, in the server log files. So they record everything that comes in and that's how you get a definitive view of what the Google bots or Google bot or the AI bots are actually looking at. That's the that's the cleanest data. >> So if you ran an audit on a website, you

01:52:39

can actually tell if AI is crawling it, if Google's crawling it, if a human's crawling it, >> you can see that. Yeah, >> the data shows up differently. Okay. >> But within GA >> GA Google Analytics when you're looking at direct traffic Google Analytics >> that needs to come from someone or something that is uh loading JavaScript, >> right? >> And right now the AI agents aren't loading that >> in which case the direct traffic we're

01:53:07

seeing now >> is actually human traffic. >> Okay. So that's that's where >> so let's say I upload a 100 pieces of content for 12 months. I have 1,200 pieces of content on my website. >> I run an audit. I want to know how much of my website was crawled by Google, how much was crawled by humans, how much was crawled by AI. >> What's a healthy ratio? What's something I should be looking for? where in some cases for some clients the AI will the inferred AI traffic is maybe a you know

01:53:38

maybe 10% of organic and you got in some cases eight times organic. So there's kind of like it's so uh I guess it's so contextual to you know the market you're in the level of competitiveness also the type of terms that are being targeted. There's so many variables in there. I think it's it's it's almost impossible for me to give one one clean answer. But all I'm saying is I think the most important point is for now just worry about organic, worry about the search

01:54:12

engine. >> You know, I think is said if you worry about that and you you keep an eye on that if the metrics are strong with organic with search then AI will follow through with that. it will >> there there is a sentiment that if I'm not paying for it can't be good like if I'm not paying for Google Adwords if I'm not paying for my content to be boosted there's no way in hell I can have a chance to even compete with the competitors who are so let's say someone's a small

01:54:45

competitor in a very busy market is is it truly a volume game and can you win without having to pay Google >> people who are prepared to invest in the content I I think they can reduce their reliance on paying Google. I think that's as to whether or not they can succeed to the ultimate level of their potential without uh Google ads. That's a big question mark simply because >> do you have a a proclivity to think that if they were to pay for ads that they would have an advantage?

01:55:20

>> Right. Like I'm imagining if I were Google >> Yeah. and there were two users, I might go, "This guy's paying me, this one's not. I'm going to reward this guy." >> As to whether >> paytoplay, whether there is a halo effect of level of spend and organic performance. Um, there's a lot of conspiracy theories that I haven't >> Give me the conspiracy theories. >> The conspiracy theories that you spend more than you will. I I think there is a

01:55:49

there is a mechanic where it can help but it's not the mechanic that people talk about in the terms in terms of oh you're paying a lot therefore all right there's a little button that says hey Dane's Dane's a good spender kind of help him out on the other side. It's rather that um there's a report in Google Ads which is called the paid and organic report. And what it shows is and we've seen this report time and time again is where if you have a presence in like the ads and

01:56:24

you have a presence in the organic results. So basically more real estate in the one right >> the click-through rates for both go up. >> So it's really about presence and recognition. >> If you were to have a guess Yeah. What should someone's minimum budget on Google be to give themselves the best chance to get traction? Oh man, it's uh I >> trust you got here like what what what's like a >> it's really hard as said it's it depends on the competitiveness of the industry

01:56:55

and we've got to remember that when people going into this there's a there's an aspect called the you know >> there's a learning phase with all this which which is actually is parallels really well with the kind of SEO strategy that I'm talking about >> where at the start you've got to accept But as with any learning learning endeavor, not just around advertising, that you got to be prepared to make a ton of mistakes. And what I tend to recommend with clients is that

01:57:26

to be really serious about it, I'm say I'm not saying you can't, but to be really serious, I would say 20 to 30 grand, you got to be prepared to completely burn. Like just try a whole bunch of things. test it out and see what's going to work, what doesn't work for you, and then hone in on that. I think that's kind of like a reasonable expectation. Um, and there are arguments to say you can start lower, but you will get insights a lot slower as a result of that because it's all about the amount

01:58:00

of traffic you get, the feedback mechanism. So, hypothetically, if I'm going to burn 20, 30k on learning, >> what should I be looking to learn to get the most out of that spend? >> Yeah. >> Um, well, I think first of all, I mean, now the I guess the the methods for targeting. So, I'm going to use like a simplistic I guess uh way to target on Google Ads. Like there's a lot richer ways but there's concept like say broad match. So essentially you start very broad and try to target you know the

01:58:39

edge of the market like as broad as possible as you can define who would be interested in terms of what kind of terms they would use what are the what's the demographic profile etc. And then that's really important to cast the wide net at the start >> because once you do that you're going to get a lot of misses. There's going to be a lot of you know okay no that didn't work and you know you got to know what is the metrics of success and typically say e-com it's easy how many add to

01:59:07

carts and then actual transactions all that stuff you got to measure it and you just got to be very quick in being able to compartmentalize okay well let's put this bucket of those you know the spend that didn't deliver anything and let's look at the spend that did deliver something and the spend that did deliver what are the common attributes of that and the common attributes of the other ones and do less of the one that did less that didn't deliver much of an outcome >> and double down on the ones that are

01:59:38

delivering outcome. >> Yeah. You're almost using it like a >> you're panning for gold. >> Exactly. Exactly. >> This one's working really well. This one you're you're AB testing. >> Absolutely. constantly and that's >> like to think that >> if you try to be too precise at the start then you end up missing >> what >> opportunities and this is why the whole >> casting a broad net for content also works why you want to go

02:00:07

really aggressively and and do a lot of content because once you've published it you'll start to see which ones perform better than others. So your theory is rather than fighting over the middle where everyone's contesting over the certain key phrases, you're saying go to the outer edges of the galaxy >> and and start there and then work your way back. >> Yeah. And triangle and then hone in. >> Why is it better to go outside in? >> It's it basically allows you to tell it

02:00:38

allows you to explore the edges of the opportunity >> because a lot of people have assumptions about this is the only kind of people that I target. So that's all I'm going to do >> and not realize that potentially there are adjacent opportunities that you never bothered exploring. So therefore, unless you cast a wide net, you will never know that that is the case. >> What's an example of someone you've done this for and realize actually like on the fringes there's some money here for

02:01:03

this business. >> Okay, let's say like car insurance example like car insurance is a term. Of course, you're going to go after people searching for car insurance. It's obvious that's obviously. But if you think about the underlying need for car insurance is really about risk adversity about controlling risk like you know people who want to buy car insurance obviously you got to do there's a legal requirement to do it but there's you know there's different levels which you know so generally it's

02:01:30

for people who are trying to manage risk and um we kind of did some experiments around that and targeted terms that aren't directly related to car insurance but people who have some sort of uh who are risk adverse who um who are trying to control risk. You know, terms like um defensive driving and or engine immobil like different kind of terms that are adjacent to someone who is who is very conscious of the risk profile of being a driver and going after terms like that rather than the most obvious one which is around car

02:02:11

insurance. And the advantage of that is that it's not an obvious one. And but you can tell that it's the same kind of underlying persona. And by going after that, it's also uncontested space because rarely do other competitors think about targeting those kind of terms. And so as a result of that, much like Longtail, it's kind of un less competitive, cheaper. So, let's say someone's searching defensive driving cuz Okay, let's say in the middle of the galaxy is car insurance

02:02:44

>> and on the far far out of brim defensive driving. >> Yeah. >> You're saying it's actually speaking to the same potential consumer even though the search query is radically different. >> Exactly. Yeah. I mean, I think um Yeah, absolutely. I mean, you have to look at the underlying customer and what is the need that you're trying to satisfy. >> Okay. Can we test this real quick? I want to see how sharp you are at just thinking about different industries.

02:03:12

Yeah, sure. >> So, let's say personal trainer, right? >> Key search term might be how to lose weight. >> What might be something on the outer perimeter? Um the way that I would diagnose this to get to uncover it is I would I'll I'll explain the methodology and you'll get an idea. So for a personal trainer, I would go to that personal trainer and go, "All right, can I talk to your 50 of your C or 20 of your customers and ask them, can you tell me the journey that

02:03:47

led you to wanting to engage a personal trainer? Like what was the event that happened in your life that triggered the underlying need for a personal trainer?" And off the back of that information, yeah, actually at least 30 people, you would be able to start to uncover the different routes that people take to whatever product and service that you're offering. >> And it's often very different from the kind of like the idea that most business owners have in their head because you're

02:04:28

so caught up in your own perspective. I guarantee if you talk to 30 of your clients, you will be surprised with the journey that they took to that led you to where they are. So whilst it's not a direct answer, that is the meth that's the underlying method. >> Like having worked in a gym myself, having spoken to thousands of people signing up, >> you know, someone would say, "Hey, I was out for a social outing. >> I couldn't get my dress to look good, >> right?" The next person might say, "Hey,

02:05:00

I wanted to lose weight, so I just Googled best supplements to lose weight." You're saying you're going to find these outer rim search terms. >> Outer rims is actually what? Like there's critical pathways. >> Yeah. There's like there's an aesthetic route. >> Yeah. >> There's a >> emotional health. There's emotional route. There's a health route >> and there's maybe a recovery route where it's less like like part of your

02:05:22

>> part of your rehabilitation because of an injury like you or something. I kind of see like a like the city of Rome and all these roads Yeah. >> out into the wilderness and you're you're just trying to figure out how many routes are leading to this point. >> I think the critical thing is to appreciate that >> the routes that you think the customer is taking. Yes, I'm sure there is, but there's probably a lot more than you really realize. And this is the beauty

02:05:44

of the longtail as well is like the routes people search, you know, are vast and much more than you can perceive because you impart your own biases in terms of how you think customers act. And until you step out of that and acknowledge the fact that actually there could be a lot more, that's kind of like the start of real insight into >> targeting your customer. Yeah, this is getting me excited because I I can I can think about most clients I speak to aren't thinking to themselves, I'm going

02:06:21

to go interview 20 of my prospects to figure out how the hell they got here. >> Usually there's some conversation in the sales process like how did you find me? Yeah. >> Usually the response is like, I don't know, I found you on social media. But you're saying investigate that entire journey thoroughly with as many people as you can. >> Yeah. I mean we work we often operate on a lot of assumptions and those assumptions create blind spots. I mean, a great example, I mean, not even

02:06:46

interviewing like we a recent case with a with a with a fashion brand that was that perceived themsself is clearly targeting uh you know, was it I believe 35 plus >> kind of like females >> and we actually looked at the demographic data that we could service in Google Analytics and we realized actually your second biggest cohort is between 25 like it's a younger cohort And this has been a business that was running for over a decade. And their entire marketing was based on that assumption about the

02:07:24

customer base. And we had the data to suggest that actually it's a lot broader than that. You know, for business owners need to have some intellectual humility and and and uh accept the possibility that you might not actually know your customer. You have to ask them and look at the data and then see, you know, um what is the truth? >> Yeah. The truth will set you free if you take the time to look at. >> Exactly. Just take the time. And it's such a simple thing like is are you even

02:08:00

targeting the right people? And uh yeah, I think that those and this is kind of like a a question we all always ask, you know, when we're engaging with clients like because um yeah, we never take it on face value because we know that >> this is how it's been for years and that's probably even more reason >> it's outdated. So dig, investigate, look thoroughly, sit with the truth, write lists, and get a little crazy with the volume of how much we create to get this attention.

02:08:32

>> Yeah. Because you'll never he said you'll never we'll never know the answer. We just have to cast wide, make mistakes, and then just quickly >> triangulate where the little opportunities are and then hone in on it. >> Yeah. >> Don't don't try to do that all before you've got real data. A lot of people give up and make assumptions here and there, cut off huge chunks of market without any validation. >> No, I think I've been a victim of of

02:08:58

just going to the middle and just going, "Well, this is our core competency. This is the core search term. Let's just sit here." Yeah. >> You got me really excited to sit on our website and have a conversation with my team and go, "What's on the outer rim?" like let's get a a Zoom call with a bunch of clients and let's just meet with them one by one and figure out what routes they took to find us and then figure out how we can create content around those things. Even not just for

02:09:21

SEO, but even for YouTube, Instagram, all formats of content >> and it's a great it's actually a great strategy for your content. Like what you do is >> nowadays it's so much easier with AI. You simply do all the Google Meets or Zooms. You record everything and then you're all of a sudden creating these really rich personas that you need to make sure you talk to in your copy so that it engages them >> and yeah, I mean it's just a really healthy un you know it all goes back to

02:09:50

unfortunately like the you know it does require work doing basic stuff >> but it's simple. We have we have a phrase that we teach our students in the personal brand classroom and it's that TQAT. So all content is is a topic. Pick a topic. >> Mhm. >> Ask a question from the audience's perspective, not your own perspective. Like from the outside in topic, question answer the question and then make it theatrical. Add some drama to it. >> Don't just answer directly. Make it

02:10:25

dangerous. Make it sexy. Make it aspirational, make it funny, add some >> conjecture to it. >> And I think it's it's >> people think content is like a very difficult thing. It's just answering interestingly. >> It's having a unique perspective uh in a way that someone hasn't delivered before. >> Yeah. Absolutely. >> Yeah, >> man. That's it's equally applicable to, >> you know, to what we're talking about. Absolutely. Although that content

02:10:53

approach very it's actually very it's quite universal at the end of the day. >> Yeah. >> If you want to have conversations with your market >> be interesting. >> You gota you got to understand what they're curious about. You don't just >> tell them stuff for the sake of telling it. >> If they're not interested in it, why bother? >> You're the guy at the party talking and everyone's trying to get away from >> Exactly. You got to listen first,

02:11:16

>> then uh Okay, great. And it's just >> Yeah. on first principles. It's actually very basic. >> Two ears, one fog horn. >> Do the math. Jeremy, it's always a pleasure catching up with you. And I always love that you're predicting the future because if history were to tell when you were here last time, the predictions you made about Google came true. And I'm sure the predictions you're making today will also come true when we have you back on next time.

02:11:45

>> Absolutely. Thank you so much. In conclusion, in tradition of the agency podcast, what does taking agency mean to you? >> Talks the idea of really of agency is about it's really about doing what you want, where you want, with who you want, when you want, for as long as you want. And that is really what agency is to me at least. I've never heard anyone describe it like that. >> But the tricky bit about that framework is that what do you really want? That's the that's the tricky bit.

02:12:25

>> You know, for anyone listening to this, if you can truly answer that question, you're in the 1%. >> Yeah, absolutely. >> Most people don't sit there and go, "What do I want? What do I want? What do I want?" And then have an articulated answer that is meaningful to them. >> Yeah. >> Yeah. >> Always a pleasure and uh you're a dear friend of mine and I appreciate you being here, man. Thank you so much.

‍