UP541 Addicted To Fake Fan Numbers: When Sport Met The Dead Internet

0:00Are we addicted to the massive numbers?

0:02Dr. Augustine Fou:Yeah. So that's been the kind of bill of goods that we've been sold for the last 15 years. It doesn't mean that it's actually going to drive real sales for advertisers. Some of these advertisers are finally waking up to the fact that the super large numbers don't mean anything. If the vast majority of them are basically bots that are following your account, bot views on your YouTube, right? It's not actual engagement. It's not real people to begin with. So none of that's going to actually translate into real business outcomes. Hello, Richard Gillis here. Welcome to Unofficial Partner. Today, we've got Dr.

0:32Augustine Fu, one of my favorite guests, been on twice over the last few years. He is a absolute expert in the subject of digital ad fraud, the advertising industry's dirty little secret. Most of what you're paying for never reaches a human being. Fu has spent 30 years watching the internet fill up with bots, fake impressions, and fraudulent metrics. And the advertisers funding it often know but keep spending anyway the incentives are too comfortable to disturb if this sounds familiar the same is happening in sport we've built valuation models rights deals and global expansion strategies on fan numbers that deserve the same scrutiny this conversation is about the mechanics of self-deception and why the moment of reckoning may finally be arriving

1:24Listen, thank you very much for coming back on. It's been too long. The episode that we did, we've done a couple of things in the past, and I'll point people towards those. And they have always been really well received. The numbers on them are really high because I always say that your work, it does my head in a bit. Let's talk about you because I want to just sort of set you up with the people that haven't heard previous recordings with you. What's the work that you do? Give us a sort of taste. Sure.

1:51Dr. Augustine Fou:These days I'm doing ad fraud research. So in digital marketing, obviously we run billions and billions of ads online. And the problem is some of those ads are actually not seen by humans. They're being displayed to bots or software programs. And a lot of times the big advertisers are chasing these vanity metrics like, oh, we bought 10 billion impressions and we got 10 million clicks. And all of these vanity metrics are what we call quantity metrics. And unfortunately, that doesn't necessarily translate into sales. So my job is to help them identify which ads are being shown to bots, where the ads go, like which sites and which apps are responsible for that kind of fraud, so that they can actually put those in a block list and not waste any more of their ad budgets on that.

2:38Dr. Augustine Fou:So these days, I'm an ad fraud researcher, and I have my own platform called Foo Analytics, which is in a way similar to Google Analytics, but Google Analytics is used for websites. Foo Analytics is used to measure the ad impressions as well as websites. And bot detection is actually built in. So it becomes an analytics tool for marketers to use and also their agencies to basically monitor more closely their digital campaigns and make sure they're not wasting a whole bunch of money on fake visitors, fake views, fake likes, and those kind of things. Can you give us a sense of the scale of what you're seeing, the scale of the bots?

3:17the, you know, we'll get to the term dead internet in a minute, I'm sure, but just give us a sense of the scale of it.

3:24Dr. Augustine Fou:The scale is enormous and it's almost to a point where it's hard to believe how large it is. But let me use an analogy that maybe folks have heard of or understand. You've heard of these DDoS or distributed denial of service attacks where giant botnets simply overwhelm a site with so much traffic that the site fails. And these include even the largest of sites, right? Whether it's Twitter or LinkedIn or any large sites, they can send so much traffic that they can take them down. So it's that same kind of traffic that is now used to generate ad impressions. So for example, if you have a whole bunch of these fake sites, there's no humans that visit them at all.

4:09Dr. Augustine Fou:So they have to go out and buy 100 % of their traffic. And essentially, you can think of that as just renting time on these giant botnets, because you can basically tell them, these botnets or whoever's managing these, we need 10 million page views, or in fact, 10 billion page views on our sites or sites. And they will just send that traffic. And then when the page loads, the ads load. So the scale of the problem is truly enormous, but it's actually been very significantly underreported because the fraud verification tech companies that exist to try to detect that fraud is basically missing most of the fraud.

4:51Dr. Augustine Fou:So basically it's marked as valid when it's actually invalid. And why is that? Why is that happening? Is that a technical thing or is It feels like there's an incentive for everyone to have a big number. Yeah, that's a perfect segue. Yeah, the incentive actually is kind of counterintuitive because for the past 10 years, a lot of the biggest advertisers and the biggest agencies love buying a lot of digital ads because it makes them look innovative, right? That's literally for the last 10 years. So if there's a lot of fraud and they could detect it, then there's not a lot of inventory to buy because you realize that's all bot generated ad impressions, not humans visiting websites.

5:34Dr. Augustine Fou:So there was an incentive to buy the services of these legacy verification companies because both of them, there's two large public ones, have published press releases that said fraud or IVT, invalid traffic, has been 1 % for the last 10 years. So very conveniently, the agencies loved buying their reports because it said, oh, hey, look, these guys said fraud is 1%. So there's nothing wrong with what we're buying. So that enabled them to continue to buy super large quantities for the last 10 years. But the problem is that's actually not all the fraud there is. I like to say that the 1 % that they detect is basically all that they can detect, not all the fraud there is.

6:18Dr. Augustine Fou:And there's a technical reason for some of that as well. The bots have gotten more advanced over the years, so it's actually harder to catch. But there's actually an even simpler explanation. It's called verification stripping. And both of these companies also have published about that. They know that's happening. What that means is the bots are actually blocking their detection tags so that they cannot mark the bot as IVT or invalid traffic. So the problem with these platforms is that they report no data as no fraud. And that's actually wrong. If they have no data, they shouldn't say that it's no fraud because it could actually be a bot that just blocked their detection tag.

7:02Dr. Augustine Fou:So they couldn't collect any data on the bot. So I think that's a critical kind of nuance that a lot more advertisers should ask them about. If they're paying these companies, these legacy verification companies for detecting fraud, and they're only catching 1 % of it, that's not a very good use of money. So the big advertisers should ask them harder questions like what percentage of the impressions did you actually measure or could you actually measure with a JavaScript tag? If there's a whole bunch, like a large percentage that's not measured, you need to disclose that as not measurable instead of not fraud.

7:40Dr. Augustine Fou:So that's been a major, major issue. And that's why fraud and IVT and valid traffic and bot traffic have all been severely underreported. And that's why a lot of advertisers don't realize how big of a problem this is. See, when I first heard of this, or when you sort of think about it, you think, okay, you've got that chain of, you've got the client, the advertiser, the big brand, whatever they are, whether they're a car company or an airline or whatever it is, at the top end. Well, take me through the chain to me as a punter. Who is in the middle between me and the airline? So the advertiser, right, whether it's the airline, the automotive company or whatever, the larger those advertisers, the more likely they would have used a media agency to buy their digital media for them.

8:32Dr. Augustine Fou:And this goes all the way back to TV advertising, right, where they would go to a big agency, a media buying company and say, oh, you know, get us a really good deal on TV ads. And so the media agency would do that. But in digital ads, a lot of that negotiating power is no longer necessary because every single ad impression is auctioned off. So in programmatic media buying, in theory, every single ad impression is bid at the market clearing price, right? Bid and won at the market clearing price. So it really makes the agency's collective negotiating power completely irrelevant in digital. However, old habits die hard.

9:11So the big advertisers ask their agency, go buy me as many impressions as possible for as cheap a price as possible.

9:20Dr. Augustine Fou:They think they're getting a good deal by getting a very large quantity at a very low price. And the reason they still think that is because in the days of TV advertising, reach and frequency were very key drivers of advertising effectiveness. You want it to reach a lot of people with your message. And then frequency, obviously, you don't want to be too frequent. You don't want your ad to run too frequently. But in digital, the reason that no longer holds, the whole reach and frequency argument, is that we have seen fraudsters and bad guys generate unlimited inventory. So unlike on TV, there's a finite number of 30-second spots.

10:01Dr. Augustine Fou:There's a finite number of 15-second spots to be purchased. There's a limit to the supply. In digital, there's no limit to the supply due to fraud. So over the last 15 years that I have observed digital, there's been a tremendous amount of money shifting into digital from TV, from print, from radio, from offline billboards, to the tune of hundreds of billions of dollars. So you would think if there's that much demand moving into digital, if there's a finite supply, like literally the people going to a legitimate publisher like New York Times, Washington Post, CNN, that the prices should go way up, right?

10:45Dr. Augustine Fou:Because the demand went skyrocketed, but the supply was finite. But that didn't happen. In fact, the opposite happened. At a macroeconomic level, we saw that CPMs, the prices for these digital ads, have declined for the last 15 years. And that means the supply outgrew the tremendous growth in demand. The supply grew even faster. How is that possible? We did not triple or quadruple the number of humans on Earth. We did not go beyond a 24-hour day in our solar system. So how is that all possible? So if you look at some of the Pew Internet data in terms of humans' usage of the internet, social media, mobile, that has all been maxed out since 2013.

11:35Dr. Augustine Fou:So humans' consumption, humans' usage of the internet, social and mobile have all been flatlined. It hasn't grown much. But yet the amount of inventory that is now bought and sold through digital advertising, these programmatic platforms is in the hundreds of trillions of ad impressions. So you can see, OK, all of it's made up out of thin air, even if those fraud detection companies could not detect it. So there's one other data point I just published recently. we looked at 250 public US companies and we looked at their revenue growth from 2009 through 2024. So over a 15 year period, most of the largest of those advertisers did not grow at all, despite shifting tremendous amount of dollars into digital, thinking that they were showing the right ad to the right person at the right time.

12:29Dr. Augustine Fou:So all of this hyper targeting, all this personalization did not lead to sustainable long-term growth and revenue. In fact, when they shifted all that money into digital, they started to neglect branding and awareness. So now you're targeting the people who have heard of you before and who have purchased from you before, and you're neglecting the other 99 % of the market who simply don't know who you are. So they're forgetting to do the basic blocking and tackling of advertising, where you have to fill the top of the funnel by generating awareness first, and then you can use digital to start harvesting some of that demand.

13:10But by overspending in digital, some of these largest advertisers have simply not grown

13:16Dr. Augustine Fou:because they've neglected to do the awareness part. So the debate within the advertising effectiveness space, whatever, you know, is that presumably has fallen hugely have they how do they explain the failure of advertising to reach its audience to generate you know they they haven't in fact uh we only now have the 15 years of revenue data to show them that the compound annual growth rate has been nearly zero right so what has happened is that the digital reports the attribution reporting that you get from the platforms from which you buy the ads appear to show very high ROAS, return on ad spend, or ROI, return on investment.

14:02Dr. Augustine Fou:But unfortunately, it's not fraudulent, but they're just incorrectly reported. And what I mean by that is they are claiming credit for sales that had already occurred. So very few advertisers have done the following, but I encourage them to do so. if you take up all of the conversions or sales that one platform like Google reports, and then you look at all the conversions and sales that another platform like Facebook reports, and then you look at a third, say YouTube, you're doing YouTube advertising, maybe TikTok, and you look at all the conversions that they're reporting to have driven, you're probably going to get four times the number of sales you actually got in that time period.

14:47Dr. Augustine Fou:So essentially, the attribution reporting is flawed. So it makes it look like you're getting sales while you're doing the digital advertising. It's actually not causation. It's not because of your digital ad you cause those sales. And that gets to a very fundamental thing that advertisers need to refocus on, which is incrementality, right? Were those sales incremental, which means would they have happened in the absence of the advertising? And the answer is most of the sales that these platforms are claiming credit for would have happened anyway. It wasn't due to the advertising. So using just a very simple offline analogy, if you think about soup and soda, if you think about paper towels, toilet paper, they're sold in grocery stores.

15:35Dr. Augustine Fou:So over here, you're doing digital advertising. Over in the stores, the sales of soup and soda, toilet paper and paper towels keep happening. so now you're happily attributing those sales to your digital advertising yeah what most advertisers have not done is actually turn off that digital advertising to see if the sales of soup and soda toilet paper and paper towels continue one of the case studies was procter and gamble wasn't it that was a moment and i was wondering i was going to ask you about what the impact of that was because you had this sort of hiatus where they you know massive probably arguably one of the biggest advertisers in the world just tell me about they they turned off their digital ads or yeah they They spent$2 billion annually in digital advertising.

16:18Dr. Augustine Fou:They turned off$200 million, so one-tenth of it. And basically, they saw no change in sales. Now, all these other proponents of digital found other explanations for that. Oh, yeah, it's not related, and it was like cost savings or whatever. So they tried to explain it away. And my question for them is, okay, well, if there's no change in the sales, why didn't you do another one? So they never followed that up with anything else. But since then, we have other case studies, right? So before that, there was an eBay case study, where they basically in the US turned off all of their search spending, paid search spending in the West Coast.

16:52And they saw that there was no change in the traffic coming from the West Coast.

16:57Dr. Augustine Fou:Okay, so they were essentially bidding on terms that people would have used to search for their search for items on eBay and come to eBay anyway. That was a Harvard Business School review from 2011. And since then, Airbnb has done the same. They have said, and there's a work article on this where they said, we have overspent on performance marketing. We're actually going to cut all that out. They actually did that during COVID, right? They stopped all the performance advertising and then simply observed that the traffic came back to 95 % of their original levels with no additional spending. So they basically from that point forward said, let's go back to branding.

17:36Dr. Augustine Fou:And you see Airbnb doing a lot more TV ads, right? And that's actually allowed them to grow again. And there's one other case study off the top of my head. It's from Adidas. They actually said, we've overspent on performance marketing and we're going back to branding. So there's been a few of these case studies. They're kind of sparse, but those are good lessons where the marketers themselves have realized, you know, when we're trying to do this hyper-targeting, this ultra-personalization, trying to get the right ad in front of the right person at the right time, it's too surgical, right? There's a lot more people who are simply not aware of your brand yet, or who need to get inspired by your new products, right?

18:16Dr. Augustine Fou:They're not even aware of the new products, the new shoes that Adidas has. You need to get your ad in front of those people. And I've actually written an article about this. I call it the green field, right? Instead of targeting the 1 % that has already purchased from you before. What about the other 99 %? If you actually just start showing ads, the other 99%, that's a group that's 100 times larger than the people who have already purchased from you before. So even if 1 % of those converted, right, you basically doubled your sales. So the message for advertisers is make sure you don't neglect doing the greenfield advertising, right going back to brand and trying to get your ad and brand in front of other people who have not purchased from you before so some of that's going to solve the fraud problem as well because you got to make sure your ads are shown to humans and sometimes tv or maybe outdoor billboards are just as effective or more effective than the surgical digital ads yeah yeah so loads of questions so one question is there's a question about the client as in they are looking at this picture and i used to think when i worked at a you know within an advertising agency people would quite often say oh well the client doesn't either doesn't care doesn't understand and let's keep them blind to this little ruse that we have because we're making a lot of money at the agency and increasingly you sort of think no they're not stupid they know it but there's an incentive for them also to keep this game going and yeah just to jump it over to sport just the analogy of sport is is fan numbers so there's a whole sort of question of how many fans a club has how many a league has we're about to have the nba come over to europe and open up franchises they're going to be someone will pay 700 million for them predicated on a global fan base survey and i can sort of it's analogous it's not you know it's it's similar but not yeah i i can i can tie those two together because over the years, you remember when Facebook first came out, a lot of these brand advertisers, I remember most famously Burberry, they set up a Facebook page.

20:24Dr. Augustine Fou:It's like, oh, well, we want to have more likes and fans. So those big advertisers literally went out and started buying likes. Initially, it started off as the right idea. Let's pay a social media agency to get us more people to our page and have them like our page. But then it's a, oh, well, we can just go buy likes by the thousands, right? There are vendors who are willing to sell that to us. And so essentially that gave rise to the first cottage industry of fraudsters who would set up a lot of Facebook accounts so that they could just go and supply the likes to whichever brand wanted to buy them, right?

21:01Dr. Augustine Fou:So that was the first wave. And over the years, whatever metric, right, whatever quantity metric you wanted to gain, you could literally go buy those things, So you can literally go online and Google for this. I want more views for my YouTube channel. Just go buy them. I want more followers on my Instagram account. Just go and buy them. I want more listens to my stream on Spotify. Just go and buy them. It is 100 % bot traffic. It is 100 % automated. It's not real. but over the years when everyone has come to accept those numbers like you said more fans bigger numbers mean higher valuations everyone's bought into that concept of bigger numbers equals better i mean a lot of that's rooted in just video gameplay right so higher scores are always better than lower scores right higher scores win so it's that basic kind of mentality that now pervades marketing, sports marketing, whatever.

21:59Are we addicted to the massive numbers? 100%. It's really hard to come to market with a story that hasn't got an enormously, incredulously large number.

22:09Dr. Augustine Fou:Yeah. And the growth rates, right? So that's what Silicon Alley has based all of their valuations on, right? They can be losing money, but if that hockey stick growth rate looks like a hockey stick, right? Shooting straight up, they can get future valuation that's worth a ton of money, right? So that's been the kind of bill of goods that we've been sold for the last 15 years. It doesn't mean that it's actually going to drive real sales for advertisers. So I think you're right. Some of these advertisers are finally waking up to the fact that the super large numbers don't mean anything. If the vast majority of them are basically bots that are following your account, bot views on your YouTube, right?

22:47Dr. Augustine Fou:It's not actual engagement. It's not real people to begin with so none of that's going to actually translate into real business outcomes so one of the the counter then is brand over performance marketing so that's one solution to this but in a way brand is a sort of ruse in itself isn't it you're you're saying right okay as long as we do some big like nike is always that adidas story nike similar type of thing where there's a you know one ceo comes in and says we're gonna stop doing we could do more performance marketing yeah digital blah blah blah and then he gets the sack and then the next person comes in and says no we're going to go back to do something different storytelling and brand and brand is a bit of a ruse because it's sort of a bit it's intangible and therefore never promises to be measured in quite the same way the promise of the internet and going back to its initial you know ad tech initial iteration was attribution and it was that person is there and you're seeing they're receiving the ad and therefore you can trade with them you know who they are all of those stories i've you know i'm assuming you've lost faith in that story but it's also brand is also problematic in some ways because i could sort of see okay i'm going to come in now and an ad agency she's lovely because i go in and say right i'm going to sell them these massive great campaign ideas and they're going to just go that way because they're not going to get caught out in the same way possibly but they don't really need to be yeah in the same way i think you know brand this goes back to one one more of the misaligned incentives so i've seen this over the years where the marketers uh if they're given a budget say 50 million dollars 100 million dollars their job is to spend it before the end of the year to spend it all before the end of the year and they are always definitely afraid that they are not able to spend it all because then that will mean they'll get less budget to use next year.

24:43Dr. Augustine Fou:So in some cases, again, it's a philosophical thing. I've seen too many marketers are just desperately trying to spend it all instead of actually looking at what it drove for them. And so similarly, the agencies are set up, their job is to help the marketer spend it all. So there's been cases where you have 10 % of the funds left over in the last month of the year, I've heard of these cases where the agencies are desperately going to the programmatic platforms and say, can you help me spend this last 10 million because I need to spend it before the end of the year? Just make some stuff up and let me spend it.

25:17Dr. Augustine Fou:Those are all misaligned incentives. And it's usually hidden from view because nobody's willing to admit that. So there's a couple of things to think about here. And I really differentiate between the theory of digital advertising and the reality of it. So the theory is the right ad to the right person at the right time. That all still works. You can do that. And the best way to do that is through search. When the person types in the search term, they are telling you what they're looking for, when they are looking for it. And if you can get them either organic content or literally a paid ad, both of those are useful to them at that time, because that's when they need that information to inform their own purchase decision.

25:58Dr. Augustine Fou:But that doesn't happen that often, right? It happens when that person is ready to buy and are looking for something, right? So the other part of the branding is that, oh, now when you say, okay, we need to go back to awareness and branding, then the agency has interpreted that to say, oh, let's go buy you as many YouTube views as possible, right? YouTube ad views as possible, as many Facebook, Instagram impressions as possible. And that's where they actually need the fraud to get those large numbers so that they can actually spend all that money. So that's where the branding, in theory, you want to go back to more awareness.

26:36Dr. Augustine Fou:But in practice, if all of that awareness is just bought accounts and fake views, it's not going to generate the outcomes. So again, the advertiser still has to be very wary of what they're buying and still have to ask very hard questions of the vendors, right? Whether it's the agency, agencies, the ad tech companies, or whoever they're buying from to make sure that they're actually getting their ads in front of humans so that it can have some kind of business impact. So I'm going to push this then into the, there's an obvious AI question. Now I am spending a lot of time on Claude and I'm thinking, okay, I'm asking that questions.