You know, I have a theory about this. So I do actually think that we are far too critical of ourselves. So I also suffered from imposter syndrome a long time ago. I still have it occasionally like on the like which is like what am i doing here i don't deserve to be here whatever. What is the thing is we kinda found a way. Telling others how awesome we are by inventing job titles that certainly mean something like i'm a ceo of something i'm like a vp of this oh my god i'm a c Look at how important I am. Look, I'm managing 35 people. So you're kind of trying to kind of put all of this into a job title, how important you are. But if you would say, look, I am important, then everybody would just, you know, they would just hate you. And oh my god, that's not socially acceptable, but flaunting around job titles totally is.

Yeah. I mean, I don't ever want to be compared to Elon Musk, right? Oh, no. He definitely believes that his shit doesn't stink and it's real weird. Like it's just gotten dark these last few years. It

has, but we're not talking about his stuff and not at Grok AI. We can maybe, I don't know. Well, actually we do a little bit. So because we're talking paid advertising today and the leading question of today is that digital advertising is never a scam. Isn't that true? Isn't that true, Rand Fishkin?

Never. That is a bold statement, I assume. I hope Google is paying you well.

Unfortunately, I don't. So we're going to talk about paid advertising. So what is paid advertising? Paid advertising is I am giving money to a company that has access to a specific market, usually also by serving some kind of content kind of free. And then they're giving me leads back. What is wrong about this model, Rand? Isn't this perfect? It sounds like a fair deal, though. Yes,

I think absolutely it's a fair deal. And I would never say all digital advertising or all advertising is a scam. Advertising works. It worked in the 20th century. I've tested it myself. You can prove to yourself that it works. But inside the world of digital advertising, there are literally hundreds of billions of dollars that flow into that ecosystem that don't necessarily have careful accounting. And anytime you get dollars of that size and incentives of those proportions, you're going to get people and companies that take advantage of them. And I would argue there's lots of people being taken advantage of in the digital marketing and advertising ecosystem today.

Okay, let's talk about it very specifically. So like, let's say I have a little bit of money lying around. I'm a company owner, I have the hottest shit in the internet, I've produced my product is really, really good, but nobody's coming. So I want to buy access to the market. This is one of the biggest problems as of today right now anyways, because it's never been easier to create any product. So the difficult part of creating a successful product is not actually the product anymore, but like getting people to actually look at it in the first place. So this is the point of like where we get paid advertising. And I've been using paid advertising also as a means to test specific AB experiments that just did not have any market traction yet, just to see like what takes better, but most companies are actually buying their traffic to get to make money. So when you're talking about there is not proper accounting, I think what you're talking about is which traffic is real, what is this traffic doing, and would they have done it even if I would not have paid money for it? Can you go a little bit into more in detail in what you're talking about here when you talk about accounting. Yeah,

I think both of those are interesting and important points. So as an example, Leah, you're probably aware, I don't know if all the listeners are aware that years ago, it used to be the case that the two largest advertising providers on the web, which are meta, Facebook, and all their properties in Google, Alphabet, that essentially when they caught traffic that was not valid, spam, bots, your competitors clicking on your ads to try and charge you money, all that kind of stuff, they would automatically identify it, automatically refund you. And that is a pretty reasonable and fair model. And they did this because they kind of had to, because there was a lot of skepticism around digital advertising, a lot of CMOs and VPs of marketing had grown up in the era of 70s, 80s, 90s, even early 2000s style of advertising, which was big brand spend. And so they saw the results of that through essentially A, B testing and Lyft. They didn't look at what today is the gold standard is digital attribution. We'll make the argument. I'll make the argument certainly that should not be the gold standard anymore. So today, Google and Facebook are so powerful. They are so dominant that they have learned, especially in the last three, four years, that they don't have to do this anymore. If a of traffic that should not be counted for varieties of reasons, right? Could be spam, could be bots, could be your competitors clicking on you, could just be the ad appeared in the wrong place, that kind of stuff. They will count it unless your team goes and contests the results. So you have to manually go in, look through your ad channels and interfaces, call out problems where you see them, and then attempt to get refunds for that. They no longer do it automatically. And I think that's so darn sketchy because they know, right? They know that it's happening. They know where it's happening. So that's the first point you made. The second point you made, which absolutely I think is an even bigger problem, is tons of marketers spend billions and billions of dollars on advertising that is shown and that appears in the journey of people who would have bought anyway. And they are afraid, the CEOs are afraid, the CFOs are afraid, the individual marketers are afraid of shutting those ads off and testing whether they add incremental lift, right? Whether you would have gotten those sales anyway because they're all addicted to the growth metrics and the attribution accounting, those pretty reports that Google and Facebook give you. So

what you are specifically talking about is let's say we have a high-intent buyer and that person wants to buy from you or like it doesn't even have to be buying by the way, right? Like so what we're paying essentially for digital advertising is usually cost per click. Let's say I want to click and I also know this specifically also from my bank. This is an interesting one because every time I log into my e-banking system, I'm going to click and I'm clicking actually on a paid link that they are buying on Google because it's on the top. Yep. Where does it appear? Because I'm just typing in like UBS. It comes up, I'm clicking on it, boom, I'm credited. So let's just talk about the scale of this particular problem. How big would you say this actual, whatever we're going to call it, malpractice or whatever? How big is it? So if I'm spending $100,000 on my advertising campaign like this, how much do you think is largely falsely attributed like this? It

varies massively. There's no one answer to this question. Different sectors, different fields, different types of ads. If you're running, for example, primarily search advertising on unbranded keywords, that percent is probably pretty low. If you're running it on branded keywords, that percent is probably quite high. If you're running retargeting and remarketing campaigns, very high. If you're running display ads on social to new audiences that have never heard of you before, it's probably relatively low. So it's all over the place. And I don't want to make some sort of claim like half of all your ad dollars are being wasted. But I do think half or more of all ad dollars collectively are probably being spent on ads that are shown to people who would have purchased anyway or would have converted, would have signed up for the email list, would have visited the website, et cetera, et cetera. And that number is a little hard to know, but my guess comes from, is moderately educated. It comes from analyzing sort of where traffic comes from and where it goes to, you know, SparkToro is powered by a clickstream panel. And as a result, like we, you know, I run lots of studies with you and I talked about them on a previous interview series that we joined together. And so yeah, you can feel that tons of businesses when they decide to actually do this, Leah, they're they shut off their ads for a month. They see 90, 95, 98% of the conversions, the traffic come through. And that is pretty scary. And it tells me one thing, which is you have to have the courage to test shutting off your ads.

You have to have the courage to shut off everything that runs in your business except maybe salary payments. I think whatever

the powering the product, you probably want to keep that on too.

And no, but I think this goes to a very good point in that it is very easy to implement processes and budgetary spends that are just recurring. That's just how it is. It's very very easy. Now, in the case of a CFO that comes into a company and looks at kind of positions to pay for, they're like, hey, does somebody use this tool still? Do we need that many company cards? That is easy. But for marketing spend, this is extremely difficult, right? So like, here's the actual problem that comes afterwards. I think, by the way, I agree with you, but from a completely different point, I'm not sure why they've ever heard this one as well, but you probably have. But like, if we talk about MQLs.

Sorry, what was the acronym? If

we talk about marketing qualified leads. So qualified leads that are coming from marketing in some kind of way, whatever they are defining them as, whatever they do. Typically, this is something like, oh, we're bringing in traffic to the website, and then they need to do something there, and then they are MQLs. And this is usually used as kind of like the counterpart to sales, where you're being measured like, hey, Rand, I want you to bring us 100 MQLs next month, and then you're successful. If you don't, then you have failed as a marketer. The problem lies in the definition of these MQLs. So a lot of companies that I have been working with is that they're defining this, like what the traffic is that comes in without any check on intent. What I mean with that is that I like to know as a company leader, in theory, how many dollars do we spend per real ICP that reaches a specific moment in the product, which tells me, and that I can measure, yes, they understood what we're doing. Usually, this is through product usage. It's not always possible, of course, in every product. But that's the gold standard because you cannot fake that. There's no bot going through in terms of like, you know, like, oh, now I'm going to click on this particular button and then I'm going to get this registered. Because it always depends on like, how do you define this kind of aha moment and so forth. So this is a very interesting one to me, because then you're also not falling for this trap of like, oh, look how much traffic we had on Google Analytics. Google Analytics, in my opinion, has deteriorated, at least in my life as a product person, from 20 years ago from being one of the most valuable tools in the world to understand what's happening on your website to absolutely useless. I do not look at the time that they spend on the page. We have so many companies who are doing analytics where you just have to create your custom events. Because it's just not good enough.

It's really frustrating. It's become, I would say, kind of a piece of crap software that is excellent at one thing, and that is integrating with Google Ads, which almost every company has to run. And Google knows that. And so they can essentially use their monopoly power in advertising, especially search advertising, to make sure that no other analytics tool can compete because they won't provide the full API into other analytics. I wish the Justice Department would look into that because I think that's a clear violation of the Sherman Act here in the United States, but I won't get into legal theorizing. What I will say is if you are attempting to carefully attribute which ICPs, ideal customer profiles came to your tool, your product, bought from you, signed up for your email list, whatever it is, and you are attempting to say, oh, they came from this channel. We should attribute some or all or most of the acquisition to this channel. You're wrong. It doesn't exist. I'm very

passionate about this particular topic, specifically in B2B when we start to get a little bit upmarket. So we're leaving this fear of one people businesses, right? So I'm a one people business, more or less. Okay, I also have an assistant. But the moment you go a little bit above that, you have a couple of people in a company, and you try to figure out what made someone buy things get incredibly complicated. Is it first touch? Is it last touch? Is it middle touch? Is it multi channel attribution? I think there is some kind of wisdom in there somewhere. But it is really, really, really, really difficult. And it's mostly stupid to think that someone is coming from one method only. The only thing that you measure is, what was the last

touch that they did to get to you? And that alone is not useful. It's not useful. And it's lying to you too, right? So a huge amount of the web's traffic, if you go into any, I don't care if you use Google Analytics or Omniture, whatever you want to use, go to your log files. It will say that a huge percent of your traffic came from direct or from Google search. Those two are going to be at least 60 and often 90% plus of everyone's traffic. And what's really happening is someone especially is a B2B example, someone talked about you in a Slack community. They clicked a link in that Slack community and Slack intentionally removes the referral tracking and so you cannot see any referral. So it shows up as direct.

Or someone

says, oh man, have you tried SparkToro? It's this great tool. And so that person goes to Google and they search for SparkToro. And now Google gets the credit. Right? And I mean, just ridiculous, right? Insane. Or yeah, the third one is they see in their chat logs, SparkToro, someone's typed that in. So now Facebook's keystroke logging is like, great, excellent. Let's make sure we show them the Facebook ads as they're browsing Instagram or Facebook or the web. Google ads does the same thing. And then Google or Facebook gets the credit for showing you the ad when in fact it was just a recommendation from a coworker on Slack or an email or whatever it is. You will never be able to know true attribution. And if you're trying, you're fooling yourself. I've never seen everyone who comes to me and says, oh, rant, we have this econometrics marketing mix modeling system. And it's powered by this AI stuff. It does incredibly advanced math to sort out. And I'm like, you know, yeah, it's like a dog's butt. Cat's butt. I believe that you are able to get some reasonably good data if the only thing you're trying to measure is which ad spend was most effective on particular ad channels. But if you're trying to actually figure out how much did PR or brand or organic content or word of mouth or news media or being on a podcast or going to a conference or event or the million other things that people do in the world, you'll never know. And you will undercount the important stuff and you'll overcount the digital advertising and that's part of where this scam happens.

So I can give a very specific example on what you just said is absolutely true also in my experience. So I've been starting to really start the content mill for my own brand about three years ago, like maybe two to three years ago. You're one of the few people that actually has more followers than me on LinkedIn. Shame on you for that. Sorry. And I did the math at some point, right? So like, how many followers

do we have? You

have, think you have 150,000 or something. I don't know where I'm at at the moment. Like, I think I'm 50, I'm at 56,468 or something. Not that I care. And no, but like, so here's the thing. This was an extremely arduous exercise to also kind of build reach and so forth. But I always believed in that this, at some point, will work and so forth. You know, like I'm also coming from this kind of school of, I mean, it's not the school of Adam Robinson. I think he started after me on this stuff as well. Like, you know, like my building in public, but you kind of believe that it kind of works. Yeah, none of this in there would help you with attribution. The only thing that you can do is, is if you have only one channel, and you do only one thing, then you can say, okay, it must largely come from this. And then you also have some word of mouth as well on the side, right? Like as soon as brand is starting. So I did the math at some point. And by all means, like my consulting clients that I have, I would count as an enterprise business seller, right? Like my B2B business is enterprise buyer. So my ACVs are at least 100,000. So they spend at least 100,000 per year on me for every advising client that I have. Now, if I'm breaking this down to the impressions that I'm getting per year, then you end up with, well, per 1 million impressions that I have on LinkedIn, I get probably one advising client. Good luck figuring out now whether if I increase with paid advertising by 50%, my impressions, that actually have money in the sense of like they need to spend a lot of money to get one kind of buyer. It's impossible. You need to trust the process. And yeah, so I can absolutely attest to this with you as well. Like, so I made today also my strategy for the next year and like how I want to build my own business further, right? Like how can I push this even further? How can I scale it? And I said together with my assistant, we actually defined only numbers that we were like, you know, like we're going to go for the followers, we're going to go for the subscribers that we have and so forth. but we're not going to directly target them because we could always buy them. We're going to observe them as a result of us increasing the amount of content that we produce. That's the connection that I believe in. The rest is garbage. It really is. It just does not work. My experience has been that the best marketing

channels are the hardest to measure.

Because they're hard to measure. Because if they would be easy to measure, the cost will come down. That's the thing. It's just basic.

Yeah. No, no. I mean, basically, so whenever something becomes measurable, especially when the business behind it becomes attributable, and there's an incentive to prove that attribution. You get a huge amount of taking credit for sales that already would have happened anyway. And you can see this, I don't know if you read the Department of Justice's documents and emails that they got from Google, but looking through there, you can see the ad executives telling the rest of the team, telling the leadership team at Google, hey, do you want to make your quarterly numbers? Because if you do, we're going to need to do these sketchy things that essentially take credit for sales that would have already happened, show people more ad results, reduce the quality of the organic results, make it harder to identify ads, force people to scroll down further, lazy load the organic stuff and quick load the paid stuff. Everything that you can imagine to juice the numbers. Unfortunately, because they're a monopoly, we have to go for it. If it was like, oh, Bing and Google and DuckDuckGo are each a third of the market, they would all have to innovate, right? They'd all have to be as high quality as they possibly could. They'd have to do a great job of showing you, hey, these sales probably would have happened anyway, because they would use that as a unique value proposition to get customers. Google's unique value proposition is where are you going to go? Right?

That is very true. That

sucks. I

think I was writing an article, I think it was my first one on my AI blog where I was just thinking or trying to figure out how does Google's business change because of AI models that just avoid Google by using Google, but without going to Google or in general. And then the broader point is we will have models, and this is already happening, that are not going over the interface that the product has. So you don't even need to engage with the product and where you can just get the value out and this kind of bypasses ads and some other stuff. So it's really very interesting on how that ends up. But so we've been. Yeah, sorry, go for it.

Well, I would just point out, right, that what we call AI today is extraordinarily similar. Maybe it's a slight advancement from the machine learning models that we had in 2011, 2012. It is more advanced, right? And machine learning has made a bunch of progress and the large language models that we have today are in advancement. But Google has been using what we call AI for 12 years. So is Facebook, the algorithmic feeds that Twitter and Facebook and Reddit and LinkedIn and YouTube use to addict us based on engagement. That's all machine learning, the same stuff that underpins large language models in AI. When people are like, how is AI going to change the web? It already has. You're seeing it now. It's been happening for a decade. I don't know what we're talking about here. What do you mean? How is AI going to change? This is the AI. It's already there. So is more stuff going to keep coming? Sure. I

think we're just eliciting a lot of emotion with people when they are confronted with a new interface to interact with something. And I do contend as well, like that AI has given me more access to things that I went were completely hidden for me before than it ever has, like understanding legal contracts, checking bullshit in seconds rather than hours. You know, gish galloping is much, much harder, like this logical kind of behavior of people that they just like give you a slew of arguments that are just stupid, and you just cannot address them anymore. So I think there's also good things in there. But before we go into this, like, so been ranting for like 25 minutes right now against paid advertising. So let's become a little bit more practical then. What would you suggest to a typical, I don't know, B2B SaaS company that is selling to other clients, they have some money or they're trying to figure out their go-to market, what should they do then? I mean, you got a point. There might be something fishy. So I think what

I would love to tell them is, gosh, you should really be investing in things like organic content and organic social and press and PR and media coverage and events, all this type of stuff that I love to do that's been very effective for us. We haven't spent any money on paid advertising and SparkToro has turned into a nice little low seven figure business, but that is unfair. I don't think that's actually the right advice. I think for many B2B companies and B2C companies, advertising is where you should spend your dollars and efforts, but what I would urge you to do is A, like we talked about, have the courage to shut off your ads and see how many of those sales would have happened anyway. And B, you've got to be thinking about digital advertising outside of the attribution that Google and Facebook are providing to you, and Apple and Amazon and all of these. And instead, I would argue what you should do is treat it the way Coca-Cola treated billboards in 1965. So in 1965, Coca-Cola would run a billboard with one message in Cleveland, Ohio, and they would run another one in Cincinnati. And since those two cities are quite similar in size and demographic makeup and amount of Coca-Cola purchased per capita, they would see which one resonated more. And then they would run that one across the rest of Ohio, and they would compare that to the California campaign. And they decide if this was going to be a national billboard, or if they were just going to run it state by state locally and expand it. And the way they would test that is by looking at same store sales lift with seasonality baked in, right? Compared to last year this time, did we see more lift than our model would have expected? Oh, we did. We saw 5% extra lift in same store sales within a five mile radius of where this billboard is. That suggests that billboard really worked. Let's try it in Cleveland. Hey, it worked again. All right, we're rolling it out to Akron. Hey, it worked a third time. You know what? This billboard works. We're rolling it out across all of Ohio. Maybe we'll expand to Illinois. That is a fundamentally 20th century way of thinking about how you test and invest in advertising. And it sounds crazy to say, hey, we have all the sophisticated data. We have machine learning. We have AI models. We've got fantastic amounts of data that you can use. And I'm telling you that you should do it the way Don Draper did it in the 60s. Sorry. I think that's the only thing you can trust. I just don't believe any of these other sources. I've seen too many examples of people shutting them down and getting the same or improved results. You can do it yourself. Go look at the New York Times case study on Citibank. Go look at the Airbnb case study from 2021, where Brian Chesky talked about how they pulled their paid ads down to 5% of their previous spend, dropping hundreds of millions of dollars in saw improved results. Look at the Uber case studies. Look at the Chase Manhattan case studies. There's just too many. Like there's too many of these public and private from almost everyone you know. If you put up something saying exactly what I'm saying right now on LinkedIn, you will get hundreds of comments where people go, yeah, I saw that too. We did this and we saw it. I'm not telling you something you don't know. I'm telling you something that your CFO and CEO won't let you do.