AI Hype vs Reality: What actually works in marketing (w/ Rand Fishkin) | S3 EP2

Speaker A: Hello and welcome to the Paid Media Lab. My name is James. I'm the content lead here at Lunio, and in this episode I'm joined by Rand Fishkin, founder and former CEO of Moz, and now the co founder and CEO, uh, of the audience research platform SparkToro. I'm sure many of you out there will already be well aware of Rand, and he probably doesn't need much by way of an introduction here, but to cover the main points, Rand has dedicated his entire career to helping people become better marketers, whether that's through his writing original research or his educational videos on YouTube. A few years back, Rand also published a book titled Lost and Founder, A Painfully Honest Field Guide to the Startup World, which is a truly unfiltered account of what it's actually like to be in the trenches of finding a tech startup, which I think these days is all too often romanticized. He is the co founder of inbound.org as well as the co author of the Art of SEO, which for many of us, including myself, was practically required reading when we were first starting out in the industry. Over his career, Rand has given more than 100 keynote talks around the world and built companies that have shaped how we think about marketing, startups and, uh, leadership. What makes Rand stand out isn't just the scale of his success. It's his honesty about what it takes to actually get there. And in this conversation, we explore one of the biggest questions clouding the entire marketing industry right now. Namely, are we in an AI bubble? And if so, what happens when it bursts? It's a big question, and, uh, to help us answer it, we discuss SparkToro's research showing AI adoption slowing down, why Google search definitely isn't dead yet, and how marketers should be thinking about attribution, performance, tracking and channel strategy as AI tools become an increasingly embedded part of the customer journey. Rand also shares his take on what's driving the AI hype train and how you can future proof your strategy no matter how things shake. This conversation cuts through the noise and gets to the actual ground truth behind the headlines we're seeing in the moment in the marketing world. And I hope you find this one as useful as I did. Rans, welcome to the Paid Media Lab. It's great to have you on.

Speaker B: Yeah, good to be here, James. Thanks for having me.

Speaker A: Firstly, I want to say this is a bit of a full circle moment for me. Uh, we were speaking just before we switched on the mics about how I transitioned from my original career as an optometrist working for specsavers here in the uk. Um, lifetime ago, it feels like now, but when I made the transition to marketing, it was your Whiteboard Friday video series that kind of really helped me learn the foundations of how Google works, especially on the content and SEO side. Um, I remember sitting at my desk and I'd have two screens and one screen would be one of your Whiteboard Friday videos, and on the other screen I'd have my Google Docs and I'd be making notes, trying to, uh, learn and keep up with what the latest was. And I noticed you're keeping the whiteboard tradition alive over at the SparkToro, uh, YouTube channel as well, right?

Speaker B: Yeah, yeah, just, uh, usually once every week or two putting together a five minute video with some information. Rarely do I use the whiteboard these days. Usually it's a digital whiteboard now, but, uh.

Speaker A: Yeah. Nice, nice. So the topic we're going to get into today is about whether we're in an AI bubble and what the fallout could be for marketers and marketing departments if that bursts. But before we get into the details of the bubble and what it means for marketers, I wanted to take a step back first of all, and get a sense of your bigger picture view on the whole issue. So, broadly speaking, would you consider yourself an AI optimist or an AI pessimist? Um, on that kind of slide and scale?

Speaker B: Let's see. I would consider myself much more pessimistic than most of the media hype around it, and probably more pessimistic than, um, most of the AI company operators who believe that they'll reach general intelligence at some point in the future. But I would consider myself more optimistic than the extremely negative folks on Blue sky or Threads or Twitter who think that everything AI is baloney. Uh, I don't agree with that either. I think sometimes it's really nice to have a spicy autocomplete that can tell you what words frequently come after other words.

Speaker A: Nice. So you're not buying into those kind of real doomer arguments that AI is going to be, uh, the end of us all? Uh, in terms of an existential threat?

Speaker B: I know Elon really wants that to be true so that the only people who populate the earth are like his children. But, um, no, I don't see that happening.

Speaker A: There's a lot we could go into on Elon's children, but that's not the topic of discussion today. So I could spend a long time talking to you about that more philosophical, existential stuff. But to bring it Back to our main topic of discussion for today. Uh, do you think we are in an AI bubble right now in marketing? And if so, what is inflating the bubble?

Speaker B: Uh, we are absolutely in an AI bubble macroeconomically, uh, especially in the United States, where essentially AI is the only thing propping up our flagging economy. We would otherwise have been in a recession for several quarters now. Um, I think we have something over a trillion dollars of expenditure in AI hardware alone. And, uh, AI optimism is propping up the stock market at this point. Is that a bubble? Almost certainly it is. Um, I don't, I don't think that we can get to a normal price to earnings ratio for the major stocks with just the AI progress that we can make. Are we in a marketing AI bubble is a weird other question. Because it's. Marketing can't be disconnected from company growth, right? Like what when companies stop growing, marketing stops getting investment, people stop paying their agencies, they fire marketers. The marketing team is often the first to go. The CMO is often the role that's, you know, most replaced in those times that will happen. I, I have no doubt. I don't know when exactly it's coming, but, um, it would shock me if it's not in the next two or three years. However, in terms of marketing use cases for AI tools and putting AI to work to help you become a better marketer or to streamline some of your processes, or to do some of the annoying tasks that you hate doing, or to give you a bunch of brainstorm ideas so that you can be more creative. Sure. Well, uh, I don't think that's a bubble at all. I think there's. It's kind of like the Internet. Right. Ten years from now, there's going to be 10,000 new use cases for AI stuff and we're going to use it in technology. Just like we used its predecessor, machine learning, in a billion ways. Just like we used algorithms, broadly math algorithms in computer science and software and marketing. That's all going to happen. So it's complicated.

Speaker A: Yeah. In terms of the macroeconomic situation, I've heard, listen to some podcasts and I've heard some people talking and drawing comparisons between the situation that we're in right now and the dot com era crash. Like I was a little, I was quite young at the time when that happened, so I didn't really have much firsthand experience of it. But do you, would you say that there are some similarities between those two eras?

Speaker B: Yeah, it's definitely similar. Media hype and, um, Stock and investment hype.

Speaker A: Right.

Speaker B: So that, you know, whatever venture capital companies in the US have put, you know, billions of dollars to work into tens of thousands of AI first software companies that will almost certainly fail. Um, that's kind of the venture model, but I think the failure rate is going to be considerably higher. And then similar to the Internet, you fast forward a decade or two, a lot of those promises that many of those companies are making will probably come true thanks to a combination of not just AI, but other types of technology and software and maturation and investment and development. So again, we overhype something. We're human beings, so we go way too far on something and then we pull back way too fast, and then eventually we find a healthy metal.

Speaker A: Yeah, I think that's fair. Just on those promises. Before we move on to the next kind of topic. For me, obviously working in marketing, using ChatGPT a lot, you know, being blown away when it first came out and thinking like, well, this is really serious shift in how the digital world operates. And then noticing the shift from like ChatGPT 3.5 to 4 and being like, whoa, like, this is a significant improvement over what we had previously. And then kind of being excited for the release of ChatGPT5 and what's going to come around the corner and it was just like, is more or less a slightly better ChatGPT4. And for me, I was really underwhelmed and I just wanted to get your sense of, were you underwhelmed in the same way that I was, or did you kind of know that it's probably not going to be this big step change when ChatGPT5 is released?

Speaker B: No, I mean, look, I tend not to believe the hype generally from these companies. Um, and I think Sam Altman from OpenAI has, and anyone who's looked into his history and behavior would say, oh, well, that's a charlatan who got very lucky. Um, and, uh, yeah, I would expect you'll hear lots more. I think what the general public would call terrible, um, lies from him for a long time. And, uh, and it'll be okay. His career is going to be just fine. Because even if he lies about 50 things, the 51st thing is not a lie and is actually useful to a lot of businesses. And, you know, he's, he's, he's a little bit Elon Musky, more than a little bit. Um, in this, in this kind of way. Right. Elon made a million promises about Tesla and getting to Mars and rockets and all this kind of stuff. And even though nine out of 10 of them turned out to be just Bald faced lies, 1 out of 10 of them were useful enough that people kept paying attention and his stuff was useful and, um, his companies made money. I think the same thing's going on with OpenAI. Um, so your experience similar to mine.

Speaker A: Nice. Um, so you shared research, um, at SparkToro showing that AI adoption is slowing while search remains steady. And I saw a LinkedIn post that you made last week about how Google still absolutely dwarfs chatgpt in terms of search volumes. Like you said, Google something like 210 times the number of searches per day. And you've made the point that ChatGPT doesn't even have as many daily searches as DuckDuckGo yet, which really brings things into sharp focus for where we are actually currently at when you strip away the noise and the hype. So could you maybe say a little bit more about the points you were making in that, uh, LinkedIn post and then maybe give us your thoughts on why you think so many people are keen to push this Google is dead narrative and the table's been completely overturned.

Speaker B: Yeah, you said it quite well, James. Right, um, so our research comes from, uh, Dados was a clickstream data provider. We're talking about desktop here. So, you know, as many folks who are listening might know, Google is way more used on mobile than it is on desktop. But ChatGPT is more used on desktop than it is on mobile. So, you know, Whereas Google's maybe 66, you know, 67%, uh, mobile, ChatGPT is more like 30, 40% mobile. So again, we're under, counting and undervaluing the degree to which Google is dominating and dwarfing ChatGPT when it comes to search. However, that doesn't mean that ChatGPT and other AI tools aren't getting used. They are getting used. They're huge, they're big. It's just that comparing anything against Google is, you know, it's like looking at, uh, any other global economy versus the top five global economies. Like, you know, next to the United States and China and Japan and Germany, like everything else just looks so tiny. You're like, oh, Andorra is really taking off. Well, sure, but, you know, whatever it is, like, I think that, um, the reason this happens, the reason this happens is because it is not in anyone's interests, personal, financial, or your agency's interest, or your business's interest, or you as a marketing team to say, you know what, things change kind of slowly and this isn't that big. Of a change. It's interesting. We should watch it. But, you know, throwing billions of dollars at it tomorrow is not a great idea that doesn't serve anybody. What serves people is you need to pay attention. You need to worry about this. This is something to be fearful of. You better throw money at it, because that money gets funneled to individuals and agencies and companies and departments and AI tool providers and AI service providers and AI consultants and experts and AI conferences. And those people, both good and bad, want to make money. Right? And, and they recognize that the fear and, uh, surprise and expectation that's been created by the media hype and by, by people exactly like you were talking about, James, that first time you use chat GPT3 and you went, whoa, this is crazy. Like, this is. This kind of works. I can't believe it. It can do a million things. Um, that was a sea change moment for a lot of people. And as more people have started to use. Feels like magic the first time you do it. And I think m. That magic feeling created, that fear and uncertainty and doubt which now gets preyed upon by people. The other thing I'll say is if you know anything about how, uh, modern media works and modern engagement works, you know that let's say there's a hundred people like you and I who are saying, hey, you know, it's not that big of a deal, but like, here's, here's the real numbers around it. Here's what's really happening. That's not what's going to go viral. That's not what the press is going to write about. That's not what people are going to talk about in the office. That's not what your Aunt Sally forwards you, you know, in a panic. What she's going to forward you in a panic is AI is going to take your job. AI is going to kill us all. AI Robots can already destroy human troops. You know, like, that is the thing that's going to get attention. And the things that get attention also achieve virality, which means they also receive, um, the lion's share of our, you know, mental, uh, attention. And that, you know, that's very frustrating. I think this is, this is why, you know, whenever any researchers look at, like, why do you extreme political opinions or extremist content, uh, happen all around the world. Why is that taking hold? It's because the extremes are what get amplified and shared and broadcast and consumed and the sort of, hey, maybe let's think about housing density and focus on doing the right things for our community. Uh, let's just push brain rot all the time.

Speaker A: Yeah, that's the thing, like, nuance just does not work well online. And like you said, like the previous kind of the early iterations of AI that we had, which were these, you know, algorithms which were controlling our newsfeed. And like you said, yeah, it's an attention economy. You need to push what's going to get attention because we are maximizing time on platform and engagement because that means we can sell more ads and we need to. And then they found out that outrage is the thing that makes people come back. Getting into fights with people in the comments section. People cannot resist coming back on Facebook or coming back into the Instagram threads or whatever it is and getting into some kind of battle with someone, that's a strong incentive to return. And I think, yeah, I think in future, like, the impact that that has had on society and then this ability for social media companies to wash their hands of any responsibility that a newspaper editor at the New York Times would have is kind of outrageous, I feel like. And the whole legal system does need to catch up with the fact that, no, these are publishers and they should be held to some kind of standards about what are they putting on the front page, because what you put on the front page matters. And there is a decision that goes into that. So again, that could take us off topic and we could make a whole different, um, podcast on that.

Speaker B: But I think this is, it really does speak to, you know, if you're, if you're a marketer, you have to understand this. And I think understanding the macro, especially, you know, like political and economic, um, and what gets attention in an attention economy, I think that's part of your job. You need to be able to explain that to your boss, your team, your client, yourself, because that is what governs how marketing works in 2025 and almost certainly will for a long period of time to come. I think it also helps immunize you against the, you know, virus of extremism or doomerism or, you know, whatever the extreme content that's got you and your Aunt Sally so scared on the Internet today. You know, whether that's political content or racial content or whether it's marketing content or AI is going to take over the world content, you know, no matter what it, um, that knowledge will help protect you. And I think that, um, one of the things that many marketers don't realize is that we have entered this era of platforms dominating influence rather than websites. And many of us, unfortunately, James, still operate in a world where we think of our website as being the place where we're trying to get people to, rather than the website essentially being a resource that other platforms will pick up and point to. AI tools and search engines in particular, but social media too. And it's almost as though the job of the marketer has switched from uh, bring people to your website and influence them there, to create influence in all the places your audience pays attention and then sort of rely on them to go find your website if, if they are interested enough to do so. That's a, that's a really different world too. And the platforms hate to send you traffic unless you pay, but they love creating engagement. And so your job is to create that engagement and influence on the platform. Yes.

Speaker A: So this ties into one of the next questions that I wanted to ask you. So obviously at Lunio, we're a B2B SaaS business and historically we've been primarily blog driven content, mostly all written content. And you know, we've, we've pivoted and, and we've started to branch out and we're putting much more focus on YouTube, you know, case in point recording a lot more interviews and uh, one, you know, one thing I want, and um, I've heard to your point what you were saying about changing your conception of what you think as the website as. And I've seen other commentators on LinkedIn saying things like now, rather than all these visits to these blog pages that might be very low intent, kind of top informational type queries and you might be sad to see that your organic click volume is going down month over month. Ah, you should start paying attention to, well, what is your homepage traffic doing and how many people are hitting your homepage and looking at these other places that people yet like you said, will hear about your brands elsewhere on these other platforms and then make their way to hopefully your homepage or some other section of your site. Um, but should if for all the marketers out there in the B2B space, you have probably seen their organic traffic take a hit. Um, like, is that something that they should be concerned about and how should they be pivoting their strategy in response to that?

Speaker B: Yeah, I think I published a blog post earlier this year that sort of summarized what was happening on a in a bunch of different examples, you know, notably with, with places like HubSpot. Right. Everybody was talking about, oh, HubSpot lost 80% of their search traffic and they must be hit so hard. What a terrible whatever SEO decision they made. And then you go and look at their quarterly stock price and their earnings per share and their, you know, like, their performance. M looks like it's. They're having record highs. What. Where's the disconnect here? And they're just, they're just one example of thousands of traffic down, revenue up. And the reason that that happens is because, you know, in HubSpot's case, in a million of these other cases, the content that is being consumed, the influence that's being created is happening on YouTube, just like you said. It's happening on Reddit, it's happening on LinkedIn, it's happening on Threads, it's happening on Bluesky, it's happening on Instagram and TikTok, right? And all these places. It used to happen on Twitter. That's not really happening anymore unless you're in like, cryptocurrency or men's rights advocacy or something. But like, all the, all of these platforms have taken the attention away, Even Google search. Think about the fact that Google is answering 60 or more percent of all searches without a click, right? So, like, uh, two thirds of the time when you go to Google and you search for how old is Paul Rudd? Or what are the best hotels in Hong Kong? Like, it just gives you a list. It gives you the answer. It tells you what you needed to know. You don't have to go anywhere else, right? You don't, you don't, you don't need to click on anything. And that's Google's future. And people say foolishly, foolishly. There was an article in Fast Company yesterday or this morning, I can't remember, that was like, AI is making your website irrelevant. And I could not disagree with that more. I think, I think that is just the dumbest headline because where the hell do you think Google gets that answer from? Where are they citing that, that original Source? Where is ChatGPT getting their answer from? Your website is more relevant than ever. It's influencing people, but indirectly in a way that you can't measure. Just because the traffic doesn't come there doesn't mean the website doesn't matter. And it really bothers me. It just pisses me off when people publish headlines like that because I think it biases marketers and teams to make the worst kind of investments. So, yeah, I mean, I don't know, James. Like, I feel like we're in this sea change. People are blaming the wrong thing. They're blaming, like, oh, AI tools must be taking market share away from Google. That's why search traffic is down. Or that's why My traffic is down. And that's not what's happening. Google is putting answers in AI and Google Flights and Google Hotels and YouTube and like, you know, and their properties up at the top. They're the ones taking their own traffic away from you. Right. And the same thing's happening on all these platforms. Facebook used to send, you know, 10, 12% of all media traffic. That's down below 1%.

Speaker A: Right.

Speaker B: Twitter used to send 5% of all traffic to media sites in the US that, that is not even half of 1%. I don't even know if it's a tenth of 1% today. So, you know, you want to find someone to blame. The platforms are where that's happening. And you have to shift your strategy because you're not, you're not going to be able to convince, hey, Facebook, like, let's play nice, just send me more traffic. Again, that's not going to happen.

Speaker A: Yeah, like you, I was watching one of your um, whiteboard. Don't know if it's a digital whiteboard. It looks like a normal whiteboard to me. But it was a point that you were making about these three different stages of development of goog. And we had the kind of early stage where traffic was being sent to your website and then we had this kind of, I think you said it was 2014 to 2022 when um, feature snippets and rich snippets were the key thing. And that was Google's first iteration of users just want the answer fast and we're going to go into the content and rip out the relevant answer to match with that query. And people really like that. And then you were saying, then we've kind of moved into this new era where it's still zero click, but it's an AI generated answer. But to your point, as you were saying, that answer has to be generated from somewhere and you need to wash the Internet and all these different platforms with your brand position next to whatever terms you want to be fined for. Like in Lunio's case, it would be invalid traffic and click fraud and this kind of thing. But you need to be everywhere talking about your brand and your offering and position it, uh, next to relevant topics. And it does kind of require a mindset shift in how you approach marketing.

Speaker B: Yeah, absolutely. I think that, you know, the um, the old world, right, of getting traffic and then monetizing it once it gets to you and trying to increase your conversion rate, that that world is dead and dying. The world of attribution is gone. Let's Just pose an experiment here. Let's say, James, right, that you, you and I run, I don't know, a hotel in Italy. And we, you know, get on all these lists of top luxury hotels in, um, Puglia, right? The, the boot region at the, at the south, uh, southeast corner of Italy, uh, boot heel region. And, and like Puglia is this, this beautiful area. We've got a lovely hotel. It's in the perfect place. It's near the sea, where we have all these features and amenities, right? And so we get onto these lists and we, you know, manage to, um, kind of get into the language model so that when people say, what are the best luxury hotels in Puglia? In Google, in ChatGPT and perplexity, in Gemini, in, um, you know, uh, anywhere, or they ask their friends on Facebook and LinkedIn and, or they ask on Reddit, you know, all this kind of stuff. And people are recommending Rand and James's hotel, which is great for us. But what are we going to see in our analytics? We're going to see almost none of that. What we're going to see is that we will not see the hidden user behavior, which is person asks, whatever friends, family, um, people in forums, people on social media sites, AI tools, search engines. And then we come up in a list in an unlinked way, and then they go to Google and they search for our hotel. Who gets credit for that? Google. Google gets credit for that. Even though they did nothing, nothing to help actually create that demand or, or, or show someone our hotel, but they get the credit in the analytics. And so, you know, as savvy operators in 2015, we'd be like, okay, Google sending us great traffic. Let's spend more on Google Ads, let's spend more on SEO. Like, let's put a bunch into our Google marketing. Google was never responsible for that. And this biases a ton of people to do exactly the wrong things. The same thing is true, by the way, on, in my opinion, meta ads. So if, for example, let's say we run a ton of ads on Instagram and Facebook, right, and threads and that kind of stuff and like, oh, hey, this really looks like it's working. Like, you can see the view through conversions are very high. Like, people who see our ad on a meta platform later book almost, you know, whatever, 40% of people who book have seen one of our ads on meta. So we might take away from that. Oh, you know, that the meta ads are really working for us. Let's keep investing more there. Which of course is what meta wants us to do. But what they're really good at, what Facebook rocks at, is knowing where you're going to go before you buy. So they, what they're doing is essentially in their algorithm, they're identifying people who are like, oh, I see them browsing around, you know, Italian travel content and eventually figuring out Puglia, and then looking in the luxury subcategory of hotels and finding Randon James's hotel coming up for them a lot on the Internet, because everybody has their Facebook pixel on their website, right? And everybody web property is using this and, and mobile apps are using it. And so Facebook is like, hey, this person, they're going to book at random James's hotel. So let's show them the ad. Let's show them the ad so we can take credit. They are taking credit for sales that would have already happened. And you can prove this to yourself just like a hundred other companies have. There've been plenty of media stories about it. You know, bank of America and Chase and um, um, uh, Uber did this, uh, Airbnb did this, where they shut off their ads for a few months and they saw, uh, 95, 96, 97% of the same conversions go through. They're taking credit for sales that would have happened. Anyway.

Speaker A: This is really interesting because in another video that I watched you, you were talking about attribution and how it leads people to invest in the wrong areas. And in that, you were saying that, um, because performance marketing and paid media is very easy to put into a spreadsheet and deliver to your finance team and say, if we increase, uh, spend by this much, we're going to get this much out at the bottom. Um, whereas in SEO and content and other, it's much more difficult to get this kind of formulaic spreadsheets where you can get investment. But based on what you said there. Now, like, is this whole point about attribution just a moot point? And it's kind of like pointless. Like, how is attribution something that you're still thinking about? Because people are racking their brains. It's causing them such trouble with all the privacy issues coming in and we're trying to attribute and we can't get it right. And what, what's your advice to people who are really struggling with the attribution problem? How should they think about it?

Speaker B: My, my best recommendation is if you are obsessed with attribution and you have to prove it, just know that you really cannot do attribution for anything other than, uh, paid media. Everything else, pr, content, SEO Email marketing, um, you know, uh, social media marketing, like all those channels, organic social, I mean, all of those channels will be under counted or not counted at all. Let's say you're listening to this podcast and you're like, oh, this Rain Guy seems smart. Let me check out SparkToro. Guess what, friends? Like, we will never ever know the value, right? Or be able to see that. But it's very possible that if we were running meta ads, we're not. But like, if we were running meta ads or if we were running Google Ads, that you would do search for SparkToro and you know, we would buy that top slot for our own brand name and then we would attribute it to Google. You know, like, attribution is not dead, it's just broken. It's so broken that it biases you to what the platforms want you to do, which is spend more money with them. And so I urge people to do two things. One, if you want to measure paid, you have to be willing to measure, uh, you have to be willing to test turning it off or turning it all the way down. You don't have to turn it off entirely. I know there's reasons ad campaigns, you know, if you shut them off and then turn them back on, Google penalize you for that. And they try and do that so that you don't do the testing I'm talking about. But if you turn it down to like a dollar a day, you'll be fine, right? Like they let you, they still let you do that. So you got to test turning it way down and see what the marginal contribution of paid media actually was to your bottom line. For some businesses it's a lot. For a lot of businesses, it's under 5%. The other thing that I urge people to do is you got to go back to 20th century style measurement. Like you have to go back to the days when, hey, we're driving down a freeway, we see a billboard for Coca Cola. And in a five mile radius around that billboard, same store, year over year, sales with, you know, seasonality included, rise a little bit. And the Coca Cola Corporation says, okay, here's how much that billboard works. That is measurement, but it's not attribution. Because they can't prove that any one highway driver, right, who saw that billboard, went into the, you know, gas station and bought an extra can of Coke because of the billboard. But they know that statistically speaking, they got 4% more sales in that 5 mile radius or 10 mile radius, whatever, than they would have otherwise. That is measurement. And that's what you have to do with every other kind of channel. You want to invest in pr. Okay, great. We're going to invest in, you know, this specific niche media in this country or this region. We're going to see what our lift looks like in that region, sales to that area, sales to, you know, people who are in the demographics, who read that or just, you know, measure it overall. Oh, man. It looks like the Instagram influencer campaign that we ran, we saw 10% more sales than we were expecting to get, than the model suggested we were going to get. That's probably that influencer campaign, right? Because it was for this particular product. We ran it for this long during this time period. And you're going to misattribute some of it. Absolutely. It will not be perfect, but it is a way better model than what you were promised in the attribution era of the first 20 years of the Internet, which kind of did work. I'm not going to argue that it kind of worked up until 2016. 17. It's just broken since then for a whole ton of reasons. Like the privacy thing you mentioned, but 50 million more.

Speaker A: Yeah, it's. Yeah, I think it just seems to become. Yeah, there's more problems and challenges with it every year and it's just like it's time to shift how you think about it. Um, so from being in the performance space and, um, thinking about AI and how that's going to affect paid, um, Perplexity have obviously been very quick in trying to get to market first with some kind of ad offering and trying to, it seems like jump the gun on OpenAI and get something out to advertisers, I've never used their offering. I don't know much about it, but based on the fact that, as we discussed earlier in this conversation, that the market shares that we're talking about here are so small. Is it your like, because m me based on hearing that, I would just think, well, it's not even worth, like, testing investing in this right now, or is there still value to be had in experimenting and seeing how it goes?

Speaker B: Um, so this is. This is why I tell people to do audience research, right? Because the broad macro trend is not your audience's trend. So, for example, let's say that you are, um, selling data to LLMs, right? Like, like ChatGPT and, um, Perplexity and Gemini. Like, these are your customers because you're selling, I don't know, language model data to them from your giant store of content or whatever it is. Well, Guess what? The directors of engineering at those companies and the VPs of, you know, biz Dev who, who do the partnership deals, they use a lot of perplexity. You probably want to be in there. You might find value in advertising directly to them in that place where they pay attention. Now let's say you're selling, you know, uh, time shares to retirees. Well, right, like, man, you know what you probably want to be in? You probably want to be in the American association of Retired Persons magazine. Like, I'm serious. Right? That and, and the two don't cross over at all. And the AARP has, you know, uh, 1/10 of 1% of you know, ChatGPT's market share or reach or whatever. Right. It's not like it's not even close. But your job as a marketer is to do audience research to figure out which publications, which platforms, which websites, which podcasts, which YouTube channels, which subreddits does my audience pay attention to. The macro is interesting. It's great for discussions like these. It's fun to talk about on stages. I do a bunch of videos about it. But what you care about is your audience and your customers. And so uh, I cannot tell everyone that Perplexity advertising is wrong for them. Just like I can't tell everyone that, you know, buying an ad in the ARP magazine is right for everyone. And this is, you know, not to toot our own horn but like you could use SparkToro for this. You could use um, similar, ah, Web has a nice product for this if you like the sort of the enterprise offerings, they have a good one. It's expensive but high quality. I like their stuff. Um, SparkToro is more the self service, like cheap version and yeah, like go use a tool, figure out where your audience pays attention. If you see that perplexity is very high, you know, in the little SparkToro graph of like which search and AI tools are popular. Make an investment.

Speaker A: Right?

Speaker B: Give it a try.

Speaker A: For sure. So I, uh, don't this kind of. Yeah, I don't know whether it's a moot point, but did you see the article earlier this week about OpenAI launching this instant checkout feature and. Oh gosh, yeah. And I just want to get your take on that and whether you think this whole agentic uh, e commerce thing is. Has got legs in it and what's your take on that?

Speaker B: You know what's so weird? It's exactly like Amazon's Alexa one for one, right? Like it is the same thing. Remember when Amazon launched Alexa and they're like, hey, people aren't going to have to go to Amazon.com anymore. They're not going to go to your website. They're just going to say, hey, Alexa, buy me 10 more rolls of Bounty toilet paper or whatever it is. Maybe Bounty doesn't make toilet paper. That would probably be rough. Um, uh. Paper towels. Paper towels. That's the one. And it never happened. They shut down the program. I think last year they shut down the program. They were basically like, yeah, nobody uses this. I don't think. I don't think anybody's gonna use this one either.

Speaker A: I don't.

Speaker B: I don't trust it. Like, it is. They're solving a problem. That is not hard. Yeah, right.

Speaker A: That's such a good point. Like, I didn't even think of it.

Speaker B: Yeah, it's. It's silly. Like, it's such a. Hey, we're gonna do this because we can, not because we should or because anyone wants it. Like, I don't. I don't really have that much trouble going to Nordstrom's website and, like, buying a new Comme des Garcon shirt. Like, you know, like, it's. It. It's easy. That. It. They already have my saved checkout information. I get my Nordstrom points there. I don't want to tell Chat GPT to do it. Alaska Airlines has all my stuff saved and, like, my, you know, whatever. Like, I'm not going to tell ChatGPT to buy my airline. I don't trust them to pick the seats correctly. And if they mess up, who's to blame? Like, do you think Sam Altman is going to be like, oh, sorry, I'll upgrade you to first class next time? No, you know,

Speaker A: I love that. Like, it's just. Yeah, that. Yeah. The problem that they're solving is not a hard problem. And I think there will be. There's going to be people out there who are super users and they use Chat GPT for everything. Just like there was a tiny minority of people. Like, my dad got really into Alexa and he was like, you know, he wasn't making purchases through Alexa, but he was using that a lot when it first came out. But he was in the minority. Like, I've point of pride, never engaged with an Alexa at all. Like, I don't own one. It just feels weird to me to interact with something in that way. And I think, yeah, I would have to agree with you that I don't think the agentic AI E commerce thing is going to live up to the way it's being hyped right now. That's not to say that, you know, some people will use it and there will be a small minority of people that find value in it. But I can't see the, you know, consumer behavior en masse, um, changing in that direction. Yeah.

Speaker B: Also the incentive is just, you know, I think businesses, e commerce businesses in particular, but sas businesses to any, anyone you buy from, right? Service businesses, they have all learned that these platforms are going to, you know, forgive my language, but they're going to F you over, right? Like if, if you own a store, you know, whatever, your Spec Savers, your Nordstrom or your, um, Marks and Spencer, like, they, those businesses have learned that Google was a Trojan horse that eventually screwed them.

Speaker A: Right.

Speaker B: And is trying to take, you know, $0.99 of every dollar of margin that they make. Um, how. Why in the world would they trust OpenAI to do anything else? It just seems wild to me that you would be like, well, I was a sucker last time, but these guys seem trustworthy. Let's do it again.

Speaker A: Yeah, um, it's, it's great. Like, so I think I know, I think I have a idea of what you, you'll say to this. I was thinking about. I've only got two questions left. So this second to last one, thinking about how to get cited in LLMs and have your brand featured in the answers that come up. And the, the one that often gets cited is you. You need to get into Reddit and you need to be engaging in those forums. And I think, like, to your point, the step before that is. No, before you go straight to Reddit, you need to understand where your audience are paying attention and where they spend their time. And you shouldn't just go, we need to be everywhere all at once. Because as you said, you don't. You need to figure out, okay, where are our audience? And then therefore, we need to make a decision of, uh, this is where we're going to focus our efforts and these are the channels that we need to spend time on. So am I along the right lines of thinking there? Like, in terms of how you want to get surfaced in those answers, that that's the right approach? Um, is there anything else?

Speaker B: Yeah, James, long term, I think you're absolutely right. Like, there's no, there's no question in my mind that the LLMs will eventually get smart enough to where they, they figure out the right source of influence for any given search or, or query or prompt or whatever, and then they pull from those specifically, rather than trying to like, Take their whole language model. But right now, if you would like to appear at the top of LLMs, you can spam it pretty easy. Um, and you know, it's like 1990s Google, like teach people you want to

Speaker A: put them m. I want to.

Speaker B: Yeah, sure, yeah. I think my buddy Will Reynolds did a, did a post that was like, basically if you search for, um, in Google, if you go search for like best agency or best, you know, web marketing agency in Citi, uh, in quotes and look for that in the title and then add the name of an agency, you can see who's spamming Google or sorry, who's spamming the LLMs, because essentially what they're doing, they're not spamming Google because it's not Google's smart enough to like, know not to put them at the top. But the LLMs are not. They just crawl and index the web and they are taking any article that gets put up on any website. And so people are putting up, you know, 50 little mini websites with like, best hyphen, agency, hyphen in hyphen, Toronto, you know, whatever. And like they're winning the AI game with that because the AI tools are not smart enough yet to be able to go, you know, specific source to specific source. A, uh, large language model is just that lots of language from the Internet. Look for words that frequently come after other words, tokens that frequently come after other tokens. Right. And so they, if you have a hundred pieces of content on the web from a bunch of different domains that say, you know that and your agency is always at the top. Yeah. When, when you ask ChatGPT, what's the best agency in Toronto, you can totally game that you can come up first right now, that works. And as a result, like, if you think about why that works, you also realize the legitimate ways to do it, which is press and PR, media appearances, podcast transcript appearances, YouTube appearances, like, yeah, this is not the only reason I do podcasts, but one of the reasons I do podcasts nowadays is because I know that I will mention audience research and zero click marketing and you know, all this stuff and SparkToro will also be in there. And so in the transcript of this podcast and dozens like it, chatgpt, Gemini and Perplexity and Claude will all see that content and then they'll go, oh, I should, whenever someone asks me about audience research, I should say SparkToro. And I'm doing that the legit way. Right? Which is going to last for much longer than the sort of spammy way that a Bunch of these agencies are doing it, but both work. You just need to be in the model more times. And, uh, they care a lot about recency too, so don't think you can do it once and then it'll last for years. If they see a big spike in the last 90 days of a bunch of new mentions of one thing over another, that'll make it into the model too. Nice.

Speaker A: It's good to know a, uh, masterclass in spamming the algorithms. Uh, but yeah, I've always been a fan of just doing it the proper way because it's like you're gonna, we know you're gonna be penalized down the line for this kind of activity. Like as you said, like Google's done that time and time again where there's sites. Did you see all the things that happened when ChatGPT first got released? And these people were able to generate huge volumes of traffic to their site through these AI generated blogs and then they all came crashing down like, you know, six, seven months later. And it's like, what is the point of trying to do things this spammy way? It's so much time and investment. But anyway, um, my final question, um, rant, um, is just for any marketing leaders out there who. There's so many new shiny objects and attribution is really hard and you know, it can feel like a hard time right now to be, um, in charge of a marketing budget and deciding what we're going to do. Um, are there any key bits of advice or areas that you'd recommend to focus on so that you don't get caught up in these things that might not actually move the needle, uh, for your business?

Speaker B: Um, okay, this is my top recommendation. It's very hard to do, but if you can bake marketing into your product, you will have way more success than someone who builds a product and then tries to market it. Here's what I mean by that. There are some products where the product itself creates an incentive for people who buy it, consume it, use it, see it, to share it, and talk about it. It's that amplification in reach and sort of virality that we talked about at the start of the podcast. If you can bake that into your product, you will have incredible success and, and you'll just multiply the, the um, value of any marketing that you do. So, uh, you know, here's an example. Like let's say you and I make, um, an incredible new, you know, Japanese steel, beautiful, uh, Damascus, you know, markings on the, on the, on the blade and we do, like, this special etching and engraving, and, like, the handle is made from this rare wood. And it's just the most beautiful knife in handmade by us. And we sell them for, like, $700 or whatever. It's a great knife. It's beautiful. Does it get people talking? Uh, maybe a little bit, but there's a lot of them. There's so many. You know, so much competition. Or let's say you do what my buddy Scott did. Um, I was just playing with the last night where he. He's a. He's like a culinary inventor, and he put ultrasonic technology into the knife handle. And so it. It passes a little ultrasonic, um, current through the blade that vibrates the blade like, 40,000 times a second, similar to your ultrasonic toothbrush, except you can't feel it. You know, if you're holding the knife, you won't. You won't feel anything. So you push this button, and it just glides through food. It's not a lightsaber, but it's not that far off. Like, it's really. It's super fun to use. Like, it's really cool, and it just looks like a knife, you know, with that happens to have a button on the underside of the handle. I can't stop talking about it. Like, anybody who comes over to my house for dinner, I'm like, oh, my God. You have to see this. Like, play with this. Try this. Like, cut through some fruit with it. Oh, uh, isn't that incredible? Like, try it versus this night. You know, I got this professionally sharp, and now try this one. Oh, my God. It's like, you can't stop talking about it. That's what I'm talking about when I say bake marketing into the product. Like, if you can change the product. And so we did this with SparkToro, where SparkToro, the old Spark Toro, had the same data as the new one we launched in May. But, James, the new one is super visual. Like, there's all these graphs. I'm sure you played with it, right? There's, like, all these graphs and charts, and, like, you can see open, you know, ChatGPT versus Google, and, like, here's how much or less it's used. And then because of that, people screenshot it and they put in their presentations, and they put it in their blog posts, and they put it in their YouTube videos, and they put it in their Instagram videos, and like, oh, my God, you know, the growth the last four months has dwarfed the growth from the previous year because people talk about it, they have an incentive to share it. So if you can do that, that is my top number one tip for marketers in any environment. Doesn't matter what algorithm changes happen, what platform changes happen, what AI does. That's always going to work.

Speaker A: Nice. Giving me a lot to think about there. And I think, yeah, we'll take back and what could we do in Lunio to try and build some of that into the product? What data can we put in front of people and how can we make that data look nice and shareable? Really good ideas. I'm going to go and talk to the product team, but I want to say thank you so much, Ron, for, uh, for joining me today. I really enjoy the conversation and I think people will find this one really useful as we look ahead into 2026. And certainly it's given me some, some reassurance and clarity on, you know, I'm not going crazy and attribution is broken. Like it is just, it's not working. Um, before we wrap up, is there anything else you'd like to mention or let people know about before we, before we finish?

Speaker B: Uh, gosh. Well, you know, James, I do run these other two companies, so if you are, uh, an indie video game enthusiast and you want to play the role of a chef in 1960s Italy who, um, you know, fights magical boars in the wilderness and then brings back your pancetta and your guanciale to make your spaghetti alla carbonara at your restaurant, you should check out snackbarstudio.com that game is going to be launching in 2027. Um, and my other company, uh, which is just launching in a few weeks here, is called alertmouse. It's a better version of Google Alerts, so it's free. You can try that out too.

Speaker A: Nice. There's been some good Italian restaurants references in this, uh, in this conversation. And yeah, I think I'm going to have to Google that knife.

Speaker B: You can tell what's on my mind.

Speaker A: Nice. Thank you so much, Ran. Really appreciate it.

Speaker B: My pleasure, James. Thanks for having me. Take care of.