Latest / Signal & Noise: Bringing Clarity to Media, Data, Tech and AI / Ad Fraud, Part 2: Dr. Augustine Fou on Hidden Fees, Phantom Outcomes, and the Agentic Al Trap
Transcript
- 0:14Welcome back to Signal and Noise.
- 0:16I'm Brett House, joined with my by my Co host area Longacre and
- 0:21our special guest, Doctor Augustine Fu.
- 0:24Doctor Fu has been on the podcast before.
- 0:26Welcome back, always a thrilling conversation.
- 0:30For those that don't know Doctor Fu, he's one of the most
- 0:32persistent and provocative investigators of digital
- 0:36advertising fraud. 3 decades you've worked at the
- 0:39intersection of digital marketing, cybersecurity
- 0:42measurement and fraud detection. Everybody knows you that you do
- 0:45a lot of posting on LinkedIn challenging assumptions,
- 0:48benchmarks and black black box reporting the industry has long
- 0:51relied on. You're the founder of Foo
- 0:54Analytics and analytics and media verification platform,
- 0:57built out marketers and publishers independently.
- 0:59Expect campaign quality and fraud and other things.
- 1:03And earlier in your career, for those that didn't listen to
- 1:05episode #1, I do encourage you to do that.
- 1:08You are the chief digital officer at Omnicom's Healthcare
- 1:11Consultancy Group, right? And you taught digital strategy
- 1:15at NYU and being in Jersey, Rutgers.
- 1:18That's great to hear. And, and you really built a
- 1:22reputation. I think, you know, Rio and I
- 1:23certainly are well aware of it. We, we respond to a lot of your
- 1:26posts, whether it's on X or on, on LinkedIn as one of the, the,
- 1:31the first people that that was really willing to publicly say
- 1:33that the fraud problem was a lot larger than what the Ana and
- 1:37other organizations were reporting as the classic 1%
- 1:41rule. And you're not.
- 1:42You didn't just say it. You.
- 1:44You proved it. And you continue to prove it
- 1:46again. He brings receipts every time.
- 1:48And again. And you show it, and sometimes
- 1:51you get in a little trouble. So we promise to keep you out of
- 1:55trouble. Doctor Boo, welcome again to
- 1:58Signal A Noise. Thrilled to have you.
- 1:59Thanks for joining. Thank you, Rio.
- 2:01Glad to be here. We got some important topics to
- 2:04cover and things the industry may not be comfortable talking
- 2:09about a whole lot, but we're going to talk about it today.
- 2:13Yeah. And did I cover your background
- 2:15well enough? I mean, I think, I think for
- 2:17those that that don't know yet, yeah.
- 2:20Let let me say a few words. So I I've also spent time on the
- 2:24agency side at both Interpublic Group and then Omnicom and I've
- 2:29been on the client side and American Express doing digital
- 2:32marketing for their small business cards division.
- 2:36There we go. Awesome.
- 2:38Well, great context again. Thrilled to have you back.
- 2:40For those who did not check out the first episode, I don't know
- 2:42if it's required listening to, in order to listen to this, but
- 2:44they're definitely related because the, the first episode
- 2:47we really covered what I would call the fundamentals, like how
- 2:49much out of fraud is out there challenging some of these like
- 2:52industry beliefs or perceptions that if there's not a lot out
- 2:56there, there, there is a lot. We, we, we talked about that.
- 3:01We talked about how it's even going to places that are may be
- 3:03considered safe, like CTV, right?
- 3:05Where premium and what's quality inventory can sometimes be be
- 3:10fake or or not quality or, or really AI generated slop, right?
- 3:14We talked about that and how the incentives, a lot of times it's
- 3:17really an incentive problem that's driving a lot of these
- 3:20behaviors and decisions people make.
- 3:21So this time we really wanted to go deeper, really look at the
- 3:23economics and the accountability question, right?
- 3:26We wanted to look at that fraud is still endemic and I think we
- 3:30all agree it it is. And if the open web is where a
- 3:34lot of it hides in plain sight, then who's actually making money
- 3:39off this, right? Who's scraping the dollars out
- 3:40of the system? How much of that is being
- 3:42obscured by blending averages by blacks box reporting, by just
- 3:48general like lack of transparency?
- 3:50And then there's that recent audit drama, which we're going
- 3:52to talk about a little bit with, with the DSP to put a brighter
- 3:57spotlight on the question. Like many marketers are, I think
- 4:01still, Brett, you mentioned this like afraid to ask what happens?
- 4:04What, what, where's the gap between what people are paying
- 4:07and, and how much is actually taken out of working media?
- 4:11You know, if we, I mean, I read a study recently that's around
- 4:1339% actually goes into media. The rest is siphoned off, is
- 4:16scraped by different players, right?
- 4:19Some of it fraud, but a lot of it, not a lot of us just take
- 4:21rates and different, different players take putting their hand
- 4:24in the, in the cookie jar, right?
- 4:25So we wanted to talk about this. We, I really think this matters
- 4:30because if we want the open web to succeed and we want it to be,
- 4:35we want it to stay alive. And it's under a lot of pressure
- 4:37right now. If, as you say, Doctor Fu, this
- 4:40is where a lot of the fraud, fake inventory leakage and
- 4:43hidden extractions remain most endemic and maybe even hardest
- 4:46to see. And then you add like these
- 4:48recent changes in agentic AI and how it's easier to create more
- 4:52content, it's easier to mimic human behavior better at scale.
- 4:56I think this if anything the stakes are going up and if the
- 4:59source of truth is already compromised, automating this
- 5:03with agentic training. These guys just.
- 5:06Calibrate the. The take rate in the front.
- 5:10So you you have said on many different topics, but let me try
- 5:13to break it down. I'm going to start with an
- 5:14example that's based on my concrete data.
- 5:17So a year ago I ran an experiment for Ozone Group,
- 5:21which is a bunch of UK publishers that includes the
- 5:24Guardian, dailymailbbc.co.uk. And the main question I wanted
- 5:30to answer was exactly that, which is what portion of the
- 5:33dollars that I paid as an advertiser actually went to the
- 5:37publisher. So I basically played the role
- 5:39of an advertiser. I asked them tell me the CPM
- 5:42that's going to be above your floor CPM.
- 5:44So I make sure they clears and I can get the ads.
- 5:47And I executed the campaigns in DB 360, which is Google's
- 5:50platform. And I specified A1 domain
- 5:53inclusion list for each of the publishers.
- 5:55So I'm going to use ABBC example. sobbc.co.uk was the
- 6:00single domain I put in the inclusion list.
- 6:03I wanted to make sure first of all, how what percentage of my
- 6:07ads actually made it to the one place that I wanted them to go
- 6:11and the ultimately what what the publisher made.
- 6:14So I bid $1.50 CPMS. Can you guess what the publisher
- 6:20actually got? $0.10. $0.50 CPM.
- 6:25OK. It's right.
- 6:27In between those two, it's $0.15.
- 6:30Oh, there we go. I was a little closer.
- 6:31I think I won that. Yeah, I, you know, I did a
- 6:34little homework ahead of time, but $0.15 and this was the total
- 6:38CPM was this incremental? See this was.
- 6:41That wasn't a winning bid, that was how much they actually got
- 6:44in the end, right? Yeah, exactly.
- 6:45So I paid it all of 50 CPMS and they got $0.15 of it.
- 6:51And you verified, validated. That yeah, 90% of it went
- 6:55somewhere else. And the reason it's that bad?
- 6:57Was that supply path, was it an issue or is it just endemic
- 7:01across? I guess you're going to explain
- 7:02that now. I'm sorry I cut you off.
- 7:03Yeah. It's it's a bunch of different
- 7:06issues, but the the key issue is that the publishers never get to
- 7:10see what the advertisers bid on their inventory and then the
- 7:14advertisers never get to see what the publishers got for
- 7:18their inventory. There's other issues which I'll
- 7:20get to in just one second. In this case, this was supposed
- 7:23to be the simplest possible scenario with the fewest supply
- 7:27paths, right? Because we all know that, you
- 7:30know, people are doing SPO supply path optimization because
- 7:32when you have hundreds of thousands of different supply
- 7:35paths, obviously the bad guys can get in there and steal your
- 7:38money. Now one of the key issues.
- 7:41So not only the publisher only got 10% of what I paid, but
- 7:45there were there was another issue which is the supply path
- 7:48problem. So from the data that I got from
- 7:51DV360, I said how many supply sources actually did I buy from
- 7:55because I didn't know ahead of time, right?
- 7:57This is all put out. So it reported 9 different
- 8:00supply sources. I went back to the publisher and
- 8:03asked how many SSPS do you actually sell through?
- 8:06They said 5. OK, So what the heck are those
- 8:08other four that had no relationship with the publisher?
- 8:13Why the heck are they allowing thedomainbbc.co.uk to be sold?
- 8:17So let me reemphasize none of this is BBC's fault.
- 8:21Yes. So they, so they have
- 8:22relationships with the DSPS, the five DSPS that BBC chose, right.
- 8:27So they must have relationships with them.
- 8:29No, these are SSPS. I'm sorry SSPS, sorry.
- 8:32So they're just adding OPS and writing adding?
- 8:34Yeah, they're adding their own markups.
- 8:35Right. So in the, in this particular
- 8:37case, these SSPS, the four others that don't have any
- 8:40relationship with this publisher, were allowing bad
- 8:44guys to pretend to be bbc.co.uk. OK, This is called domain
- 8:49spoofing and that basically diverts money away from the
- 8:52publisher in a way they can't even see because it never got to
- 8:56them. So sometimes the publishers are
- 8:59saying, OK, we get accused of high fraud.
- 9:02We look at our site, there's no bots or anything.
- 9:04We're not buying traffic. What is going on, what they
- 9:07don't understand is that all of that high fraud, it could be
- 9:09100% bot traffic is occurring on the fake sites that are
- 9:13pretending to be them. OK, so let me let me be clear on
- 9:17how how this kind of grift happens.
- 9:19If the SSP doesn't actually catch the domain spoofing at the
- 9:23very first hop, it will become laundered by the second hop
- 9:28onward. Because in that very first hop,
- 9:31if the fake site pretend or says it's bbc.co.uk, they're still
- 9:36going to leave their own seller ID in there because the fake
- 9:39site wants to get paid. OK, so they're just lying about
- 9:41the domain if that discrepancy which should be caught by ads
- 9:45dot TXT. If as a THT were not enforced by
- 9:49that very first SSP, then by the second hop some legitimate
- 9:53reseller could put their own S seller ID in there.
- 9:57So then. Which is what you mean by
- 9:59laundered. It gets obfuscated and suddenly
- 10:01you. Can't see it.
- 10:02So by the second hop onward you you won't be able to catch that
- 10:06discrepancy. So the problem is that a few of
- 10:09the s s PS are literally looking the other way and bad guys know
- 10:13exactly which s s PS to onboard their fake bid request.
- 10:16Once it gets on and it starts distributing through the
- 10:19ecosystem because there's so much reselling going on, then
- 10:23you lose track of where it came from.
- 10:24So to kind of bring this back to the BBC story, all they had to
- 10:29do was go to those 4 SSPS that they had no relationship with
- 10:32and said just block our domain because no one from your SSP is
- 10:36allowed to sell that because we don't have a business
- 10:38relationship blocking that domain.
- 10:41Bad guys can't pretend to be them.
- 10:43So we solved that but. The publisher, like the
- 10:45publisher, wouldn't even know these SSPS are working with them
- 10:48at all. They might not even know that
- 10:49right? Because it's not even in the
- 10:50reporting that they would see. So that's and because and
- 10:53because of the spoofing, the SSPS that they have
- 10:55relationships with have no contractual relationship with
- 10:58these additional SSPS that's. Correct.
- 11:01Yeah, they probably wouldn't. But again, because everyone is
- 11:04interoperable with everyone else, because they just want the
- 11:06volume, that's where the problem comes in.
- 11:09So nobody has run the experiments that I run in the
- 11:13past. You actually have to work with
- 11:15the publisher, see what the floor CPM is.
- 11:18Specify A1 domain inclusion list.
- 11:20Right? These are highly specific
- 11:22experiments where we actually got the data and this wasn't
- 11:25ABBC problem. Every single publisher has this
- 11:27issue. And doctor food, no one has one
- 11:29domain inclusion list. So what you you try to keep it
- 11:31as clean as you possibly could and you still, you know, the
- 11:34publisher still lost like 90 plus percent.
- 11:37Yes, yeah. Well, imagine the compounding
- 11:39complexity when you start, you know, doing a real buy.
- 11:42Yeah, exactly right. Nobody does A1 domain inclusion
- 11:45list and nobody cuts down all their SSPS down to one supply
- 11:48source. So basically I've told my
- 11:51advertiser clients the way to simulate a direct buy is to
- 11:55specify one SSP right? Just talk to the publisher.
- 11:58Which SSP do you prefer to sell through to specify 1 and specify
- 12:02one domain or one seller ID? So that way you can simulate a
- 12:07direct buy while using programmatic pipes.
- 12:10So that way other people can't come in and just pretend to be
- 12:13that publishers domain. That's the main problem.
- 12:17And let me tie it to your second point.
- 12:18Rio is where the agentic stuff comes in, right?
- 12:21So for 15 years I've seen bad guys just use WordPress
- 12:25templates and stolen content to create fake websites by the 10s
- 12:29of thousands. Now with agentic they don't even
- 12:32need to use a WordPress template right?
- 12:34The agent will make the website for them and populate all the
- 12:38content for them because it's all just.
- 12:40You can clone a website, just have it have your agents do it
- 12:42in the background now. Yeah, exactly.
- 12:43So, so basically we've seen a whole bunch of these recipe
- 12:48websites and you know, it's all AI generated photos and recipes.
- 12:52It's real, but you might wonder why recipes?
- 12:55It's because on recipe sites you don't have keywords that would
- 12:59trigger brand safety filters like war or shooting or
- 13:02whatever. So it's easy to get by
- 13:05everyone's filters and start making money.
- 13:07So that's why we've seen a whole bunch of these fake recipe sites
- 13:11come into the supply recently due to Agentic.
- 13:15I wonder how much of open web spend is actually recipe sites
- 13:18these days, because there's there's you're right, there's a
- 13:20proliferation of them. And if you and if you do a
- 13:21search for recipe, you'll get like a gazillion come up, right.
- 13:24And I think most of them are you scroll.
- 13:26Most, most of them are completely made by AI.
- 13:28When you read them, you can tell.
- 13:30Yeah. And that was validated by
- 13:32Sigorsky at double verify. They were saying that that's
- 13:34where a major source of the problem is, because it's very
- 13:37little of it's human generated. And it doesn't trigger the brand
- 13:41safety keywords, right, like a new site would or any other site
- 13:44like a medical journal where it has the word blood in it and
- 13:47it's automatically blocked, right?
- 13:48So our recipe sites just get through much easier and that's
- 13:52why bad guys are focusing on that.
- 13:54And more than 40% of I think the legit news content is, is
- 13:58blocked by brand safety due to keyword blocking, domain
- 14:01blocking and all the and all the filters you just mentioned.
- 14:04That's at least 40%. Super rudimentary, back in 2020
- 14:08when the pandemic hit, we have screenshots of the home pages of
- 14:13CNN, Washington Post, New York Times be blocked by the brand
- 14:17safety vendors. And that's not correct, right?
- 14:20The highest traffic volume page on those sites, those publisher
- 14:24sites is the home page. And because it had the word
- 14:27COVID in the URL or on the page, they got blocked and they got
- 14:31demonetized. So all of those and and we, it's
- 14:35documented, right? These brand safety vendors then
- 14:38blamed their own advertisers for not whitelisting those news
- 14:42sites. It's like, well, your technology
- 14:44is supposed to be better than just the keyword, right?
- 14:46They claim Oh yeah, it's. Well, yeah, dynamic keyword
- 14:51block list optimization. That's got to be a dynamic
- 14:53process because they get stale and old fast, based on news
- 14:56cycles, based on political cycles, all that sort of stuff.
- 14:59Yeah, Brett, there was that recent article.
- 15:01Anthony Vargas wrote this article in ad exchange and if
- 15:03either of you've seen it yet, but it was good they talked
- 15:05about how more than half in fact this is coming.
- 15:07We put us in a show notes and he used 2026.
- 15:10The problem hasn't gotten better was his these I guess conclusion
- 15:14that it's more than more than half is being blocked today and
- 15:17these are brand suitable pages, right?
- 15:19That were flagged. And on the flip side.
- 15:21Toward block lists. Yeah, and on the flip side,
- 15:23right on, on this side, the the good publishers are being
- 15:26demonetized. On the flip side, the same brand
- 15:29safety vendors are not blocking ads going to porn sites, piracy
- 15:32sites or even worse. So they're basically allowing
- 15:35the funds to go through to all these inappropriate places even
- 15:39though their claim to their own customers, the advertisers that
- 15:42we're going to help by buying this brand safety services from
- 15:45from us, we're going to help you prevent your ads from going
- 15:47there. And there's been others have
- 15:48documented all of these problems over the years, and not just me.
- 15:53Yeah, and you wonder too, like if a legit BBC page got blocked,
- 15:56right? And then like this the spoof,
- 15:58because you know this, the scammers when they spoof a
- 16:01domain have the the bogus website up there.
- 16:03It's not going to have any of these words, these keywords in
- 16:05it. Yeah, So, so switching to, so we
- 16:08talked about SSPS, we talked about your experiment, which was
- 16:11was super interesting. Did you put did you do a link
- 16:13post LinkedIn post about that one doctor?
- 16:15Yes, that was. Yeah, I'll share.
- 16:16Another one that I read recently was was on on DSP take rates,
- 16:21which I thought was super interesting and I'm sure you're.
- 16:24Pretty much. People don't.
- 16:25People don't talk about this one enough.
- 16:26I'm real. I'd love to get into this.
- 16:27And you argue, let me just say, let me read it back to you to
- 16:30make sure I'm I understood exactly what your argument was.
- 16:32I certainly read it. You argue that blended averages
- 16:35kind of hide the real story and in one case this leading DSP
- 16:38captured 99% of of incremental CPMS as the bids increased.
- 16:43Is that the way to interpret it exactly right.
- 16:45So the bids increase. So there's this amount over the
- 16:47ask ask price, right? That the, the publisher at the
- 16:51end of the, the the supply chain didn't see any of that 99
- 16:55Percent, $0.99 of every dollar for that, that increased bid
- 16:57amount didn't go to the publisher, correct.
- 17:01So tell us a little bit of how that works and.
- 17:03Yeah, let me let me talk about that experiment.
- 17:05So in that particular case, what we did is we upped the CPM,
- 17:10right. So we started with a lower CPM.
- 17:12And the theory was that if you increase the CPM bid, you're
- 17:16you're supposed to in theory get your ads on better publisher
- 17:19sites, right? Because the good publishers
- 17:22can't afford to sell ads at a very low CPM, only the fake
- 17:25site. Premium inventory, right, Yeah.
- 17:27So in that particular case, as we up to the CPM, the take rates
- 17:32also got dynamically adjusted up so that every cent that we were
- 17:37willing to pay extra went to the DSP in the middle, not to the
- 17:41publisher. So you weren't actually
- 17:43advertising on better invent better websites, you ended up
- 17:46just in giving DSPA higher take rate?
- 17:48Exactly. So even if you pay a higher CPM,
- 17:50so a lot of a lot of advertisers say, oh, if we're paying a
- 17:53premium CPM, we got to be getting good publishers.
- 17:56Then you look at the data, it's all the same crappy publishers.
- 17:59The reason for that is in in the middle, the DSP can be
- 18:03dynamically adjusting their take rates on a per impression basis.
- 18:07This gets hidden in the average, right When you look at their
- 18:11public filings, their quarterly investor reports, they'll say
- 18:14our average take rate is this. I'm not even going to say.
- 18:16They'll say some of them like we'll say so it's 20% or it's
- 18:20whatever it is, right. So you're so you're saying
- 18:22that's just average in certain instances is much higher.
- 18:25Yeah, 99%. So you don't ever see that
- 18:28because not all of it's 99%, but for certain campaigns, all of
- 18:33the incremental CPM that you paid went to the DSP and not to
- 18:38the publisher. So you didn't?
- 18:39Get it? Just it seems like it seems like
- 18:41fundamental short termism. You and I were talking about
- 18:44this on LinkedIn, because if, if ADSP that's, that's servicing
- 18:47the open web, right? That it's to their benefit to
- 18:51grow the pie, right? To grow the size of that
- 18:53inventory pool and and the validity of that inventory pool,
- 18:56lest it in the future threaten their business.
- 18:59Yet that seems this this type of thing undermines that which to
- 19:03me has to be just a short termism process.
- 19:06And it is pervasive because every single ad tech company
- 19:11that is a middleman, that is a toll taker has to do this,
- 19:14including Google and including Facebook.
- 19:17And what they do is whenever they need more profit for
- 19:20themselves, they just keep more of it.
- 19:22It's a total black box. You can't go back and track what
- 19:26they kept. If you look at some of the
- 19:27Reddit threads you will see some of the publishers say, oh anyone
- 19:31else noticed that the the was it called?
- 19:35The revenue share rate just dropped dramatically this
- 19:38quarter. So basically moving from Q4 of
- 19:40last year to Q1 of this year. Anyone noticed a 30% drop in
- 19:44revenue share? It's like what the heck's going
- 19:46on? The publishers have no control
- 19:48over that because whoever is the middleman, they can adjust the
- 19:52take rates to whatever they want to maximize their own profits.
- 19:56So that's not ADSP issue, it is every middle man can do that.
- 20:00Who else is doing that? So let's say so DSPS are doing
- 20:03this. Is every DSP doing this or some
- 20:05of them? And then I'm not going to ask.
- 20:07Every, literally everyone can do that.
- 20:10So it's like, do you want to rely on them to be trustworthy
- 20:14and to be honest and above board?
- 20:16You shouldn't because every single one of these ad tech
- 20:19companies have shown that they can go do that and just keep
- 20:22more money for themselves. So what about DSP is with, with
- 20:25owned and operated inventory? Do you think that's different?
- 20:27Do you think because they have the, they're actually, they have
- 20:29the owned and operated inventory, they're, you know,
- 20:31they're actually selling their own inventory through their own,
- 20:33their own platform. Does that make a difference or
- 20:35have you seen the same thing? Yeah, this is different.
- 20:37I mean, owned and operated, you're assuming you're buying
- 20:39direct from that particular publisher or sell direct buys.
- 20:42This is, this pertains to middle men, right, Toll takers.
- 20:45So just think of it in that context.
- 20:47Whoever's a toll taker, you can dynamically adjust whatever they
- 20:50want. And you, you don't have the,
- 20:52the, the data to see how they did it and what they, it's all
- 20:55black box. So they can do whatever the hell
- 20:58they want. And then there's no explanation.
- 21:00So on, the advertiser said. What would you inspect first
- 21:03when you're auditing the supply chain, right?
- 21:07If you're buying direct from the publisher, you at least know,
- 21:10OK, somebody one or more mouths to feed in the middle took 15%
- 21:15and another 15%. So at least the publisher got
- 21:1770% of what you paid, and that should be relatively constant.
- 21:21If there are middle men, then you can't tell.
- 21:24Literally you can't tell. That's where in my case, I
- 21:27bought one domain inclusion list, but still the publisher
- 21:31got $0.10 of what, 10% of what I paid.
- 21:35So if you don't buy direct and you're buying through ADSP with
- 21:41an SSP and there's unknown supply paths, that's where the
- 21:46problem comes in. That's because the bad guys are
- 21:49pretending, you know, they're putting their fake stuff in and
- 21:51pretending to be a legitimate website.
- 21:53And also, I don't know if you know this, the SSPS are bidding
- 21:56on someone else's inventory because all they're doing is
- 21:59arbitraging. So as an SP, like you're not
- 22:02supposed to be bidding on someone else's inventory, right?
- 22:04You're. Supposed to.
- 22:05And that's just an audience extension play where they get
- 22:07the arbitrage and they make the margin difference.
- 22:09And it's just just an issue for like open exchange or is this
- 22:12like like let's say, let's say if a, you know, there's a trend
- 22:16towards more PMP deals or, or or let's say programmatic
- 22:20guarantee, are there specific buying model, buying formats
- 22:23that can minimize this? PMPS will not do that.
- 22:28PGS won't do that. OK, let me describe why.
- 22:31So PMP means private marketplace.
- 22:34It is only as good as the publishers that are in that
- 22:37private marketplace. So if you have APMP, there's a
- 22:40bunch of junkie or crappy websites and mobile apps in it.
- 22:44It doesn't prevent any of the fraud.
- 22:46So the buying model doesn't do anything.
- 22:48You have to look at who's in that PMP the second.
- 22:51If you have legit publishers of the PMP, can they still be
- 22:54subject to yes spoofing? Yes, like we just discussed and
- 22:59just pretend. So let me put it very simply, if
- 23:01you're the owner of fakesight123.com, you will never
- 23:05put your own domain in the bid request because you will never
- 23:07get any bids. You will always have to lie and
- 23:10pretend to be some well known site like CNNESPN, whatever.
- 23:14So in those cases now it's also ties to the MFA sites, right?
- 23:18So Ana came out with their 2023 or 2024 reports saying, oh,
- 23:22there's a whole bunch of MFA sites.
- 23:24And then Jounce Media started selling a list of MFA sites to
- 23:26block. The moment someone bought that
- 23:29list of sites to block and added to their block list is the
- 23:32moment those sites started lying.
- 23:34Yeah, it's defunct. It's meaningless.
- 23:36It's meanless, they can just get around the block list just by
- 23:39lying about the domain. And again they catch us if you
- 23:42can. If you don't bother to check the
- 23:44ads on TXT, it's not our fault that you let it through.
- 23:47But The thing is, after people bought the block list, they saw
- 23:52that in their place reports, all those MFA sites went away.
- 23:55They did not actually go away. They simply don't show up in the
- 23:59data. Because if you're an MFA site
- 24:01and you started saying it's ESPN, all the fake stuff is now
- 24:04commingled in that same row as the real ESPN inventory.
- 24:08So it just gets hidden. So the fact that the MFA sites
- 24:11went away and and they published a study, a follow up study a
- 24:14year after saying, Oh yeah, the MFA site MFA problem is solved
- 24:18because they all went away. OK, It's all still there.
- 24:21It's because they're looking at log level data and place reports
- 24:24that use the domain and the bid request, not the domain where
- 24:27the ad actually went. So all of that was misleading
- 24:30and I had to write back to Ana and said, no, that's misleading
- 24:33it the MFA sites are still in there.
- 24:35I can see it in the data. It's.
- 24:37Just respond. No response because they didn't
- 24:41want to. Just went up to hear about.
- 24:42That so, so your conclusion at the back to Taria's question is
- 24:44that the curated supply PMP's and other and other, it doesn't
- 24:48really reduce waste or fraud. It just kind of rearranges.
- 24:51It a bit correct and the only way to really solve it is to
- 24:54measure it with a post bid JavaScript like mine through
- 24:57analytics where you can actually see where did the ad actually
- 25:00go. So then you compare that to what
- 25:02they said in the bid request and if it doesn't match up you know
- 25:06they lied to you. Simple as that.
- 25:08Yeah, sounds. Like it's a whack A whack A mole
- 25:10problem, though, right? I mean, you have to be like
- 25:12constantly surveilling it, identifying, you know, where the
- 25:15ads come up. And if you if you identify that
- 25:17you know there's a potential problem, investigating it more
- 25:21letting your vendors know. But The thing is, when you have
- 25:24postpaid JavaScript, you can actually see, OK, because these
- 25:27bad guys rotate through a whole bunch of domains and mobile
- 25:29apps. So we can see, oh, it's the same
- 25:31seller ID doing all of this. You just block it at the seller
- 25:34ID because we know that seller is a bad actor.
- 25:37So then you block it at the seller ID level.
- 25:39And then no matter what domain they spoof or what app they
- 25:42spoof. We've.
- 25:43Blocked them. So that's how DSPS are using Foo
- 25:46Analytics right now as a feedback loop, right?
- 25:48We don't do anything in the Prebid because Prebid filters
- 25:51don't work at all. We measure post bid, so where
- 25:54the ad actually served. And then if we can see the same
- 25:58seller is doing all these shady things, right, using bot
- 26:01traffic, lying about the domain, whatever, then we just cut them
- 26:04off. We kick out that seller so we
- 26:06never bid on again. So then no matter how many
- 26:09domains or apps they put in, we've cut them off.
- 26:13Yeah, the so, so, so back to kind of buying type.
- 26:16I mean, there's a lot of talk around, you know, Google's
- 26:17performance backs Meta's advantage plus Amazon's
- 26:21performance plus, you know, they're also black box systems.
- 26:24Fundamentally they're walled gardens, right?
- 26:27Do you think these are attribute, you know, fraud
- 26:30multipliers? Do you think that this is a
- 26:32safer environment for advertisers to be spinning
- 26:35against them? And what's your on?
- 26:36That So there's a couple of things you have to understand.
- 26:39The walled gardens in two parts, the main property, say for
- 26:42example Facebook app and Instagram, that's the main
- 26:46property. And then the audience network.
- 26:48So Facebook has something called FAN Facebook Audience Network,
- 26:52and those are all the outside sites and apps that also run ads
- 26:56for them. And those are the sites and apps
- 26:58that get a share of the revenue. That's all the offsite
- 27:01inventory. Yeah, correct.
- 27:03And that's not on the main sites.
- 27:06So all of them have the motive and the means to commit as much
- 27:09fraud as they want. So if you turn off the audience
- 27:12network, you would have avoided 90% of the fraud.
- 27:16Of course, on Facebook itself, on Instagram, on YouTube,
- 27:20there's still a whole bunch of fake accounts.
- 27:22But if you want to advertise on those platforms, that's the kind
- 27:25of cost of doing business you have to accept.
- 27:27But at least not the 90%. Right, you're since you're safer
- 27:30going direct again in, in in filtering out the audience
- 27:33extension plays outside of their.
- 27:35Network OK do you think and I'm going to explain why in a second
- 27:38why I'm asking this question so if managed advantage plus do you
- 27:41think it's steering buyers towards that let's say the
- 27:44shader inventory and like the bad behavior.
- 27:46The reason why the reason why I asked this is I ran a member.
- 27:48I was running a campaign with a client last summer.
- 27:50This is for on for an auto OEM and we were specifically
- 27:55uploading audiences to meta of people who who purchase certain
- 27:58vehicles because they would qualify for a certain post.
- 28:01You know, let's say post purchase service, right?
- 28:03These are not big, not big audiences 10,000 at a time,
- 28:06right? And we were specifically
- 28:08targeting them with like with personalized creative based on
- 28:12their make and model and what and and and the specific offers
- 28:15that availed of their of their Geo right of it was interesting.
- 28:19Meta without asking us flipped on, flipped on the AI flipped on
- 28:24and we went from a few thousand impressions to 10s of thousands.
- 28:27The CPMS dropped a lot, but so it wasn't, so we weren't really
- 28:30paying much more, but we had no idea who is seeing these ads.
- 28:32We had no idea if they were like there was potential lawsuits.
- 28:36We shut it off immediately. We, we, we couldn't believe what
- 28:38happened. And so curious like, like is
- 28:41that been like kind of your experience and why do you think
- 28:43that's happened? 100% because, and in fact I'll
- 28:47use a simple story, a person was managing his own music accounts.
- 28:52He was selling music and memorabilia and that kind of
- 28:54stuff. So he was managing his own
- 28:56Facebook campaigns for five years.
- 28:58At the beginning of the five year period, 90% of the ads were
- 29:01run on Facebook itself, 10% on audience network.
- 29:05By the end of that, the algorithms had just put 90% of
- 29:09his ads on the audience network, and only 10% was left on
- 29:13Facebook itself. What what they're doing is
- 29:15they're optimizing for larger volume, lower CPU.
- 29:19Larger volume is going to be off site and lower CPM is going to
- 29:21be off site. Correct.
- 29:22There's just unlimited volume off site, right?
- 29:24There's still a number of people who are logged into Facebook all
- 29:27day long, right? So to get the massive volumes
- 29:30you have to go off site. And so that's.
- 29:32What comes in, Yeah, so we're seeing I'm getting like 10 times
- 29:37the number of impressions, 10 times the number of clicks.
- 29:40But my sales went away. So I said go look at your
- 29:44interface. Is the check box for Audience
- 29:46Network checked or unchecked? It is checked by default because
- 29:49Facebook wants. So he unchecked it and then he
- 29:53got a frantic call from Facebook saying the M's are going to go
- 29:57up, your volume's going to go way down.
- 29:58We're not going to be able to deliver in full for you.
- 30:01So you see, so Facebook wants them to use Audience Network
- 30:05because they can just run through all their money as
- 30:07quickly as possible, generate a ton of clicks.
- 30:10And if you're not smart enough to look at outcomes and just
- 30:12look at we, yeah, we got a whole bunch of clicks, you're going to
- 30:15say, wow, that's looking like it's so good, but it's
- 30:17completely useless. All of the clicks are fake.
- 30:21So again, going back to just turn off the audience extension,
- 30:24Audience Network to avoid 90% of the fraud and at least on these
- 30:28wall gardens. So Facebook has the benefit of
- 30:31people logged into Facebook and Instagram all day long, right?
- 30:34So if your ads remain on the main properties, you'll get your
- 30:37ads in front of humans. Same thing with Google, humans,
- 30:40Google things. So if you leave it on Google and
- 30:43then turn off all the outside search, partner network and all
- 30:46that, you're going to get your ads in front of humans.
- 30:49So it's a Facebook example like so like with that, they RIP
- 30:51through your budget quicker, right?
- 30:52So like it just trashes your pacing, right.
- 30:54But I guess the, but The thing is for all of the off site
- 30:57traffic, they do have to pay. They you know, they have their
- 31:00own own, they have their own audiences owned and operated
- 31:03inventory, which is Facebook, right?
- 31:04But I guess you're right, there's only so many people
- 31:06logged in. So it's more questions.
- 31:08But you like, but they do have to pay like they do have to pay
- 31:11out to these other advertisers for the site.
- 31:13They're saying so. They're right.
- 31:16They get half the revenue share right because the other half
- 31:18goes to the site, but it's incremental money for them.
- 31:21So they don't have enough owned and operated on the main
- 31:24property to be able to absorb all the irrational budgets that
- 31:27PNG, Unilever, all these guys. So they have to use the audience
- 31:31to just burn through all of it. Yeah, yeah.
- 31:33And this is that you made that point in the first episode with
- 31:36the, the, the, these budgets don't justify the actual
- 31:39available inventory, right? And humans on Earth, there's not
- 31:43more than 24 hours in a day. And even with parallel
- 31:46consumption, where every human is using their phone while
- 31:49they're watching TV, you still can't account for the trillions
- 31:53upon trillions of ads being sold through programmatic today.
- 31:57Yeah, in the case of that chart that shows by 2013 or so, humans
- 32:02consumption of all these types of media have completely maxed
- 32:06out and plateaued, whereas the digital spend curve keeps
- 32:09shooting up like a hockey stick. So that gap is all made-up with
- 32:14bots and other forms of fraud that you can't explain by human
- 32:17activity. Yeah, well, you, you, you just
- 32:20did a webinar. You shared the the deck with me.
- 32:22That was literally called the agentic AI area and AD fraud.
- 32:26What what changes when fraudsters, intermediaries,
- 32:29buying systems all get access to the scene sort of class of
- 32:32autonomous tools like? The term I used was kind of a
- 32:37democratization of the fraud. So it's like if anyone else can
- 32:41get, that's a negative. Way to talk about
- 32:43democratization. Right, They don't need any
- 32:45skills like programming skills, right?
- 32:47So yesterday I saw on YouTube what's it called?
- 32:50Vibe Hacking, you don't need any hacking skills.
- 32:53You just tell the AI to go hack this competitor site for you and
- 32:56they'll do it for you. So that's really the problem
- 32:59because a lot of these humans, they don't even have the right
- 33:01skills to do this. Now with AI, it's democratized
- 33:06ad fraud to literally any script Kitty who wants to get into it.
- 33:09So again, hopefully the advertisers are a little bit
- 33:12smarter, asking harder questions and don't just let the platforms
- 33:15use their black box and spend your money and then you have to
- 33:19rely on their reports, which is basically them grading their own
- 33:21homework. Of course, what are they going
- 33:23to tell you? Oh, everything's fine, don't
- 33:25worry. Just spend more.
- 33:26But I saw this post by by Tim Rowe, who's, you know, president
- 33:28of the state of streaming. We've had him on the pod before
- 33:30too. And like he was talking about
- 33:32how I mean, for streaming anyway, for CTV, that he was
- 33:34looking at the amount of content that was supposedly viewed.
- 33:37And it was, it would have meant everyone on Earth had viewed
- 33:41several hours of contents like in a day.
- 33:45It was just an impossible number.
- 33:46So like, I mean, I know you've, you've posted some stuff about
- 33:50that too, Doctor Fu, about how like these numbers just don't
- 33:53add up. Do you have any numbers you can
- 33:55share with the viewers about like what you're seeing and how
- 33:57it just doesn't make sense? So I'm going to cite a chart
- 34:00from T Vision, right? So they basically have a little
- 34:03set-top box that measures people sitting there in front of the TV
- 34:06watching streaming. So the curve that they published
- 34:09is that most humans watch streaming after dinner.
- 34:13So between the hours of 6:00 PM and midnight or maybe 1:00 AM,
- 34:17that's when the majority of the streaming occurs.
- 34:20And then they also publish, OK, these are the actual shows that
- 34:23people watching. So those are great data points
- 34:25to have. So if most humans are watching
- 34:29like those 15 actual CTV streaming apps like Peacock or
- 34:35Paramount or Netflix or whatever, Netflix and YouTube
- 34:38being the largest too, what the heck are these hundreds of
- 34:42thousands of other CTV apps and nobody's ever heard of before,
- 34:46right? So then we get into the the
- 34:47ratios. So I've said before, these two
- 34:51statements are true, right? They seem 100% opposite, but
- 34:54they're true at the same time. How can CTV be 100% fraud free?
- 34:59Because fraud is impossible at the same time as CTV is 100%
- 35:05fraud. How can both of those statements
- 35:07be true at the same time? Yeah, I think I know, but an
- 35:13answer for. So basically when you're buying
- 35:16ads from Hulu, from Disney, from Paramount Plus directly, there's
- 35:21no fraud in CTV because it's impossible, because bad guys
- 35:25can't get money out. So they're not in that stream,
- 35:28right? If you're buying from Paramount
- 35:29and you uploaded the creatives to them, they're running it for
- 35:32you. That's it.
- 35:32Yeah, Paramount's probably not trying to RIP you off.
- 35:36Whereas if you're buying anything CTV on big DSP, it
- 35:41could be 100% fraud. And that comes from the fact
- 35:45that, you know, the earliest case of CTV fraud was where
- 35:48Grinder, the mobile app, simply lied and said it was CTV, they
- 35:53just falsified the bid request. So instead of bidding out for
- 35:58display ads, they started lying and saying it's CTV, just change
- 36:01the device type and they got away with it until they were
- 36:04caught. So I don't think Grinder was the
- 36:06only mobile app doing that because you're getting way
- 36:10higher CPMS for CTV than Display or OLV.
- 36:13So every mobile app, every JavaScript in an ad slot, right?
- 36:18Even smart refrigerators are fabricating fake CTV bid
- 36:22request. So that's how you're getting the
- 36:2410s of billions of CTV impressions.
- 36:27Yeah, CTV, that's not CTV, but but charged at CTV prices.
- 36:31Yeah, Here's what Tim. Here's what Tim had, probably.
- 36:33This is great. I found it.
- 36:34So he was. Nielsen tracked with 60.7
- 36:36trillion minutes of television streamed in 2025.
- 36:40That's 62,081 minutes stream per person, who's over 16, which is
- 36:455173 minutes stream per person per month, which is 170 minutes
- 36:50per person per day. There's no way that's even
- 36:52remotely accurate, right? I mean.
- 36:54No, I mean some basic macroeconomic views like that
- 36:57will tell you. I mean, I did the same, which is
- 36:59basic display ads. And it's been, you know, for the
- 37:0215 years since programmatic became mainstream.
- 37:05It's been completely out of whack with reality, but no one
- 37:08wanted to hear that and they preferred to hear that.
- 37:11The legacy fraud verification company said, oh, it's 1% fraud,
- 37:14don't worry about it, just keep spending, right?
- 37:17So that's been the cover up that's been happening for the
- 37:20last 15 years now. I don't necessarily think the
- 37:23Ana or the advertisers or the publishers are being bad guys,
- 37:27right? It is simply that they were not
- 37:29given the right information. So what I tell them recently is
- 37:33that, OK, if you're using the legacy fraud verification
- 37:35companies, they've told you fraud is 1% for the last 10
- 37:38years. Obviously there's nothing to do
- 37:40if that's so low, right? It's within the noise level,
- 37:43right? So now if you upgrade your tools
- 37:46and you have better analytics, if you can see better, you can
- 37:50do better. And so that's where when they
- 37:52upgrade to Foo Analytics, they can say, oh wow, I never saw
- 37:55that there was 50% red in my campaigns.
- 37:57No one ever told me that before. So now I said don't panic.
- 38:01Now that you can see where the fraud is coming from, you can
- 38:04systematically clean it up. So for example, one week you add
- 38:08the top 10 bad guys to your block list, next week you add
- 38:11the next top ten bad guys to your block list.
- 38:14And that way you can progressively clean your
- 38:16campaigns. And in doing so, your CPMS are
- 38:19going to creep up, right, because you're buying better
- 38:21inventory, but the quantities that you buy are going to go
- 38:24down because you don't need to buy that much.
- 38:26So overall, you're still saving money.
- 38:30Doctor for turning back to the DSP take rates again, Excited.
- 38:32We talked about a little bit, I'd love to say.
- 38:34I know they vary quite a bit. I think like Amazon was saying,
- 38:39you know, because they have owned and operated, they're, I
- 38:40think they're saying 1%. I don't know if that's true or
- 38:42not. And then other ones, I mean, I
- 38:44see others have said they're around 20%.
- 38:46Like how can, and like clearly it's opaque and the numbers
- 38:51maybe can't be trusted. How can brands be sure that they
- 38:54know how much DSPS are taking? How can they, how can they pick
- 38:58the right one? And then even looking at like
- 38:59the SSP problems, how can they make sure when they work with
- 39:02publishers that they're limiting the number, only working with
- 39:05legit ones, ones that aren't doing this behavior, What are
- 39:07some best practices you can call out?
- 39:10Yeah. So when you're dealing with the
- 39:12DSPS or any kind of middleman, you just can't know because
- 39:15they're not going to give you the data even if you insisted,
- 39:18even if you sued them for it, whatever.
- 39:20Because there have been cases where the advertisers have asked
- 39:23for log level data and asked for the data behind the verification
- 39:27firms and they'll say, no, your contract doesn't allow you to
- 39:30see that data. Sorry.
- 39:32OK, So if you're dealing with middle men, there's many ways
- 39:35for them to avoid giving you the data.
- 39:37So the best way if you're if you still want to use programmatic
- 39:40pipes is just do what I said before, which is go talk to the
- 39:43publisher and say, we know you sell through multiple SSPS.
- 39:47Just tell us the one you prefer to sell through and you specify
- 39:50that supply. And just limit it to that one.
- 39:53And to the domains, right? So then you're then you don't
- 39:57let any other domains get in there.
- 39:58So you're essentially doing an inclusion list, not just of the
- 40:01domains and apps, but also of the supply source, right?
- 40:05One SSP, not 45, not 98 or whatever, right.
- 40:10And I've done those experiments as well.
- 40:12If you don't limit that, then there's all these supply path
- 40:14complexities which allow fraud to get in.
- 40:17OK, so that's one. The second is actually go back
- 40:20to buying direct. So Disney, Paramount, Hulu, they
- 40:25all have self-serve interfaces where you can literally create
- 40:28an account, login, upload your 32nd ad and then pay the fee.
- 40:35That's the most. Direct through their their
- 40:37interface or their DSP instead of like a let's say a generic.
- 40:40DSP. But does that scale when you're
- 40:42thinking about a multi channel, multi media let?
- 40:47Me ask you? Yes, of course.
- 40:49So let me ask you, how many CTV apps do you even know about?
- 40:55Yeah, it's, it's a small number, right?
- 40:57The legitimate Yeah, if you look.
- 40:58At Tvision Data, it's basically 15 that most humans actually
- 41:02view. So OK, you can set up 15 deals,
- 41:05you can log into 15 sites. I mean some of them are actually
- 41:07together like these Disney and ESPN and who are all the.
- 41:10Same bundle. Yeah, yeah.
- 41:11Do you just log into one place? But that's a lot of swiveling
- 41:14though, like in All in all fairness, there's a lot of
- 41:17swiveling back and forth. That's a lot of different
- 41:19logins. I mean, you know, for let's say,
- 41:22I guess an agency can handle it, but it's still a lot even for an
- 41:24agency because people have to do this.
- 41:25But maybe in the future, once this is like if we could build
- 41:28some like meta DSP agentic interface that can kind of
- 41:31because they all have API's now, in theory you could build 1 on
- 41:33top of all of them. Is that starting to happen now
- 41:35You. Had to go direct buy to all of
- 41:37these automated, you know from. Agencies not willing to to
- 41:41upload to 15 places Then you should fire the dam agency
- 41:44because they're not. They're not.
- 41:45They shouldn't be paid. What else are they doing?
- 41:48What else the heck are they doing?
- 41:49So so Long story short, is just simulated direct.
- 41:52Buy as best you can using programmatic pipes directly.
- 41:55Buy where you can to avoid all the fraud.
- 41:58And that gets back to my statement before.
- 42:00Fraud is impossible in CTV because you're buying from the
- 42:04direct publishers who are not deliberately trying to RIP you
- 42:06off and fraud. If you're buying through large
- 42:10DSP you literally don't know. So you can consider that 100%
- 42:14fraud until proven innocent. Yeah, you're you're forecasting
- 42:17sort of a future that for all this to fix itself the the
- 42:21middle men and you know, you've heard it here on signal and
- 42:23noise. The middle men are the problem,
- 42:26right? And it doesn't seem like the
- 42:28incentive structures the. Props the incentives.
- 42:31Are aligned to. Incentives are the problem.
- 42:33To change, it's the incentives. I mean, they're, they clearly
- 42:36know they can get gravy on top of incremental ads that take
- 42:39great perspective. Yeah, they, you know, without
- 42:41anybody knowing and it and it just boosts their numbers for
- 42:45their publicly. Reported I mean, it makes you
- 42:47think if we could rewind a clock 2530 years, right and say OK,
- 42:50like, you know the whole incentive problem, the fact that
- 42:53media's compensated based on tolls or take rates.
- 42:56I mean, I think that that that incense a lot of not to say
- 42:58there wouldn't be problems if we had a different cost model.
- 43:00But I think when you look at a lot of things we're dealing with
- 43:02today, it's because it's there's a perverse incentive to spend
- 43:06more and more and more because you're because and because
- 43:09you're taking maybe not, maybe not fixed them out, but you're
- 43:12always taking a cut of a cut of the, you know, all these middle
- 43:14men who jump in and doing that. Yeah, you're taking a percentage
- 43:17of and most of those pricing models are CPM plus, which means
- 43:20they add something on top of the media CPM and that's where it
- 43:24went wrong, right? So let me now layer in one more
- 43:27thing, which is the agencies, OK?
- 43:29So if you thought that the dollar 50 going to $0.15 was
- 43:32bad, what actually happens is the agency is charging the
- 43:35advertiser their own client $15. OK, that's the CPM that
- 43:41advertisers pay for display ads then the same agency.
- 43:45A 10X on the on the base CPM. It's $15 CPMS for display ads.
- 43:51The agency turns around and bids $1.50 so that they pocket the
- 43:55other 90% and then that dollar fifty, only 15 cents goes to the
- 43:59publisher. So the reality is actually the
- 44:01advertisers paying $15 and the publishers getting $0.15, which
- 44:06is 1 cent on the dollar. And again.
- 44:09The agency is keeping this on top of their agency fee.
- 44:12Let's say it's 15%, right? So they're taking if they're
- 44:14charging 15% to actually plan by the media, but then they're
- 44:17they're taking. No, it's actually hidden now.
- 44:19So a lot of the agencies I've seen recently say, oh, we won't
- 44:23even charge you any percentage because they want to win the
- 44:26business. Where do you think they're
- 44:27getting that money? They're getting it from hidden,
- 44:30undisclosed margin and. For example, it's funny, I
- 44:33actually saw an RFP this several years ago that actually said we
- 44:37want you to do it for 0 for for for nothing because we expect
- 44:41you'll find the money, I guess. But that's like almost saying we
- 44:43know that you'll take rebates, We'll know that you you'll take
- 44:46you'll mark up things that undisclosed amounts.
- 44:48I mean, that's, I mean, it's almost whose fault is it then?
- 44:50I use it to. Oh I'm blame the advertisers and
- 44:53specifically their procurement departments, right?
- 44:55You have forced the agencies to buy fraud and do shady things
- 44:59because you're not compensating them properly, right.
- 45:02So advertisers have gotten into the habit of we want 10 times
- 45:06more quantity at 110th the CPM price.
- 45:10Yeah, it's all. As much reach for his little
- 45:12dollar. Yeah, and just because they were
- 45:13picking the agency based on CPMS like big pubs.
- 45:16That's it, right? Yeah, it's, it's weird.
- 45:18It's almost like diametrically opposed to how television
- 45:21advertising work, television advertising work.
- 45:23They could, they could demand a premium price because they've
- 45:25got a bring, you know, a premium captive on.
- 45:27Well, you would you would up front where you'd sell 90% of
- 45:30your inventory. Then you know the scatter market
- 45:32was for that remaining 10%, the scatter market was usually more
- 45:34expensive too, which is interesting, right.
- 45:36So I think digital we've it's on its head like most of it is
- 45:39program at 65% is programmatic. Yeah, even more than that, I
- 45:43think it's 80% now. But in TV there was actually
- 45:46finite supply. So it made sense for your hold
- 45:50code to go say we have all these big clients, we're going to
- 45:52bundle all their spend together and then go negotiate against
- 45:55NBC or ABC or whatever. That worked.
- 45:59But in digital it's a complete opposite because there is no
- 46:02limit on the supply. We have no, but there is, there
- 46:06is limit. Yeah, there's limit.
- 46:09No, but minus sans the the bad guys, there technically is a
- 46:13limit. On good supply.
- 46:15For good supply, there is a limit, and that's what's broken
- 46:18right now. So for example, the reason a lot
- 46:21of these advertisers got convinced to to stop buying from
- 46:24New York Times and buy from exchanges is New York Times has
- 46:27so little inventory because it's based on real human visitors,
- 46:30OK. So it's finite and it hasn't
- 46:32grown. In fact, it's declined
- 46:34percentages at year after year. So all of the growth came from
- 46:38fake sites and fake apps. So that's why the advertised,
- 46:42that's what I mean by unlimited inventory.
- 46:44So if you look at the last 15 years and let me talk about the
- 46:47economic macroeconomic level, there have been billions and
- 46:51billions of dollars shifting into digital from traditional
- 46:53right offline media, whatever. So if.
- 46:55There's 70% of all AD dollars spend, right?
- 46:58Now then, the prices should have gone up, right?
- 47:01Based on basic economic theory, right?
- 47:03If the demand goes up and supplies finite, prices should
- 47:06go up. But in fact, we've seen 15 years
- 47:09of price declines because the supply grew even faster than the
- 47:14tremendous amount of dollars shifting into digital.
- 47:16OK, so all of that gap is made by fraudulent fake inventory.
- 47:21Now on top of that, I published an Article 2 months ago, when we
- 47:25look back at the at the last 15 years of revenue of 255 public
- 47:31companies in the US, the largest ones like PNG and Unilever that
- 47:36that shifted the most aggressively into digital
- 47:39actually grew 0 or negative OK revenue growth.
- 47:43Interesting. So.
- 47:44And just let me just interrupt you for a second just to give
- 47:46people a size like an idea of the scale of this that 2026 is
- 47:51about 830 to 860 billion globally is digital, digital
- 47:56specific of a total 1 to $1.25 trillion ad market, right, 70%
- 48:01of all those ad dollars, so 70% of all ad dollars, right, 4. 100
- 48:05and increasing every year, like every year like traditional.
- 48:08So this is a large piece of the overall pie of total ad spend.
- 48:14Yeah, it's really interesting. And then if we look at just the
- 48:17US companies, because they all have public filings, right?
- 48:19We can see it in their public filings.
- 48:21And then some of them, you can look up the ad age numbers,
- 48:24which is their split between traditional and digital.
- 48:27The ones that are majority digital, where could be 90%
- 48:30spend in digital and 10% in offline.
- 48:32Those are the companies that have grown the least
- 48:35categorically. But then you have other
- 48:38companies like Progressive and GEICO, right?
- 48:40You all seen their TV ads. They've maintained A sizable
- 48:43share of their spend in TV. And those are the ones that
- 48:47grew, in some cases double digits compounded annual growth
- 48:50over the last 15 years. So the linear advertisers, the
- 48:54television advertisers listening to the pot are just like
- 48:56celebrating. Right now they're just because
- 48:58they want, right? They want because all the guys
- 49:00who thought they were innovative and say, oh, we're aggressively
- 49:03shifting everything into digital, you know, because they
- 49:05thought, oh, it's precision targeting.
- 49:07It's one ad to the right person at the right time, right?
- 49:11They didn't account for the fraud, so between the inaccuracy
- 49:15of the targeting and the 90% fraud, all the money spent in
- 49:20digital didn't do anything for their actual revenue growth.
- 49:23Put CTV save the open web, right?
- 49:25I mean, it's no, it's been the biggest growth.
- 49:29CTV is not you don't considerate.
- 49:32No, I mean, there's Peacock, there's Disney, there's Hulu.
- 49:36Hoku is really dangerous, so I'll leave that to a second
- 49:40different pod. But there's only a finite number
- 49:43of streaming channels or apps that humans use.
- 49:46Those are the only ones you need to buy from, right?
- 49:48So Netflix now has ads, buy on that.
- 49:51Amazon Prime has ads, buy on that.
- 49:54YouTube, humans are logged into YouTube, buy on that.
- 49:57That's how you get your ads so. Because of its limited, because
- 50:00of its limited inventory and in fact you can go direct to these
- 50:04streaming services publishers, right and you can buy.
- 50:06So you're not including them because many people would say
- 50:08CTV is part of open website. You don't think it is?
- 50:11I don't think it is because on the open web, when someone
- 50:14visits a recipe site, they're not logged in, right?
- 50:17They just want that one recipe and they leave that to me is
- 50:20open web where the users are anonymous, they're not logged
- 50:23in. And that's what gave rise to
- 50:25surveillance marketing, which is where they're tracking users by
- 50:28cookies. And that's when the targeting
- 50:30went to crap. Because in the beginning, when
- 50:33you have a very finite number of sites, you could probably
- 50:36guesstimate the person, OK, if they visited Victoria's Secret,
- 50:40Maxim, Playboy, you know, those have that.
- 50:42You can probably guess. It's some dude, right?
- 50:44It's some guy. So you can guess.
- 50:47Visit playboy.com What? About retail, what about retail
- 50:49media? I mean, because they kind of
- 50:51like don't or I mean, they have some of the elements of, you
- 50:55know, of you're logged in generally because you're
- 50:57shopping, right? All retail media sit a St. or
- 51:00Sinner and they're mostly sinners because from the get go
- 51:04they blended in audience extension.
- 51:07Yes, yes, exactly. It was, it was sort of like.
- 51:10Yeah, they're acting like. Retailers when you're in the
- 51:12media. I want my ads on the screens in
- 51:14Kroger. Don't give me yeah.
- 51:16It is exactly you, it's, it's a, it is a retail experience.
- 51:20You're in Walmart, but you're in a digital environment, right?
- 51:23And so you want ads within that ex that shopping experience,
- 51:26just like you would see an end cap in a store.
- 51:28And yet they're going out to the open web and they're pushing the
- 51:30stuff for audience extension. And that's when everything goes
- 51:33to crap. And it.
- 51:34Went do you know what percent is audience extension versus let's
- 51:37say on site? 90% of them. 90% of retail media
- 51:41is. Because they want the volumes,
- 51:44the in order for them to match the CPMS and the volumes, they
- 51:47had to do that and so no. Is it the same across the board?
- 51:51I mean, is it the same across the Amazons and the Walmarts?
- 51:53I don't, I don't know specifically, but we have seen,
- 51:56I've measured some of these retail media networks and we had
- 51:59to go to them specifically because I work with a big
- 52:01agency. I said cut that out.
- 52:03That's not where we're buying. Yeah, because.
- 52:05You think you think that 80% of the spend is in Walmart?
- 52:08I think it's 80% of total spend is in.
- 52:10There's Walmart and Amazon combined, and then there's
- 52:12there's 200 other. There's a new one every day,
- 52:16yeah. That the smaller players just
- 52:18are not inherently going to have the inventory, so they're going
- 52:21to almost be forced to get from an account and it.
- 52:23Creates the incentive for them to do audience extension to get
- 52:26more inventory. Same thing happened with local
- 52:28media sites. So for example, a television
- 52:31station has a website, there's finite number of people that go
- 52:34there, right? You're if you don't live in that
- 52:36town, you're not going to the television company's website.
- 52:39So they were from the get go, they were attempted by all these
- 52:42adtech companies. Oh, hey, look, we can give you
- 52:4410 times more inventory if you do audience extension.
- 52:47So that's when all of that went to crap.
- 52:50And we actually found some of these monetization partners
- 52:53actually juicing all the numbers.
- 52:54So the monetization partner was the one buying all the bot
- 52:57traffic and laundering it through the local TV station
- 53:01website. It's like, how many people live
- 53:03in this town? How can there be 10 million
- 53:06people visiting this local TV station?
- 53:09Website, yeah. OK, that's not even possible.
- 53:12So. All of these came from the
- 53:13audience. What do you call it?
- 53:14The monetization partner was cheating and then using their
- 53:16good name, did laundry. Yeah.
- 53:18So your, your hypothesis that that when advertisers shifted to
- 53:22digital, right, let's say digital and aggregate, they
- 53:26often times saw no improvement gains in performance versus what
- 53:30they originally did in linear. I wonder and tell me if there's
- 53:33an experiment in place already, if, if advertisers said we're
- 53:37going to focus on direct buys, direct with publisher buys,
- 53:40premium and inventory, if you would see a comparable type of
- 53:44performance outcome as you did? So I'm going to recommend we
- 53:49don't use the word performance because in a lot of advertisers
- 53:52minds, performance meant number of clicks they got.
- 53:54OK, that was raw. OK, we're going to talk about
- 53:57sales and not even outcomes because outcomes are easily fake
- 53:59now due to attribution, reporting, all that kind of
- 54:01stuff. So I can cite a case study.
- 54:04My client, Georgia Pacific, they sold paper towels, toilet paper,
- 54:08paper plates and other paper products on Amazon product
- 54:11pages. OK.
- 54:12When they did CTV, when they did OLV and actually targeted to
- 54:17humans, they saw a lift in sales of toilet paper, paper towels.
- 54:24How'd they calculate that? Did they actually get data from
- 54:26the retailers? And they did a, you know, they
- 54:27did a report to. Look at on Amazon.com.
- 54:30This is on Amazon.com. OK, so for big CP, so for big
- 54:34CPG, it must be challenging. All right, so, so they're you
- 54:36know, they're buying, they're buying, they're buying media
- 54:38different places. And you know, let's say they
- 54:40don't have brick and mortar, right?
- 54:42They sell through channels. I mean, it must, I've gone
- 54:44through this exercise before where you're trying to get
- 54:45you're trying to get data from the retailers.
- 54:47It's really challenging. They don't like sharing it.
- 54:49It's usually trailing is who's incomplete.
- 54:51You have to model it out. I mean, I think, I think this
- 54:53makes it harder to go through the exercise you just described
- 54:57because it's it's it's hard, difficult to understand cause
- 55:00and effect like you should showing people ads versus are,
- 55:02is it actually driving behavior, right?
- 55:05Yeah, 100%. I've dealt with this for 15
- 55:06years as well. So grocery stores, right?
- 55:08If you're just selling soup and soda, it's sold in grocery
- 55:11stores and the manufacturer doesn't even have contact with
- 55:14the grocery store because they sell the regional distributor.
- 55:17So again, they don't get that data back soon enough to know
- 55:20how to optimize their campaigns. In this particular case, they
- 55:23had Amazon product pages. They could see the lift in sales
- 55:26once they started optimizing away from crap and optimized
- 55:30towards humans. The other thing they can do is
- 55:32regional turn off experiments. So this is what eBay did in
- 55:362011, 2012. Yeah, they turned off the entire
- 55:38W. That's how you build your
- 55:39control. And then all this, all this take
- 55:42came came in at still right. None of it diminished.
- 55:45It's just, it's a classic holdout test, right?
- 55:47Like it's like what's old is new, right?
- 55:49These are things we've been doing in marketing forever,
- 55:50right? So.
- 55:51Yeah. Yeah.
- 55:53So, so, so, so like so now that we're on the topic of outcomes
- 55:57or, or attention or whatever we would have called performance,
- 56:01right. You've drawn a line, I think it
- 56:02was in your webinar that between attention and and offenders love
- 56:06to sell, which which seems to like a reach metric to, to what
- 56:10you call it attentiveness. Yeah.
- 56:12Let me explain. Which which can be observed
- 56:14through kind of stronger behavioral signal, Yeah, how,
- 56:18how do advertisers optimized towards that?
- 56:21Yeah. So what does that mean?
- 56:22Very easily, yeah. So let me start with kind of the
- 56:25evolution. So we went from viewability to
- 56:27then attention to, I'm going to recommend the next step, which
- 56:30is attentiveness. So viewability has to do with
- 56:33the ad itself and whether 50% of the pixels were in view.
- 56:36So it has to do with whether the ad was viewable.
- 56:39The attention is a nice concept because these attention vendors
- 56:43have done eye tracking studies in a lab.
- 56:46So, for example, a larger ad would get more attention than a
- 56:49smaller ad. Totally makes sense.
- 56:50An ad that covers your entire screen would get more attention
- 56:53than a tiny banner ad. Totally makes sense.
- 56:55But the problem is these attention vendors cannot
- 56:58actually measure attention when they're running real ad
- 57:02campaigns. Because when you have an ad
- 57:05running and you have a Pixel tracker in there, you can't
- 57:08actually turn on the camera of your phone or your laptop to see
- 57:11if the person was actually looking at the screen, because
- 57:14that's going to throw a user prompt and you can't do that.
- 57:17OK. Yeah.
- 57:17So they can't actually that's also.
- 57:19Super creepy. Yeah, Yeah.
- 57:21Like an ad should not be turning on your camera.
- 57:24Yeah. So they can't measure that.
- 57:26And so essentially, when you're optimizing for attention,
- 57:29meaning higher attention, you're just optimizing for larger ads
- 57:32compared to smaller ones. You're optimizing for ads above
- 57:35the fold compared to ones that start off below the fold.
- 57:37That's not. Unrelated or related.
- 57:39But. Question Doctor Fu, are these
- 57:43are some of these devices actually, you talk about
- 57:45surveillance capitalism, right? Are they, are they listening?
- 57:49Are they filming? Are they, I mean, have you seen
- 57:52evidence to that or is that all conspiracy theories?
- 57:54What's your? Opinion.
- 57:55I think a lot of people give the permission without knowing they
- 57:58gave the permission right? So Google has this thing that
- 58:01listens for songs and helps you identify what song is playing.
- 58:04So the speaker, the microphone's on all day long.
- 58:07OK, so you don't know that because that's how it works in
- 58:10order for you, but that's kind of a side topic.
- 58:12Let me just finish the topic of attentiveness.
- 58:15So what we can measure is actually on the landing page of
- 58:19the advertiser. So when the person clicks
- 58:21through from the ad and gets to the landing page, basic common
- 58:25sense tells you if a human deliberately clicked on an ad,
- 58:28like a search ad, they're looking for something.
- 58:30So they're probably going to do something like move their mouse,
- 58:33click something, at least touch the screen, scroll up and down.
- 58:36South. Attentiveness is the percentage
- 58:38of the users who arrived on the landing page that actually did
- 58:41something. And I can give you a very simple
- 58:43contrasting example from Google search.
- 58:46The people who click through from Google search that arrived
- 58:49on the landing page and we see UTM source equals Google 70 to
- 58:5390% of those users actually did something else like move the
- 58:56mouse or click something. That's great.
- 58:58Users coming from programmatic or users coming from TikTok,
- 59:02only 1 to 2% of them did anything on the landing page.
- 59:06OK, So if you're an advertiser, which is the more valuable
- 59:09visitor for you, right? The ones that actually did
- 59:11something. So now to tie this back to
- 59:13attend attention and viewability, attentiveness
- 59:17already takes into account both of those because the ad would
- 59:21have had to have been viewable. The human would have had to paid
- 59:25attention to it to then get influenced by the creative
- 59:28message and then want to click through and then come to your
- 59:30site. So if the person came to your
- 59:33site, first of all, it's human and they did something else.
- 59:36We've already accounted for viewability and attention and
- 59:40this is directly measurable on the advertiser site and it gives
- 59:43them a direct actionable optimization, which is we're
- 59:47going to spend more money on the page.
- 59:49On Google, Google search wins again.
- 59:52Well, that particular example, yeah, yeah, that particular
- 59:55example and the. Greatest business ever built?
- 59:57Google search, Yeah. Really.
- 59:58Yeah. So bottom line is they can
- 1:00:00optimize towards the pay channels that are sending more
- 1:00:02attentiveness, right? Users would.
- 1:00:04Require, yeah, yeah. And that and that same logic can
- 1:00:06apply back to like an A chat bot once they start introducing
- 1:00:10advertising. Yeah, well, they have already.
- 1:00:13It's the same kind of path that you're following through to see
- 1:00:15what they do when they actually arrive and how does that compare
- 1:00:20versus other types of traffic which you just outlook?
- 1:00:23Interesting too, because you think about like what you just
- 1:00:25described could obviously be mimicked by bots pretty easily,
- 1:00:27but obviously with some of the filters and that you're talking
- 1:00:30about, you could, you could filter out or not not count bot
- 1:00:32traffic, right? But then, but as people start
- 1:00:34having at some point, if you think that we're going to have
- 1:00:36our own agents going doing things, we've seen a lot like
- 1:00:38even a signal of noise today. I we've seen a lot more bot, you
- 1:00:41know bot traffic going clicking around other things is
- 1:00:43interesting so well you. Can add analytics on your site
- 1:00:47because we can actually tell that apart so there have been
- 1:00:50campaigns from P Max where the bots actually are faking the
- 1:00:54clicks on the landing page but not only could label them as
- 1:00:57bots beforehand but we can look at the click patterns and say OK
- 1:01:00that's not a natural human click pattern and simple we're just.
- 1:01:03Saying click pattern again and again and again, right?
- 1:01:05Yeah, yeah. So, so there are things now the
- 1:01:07bots will eventually optimize that.
- 1:01:10And I mean, we've seen bots actually copy off the human
- 1:01:12mouse movement patterns and just replay them.
- 1:01:14That's a simple way to get around detection, but they would
- 1:01:18have to fake a dozen different things correctly for me to not
- 1:01:21catch them. Yep, and can you catch them?
- 1:01:23You've also mentioned that sometimes fraudulent reporting
- 1:01:26passes over through to like Google Analytics, like Georgia
- 1:01:294I mean have. You seen that the problem with
- 1:01:31Google Analytics is that they are required to filter out the
- 1:01:35bots. And you might think on the
- 1:01:37surface that might be a good idea, but actually it's not,
- 1:01:39because you actually want to see that there are bot clicks coming
- 1:01:43from different paid campaigns so that you know you have a
- 1:01:46problem. If Georgia just doesn't show you
- 1:01:49the bot clicks, you won't know that you have a problem with
- 1:01:52those campaigns. So that's where clients are
- 1:01:54adding food analytics on site tags next to GA, not in place
- 1:01:59of. So when there are discrepancies
- 1:02:00either too high or too low, they can use food analytics as a
- 1:02:03source of truth to help troubleshoot those.
- 1:02:06And they're comparing the 2 they can see like what GA 4 is
- 1:02:08reporting and say, OK, what food analytics is reporting.
- 1:02:10So, OK, there's a spike here and I'll food analytics is
- 1:02:13reporting, you know, increase in bot traffic correlates with
- 1:02:16this, OK. Exactly.
- 1:02:17Makes sense. Nice product.
- 1:02:19Placement there doctor food, nice product placement.
- 1:02:21I like that. That the bots are getting way
- 1:02:24more sophisticated though, right?
- 1:02:25I mean like, as I say, I improves.
- 1:02:27I mean how how do you how do you keep up with it?
- 1:02:29I mean it's, it's, I mean every day there's their new.
- 1:02:31Events, it's a little I'll give you a little secret here, and
- 1:02:35I've said this before, so it's not I don't worry about the
- 1:02:37fraudsters listening to this. So it's usually easy for a bot
- 1:02:41maker to tell the bots to click on a bunch of stuff and you
- 1:02:45would think that that's basically a random pattern.
- 1:02:47So if we see that the clicks are spread too evenly across the
- 1:02:51page, that's not humans. OK, on the on the flip side,
- 1:02:54it's easy for the bot maker to say click on this button or this
- 1:02:57XY pixel repeatedly. So again, that's not human.
- 1:03:00So if it's too ordered or if it's too random, it's not real,
- 1:03:05right? Humans are somewhere in between.
- 1:03:07And so I've said this before, we can still measure it.
- 1:03:10Again, they would have to fake a dozen different things correctly
- 1:03:13in order to get by my detection. Yeah.
- 1:03:16So, so looking at patterns, I mean, I bet that the bot traffic
- 1:03:19ends up being very similar over time when you're looking at it
- 1:03:22in the aggregate, right? Because I mean, that's what
- 1:03:23these, these, these, I mean, when you look at what these LLMS
- 1:03:26do, I mean, they're. Averaging kind of everything,
- 1:03:28yeah, basically. So, so, yeah, so let's, let's
- 1:03:32get to the, the, the sort of, So what, what can marketers that
- 1:03:35are listening to the podcast do this quarter and like the
- 1:03:39immediate future? And you know, if you were
- 1:03:40briefing in CMO, you do this all the time, right?
- 1:03:43A procurement leader, a media lead, whatever it might be.
- 1:03:45What 3 immediate changes would you mandate this quarter or
- 1:03:48let's say the first half to reduce waste and improve signal?
- 1:03:521st is ask harder questions. Don't assume that any vendor
- 1:03:55that you're buying from, right, whether it's the verification
- 1:03:58vendors or the DSPS or whatever, is telling you the truth.
- 1:04:02Not because they're malicious. It's simply that they have no
- 1:04:04incentive to tell you the truth, right?
- 1:04:06They're going to tell you what you want to hear.
- 1:04:07So you have to have independent measurements so that it gets to
- 1:04:10point #2 which is like have the vendor show you the data.
- 1:04:16And I'm asking that specifically because if the verification
- 1:04:19vendor and you ask them, can you explain why you mark this as a
- 1:04:22bot or why you didn't mark this as a bot?
- 1:04:24They say, oh, we can't explain it to you because it's our
- 1:04:26secret sauce and we don't want bad guys to know they don't know
- 1:04:29what they're talking about. South, stop buying from those
- 1:04:31vendors. You need a vendor who can
- 1:04:33actually explain everything to you, give you the supporting
- 1:04:36data so you can understand why something is a human versus why
- 1:04:40something is a bot. OK, so once you have that, then
- 1:04:43you can have confidence in the data that you're seeing.
- 1:04:46Then you can say, oh, well, obviously these are bad sites
- 1:04:49and bad apps eating up my impressions.
- 1:04:51And then the third step, very simple, start blocking these bad
- 1:04:54guys, right? I wrote an article called The
- 1:04:56Maturity level of advertisers using Food Analytics. 1st is
- 1:05:00lower the red right, start optimizing away from the fraud.
- 1:05:04Second is start optimizing towards the humans, right?
- 1:05:07Increase the blue because you want more ads to be shown to
- 1:05:10humans. And then the third is optimize
- 1:05:12for attentiveness on the landing page and outcomes if you can,
- 1:05:16right? So if you're an e-commerce site,
- 1:05:18you can see the outcomes of sales right there.
- 1:05:20If you're CPG or soup and soda, it takes a little too long
- 1:05:24because you're selling in grocery stores, but still, you
- 1:05:26can use attentiveness as a gauge for yeah, that person was
- 1:05:29interested, they would actually do something on your side.
- 1:05:31Yeah, you might be getting lag lagging indicators, but it's
- 1:05:34still going to give you better insight than if you're not doing
- 1:05:37that as a CPG player to plan ahead for your for your next
- 1:05:40buying cycle. Doctor, for what percent of
- 1:05:43overall ad spend is is fraudulent?
- 1:05:45I mean, just just thumb in the air.
- 1:05:48No, I'm not going to say that because that would essentially
- 1:05:51be an average. OK.
- 1:05:52So I will give you the range. The range could be 1% fraud or
- 1:05:56100% fraud and I've seen all of it.
- 1:05:59So again, because I don't have 100% measurement for the entire
- 1:06:03digital ecosystem, it's irresponsible for me to
- 1:06:07extrapolate and tell you a number, whereas everyone else
- 1:06:10does that. OK, Those numbers are not true
- 1:06:11because if they're just extrapolating from what they can
- 1:06:14see. I shoot to piggyback on what you
- 1:06:16said before, I averages may be part of the problem, right?
- 1:06:19Because if you think about it like you can hide a lot of bad
- 1:06:22behavior. If you come up with an average,
- 1:06:23if let's say the average, let's say the overall number is 2025%.
- 1:06:26Oh, that's bad, but not that bad, right?
- 1:06:28But but in certain campaigns, to your point, it could be 9900%.
- 1:06:31Yes, I know. And I like the question, Doctor
- 1:06:33food because I mean, I think Rio was leading you to say 99%,
- 1:06:36right. It just the most controversial
- 1:06:38statement that we'll put in the LinkedIn post.
- 1:06:40Yeah, Doctor Fu. Claims we wouldn't do that.
- 1:06:43Come on. You can actually do that.
- 1:06:45You can actually do that and we'll get a lot more views,
- 1:06:47right Doctor? We probably will.
- 1:06:49Be 9%. So all the haters will come out
- 1:06:51of the woodwork and start hating on us.
- 1:06:53But but Rio, this actually goes back to what I why I got into
- 1:06:57this 13 years ago when I was at Omnicom, we were actually seeing
- 1:07:01average click through rates on search campaigns to be 9 to 15%.
- 1:07:05That's the average. But then when you got police
- 1:07:07reports that broke out each individual site, some of those
- 1:07:10sites, the fraudulent ones were 100% click through rates.
- 1:07:13Again, the problem was hidden in the average.
- 1:07:16So you said yay, my search campaign was working great, it's
- 1:07:1910% click through rates. But then when you broke it out,
- 1:07:22you could see the problem. So just to close this out,
- 1:07:24you're right, averages have not done anyone favors because it's
- 1:07:29actually hidden the problem from view.
- 1:07:31Yeah, and you can get that in your most basic platforms.
- 1:07:33I mean, we've run e-mail campaigns, for example.
- 1:07:35And I've seen what I've been talking about this a lot is, you
- 1:07:38know, you see a pretty healthy click through, but then you dig
- 1:07:41in. I noticed that it was like 2
- 1:07:42domains in my entire e-mail list that were driving 90% of the
- 1:07:47clicks. And I'm like.
- 1:07:48Yeah, we thought, we thought we had really good results.
- 1:07:50Yeah, it was like 500 clicks from one domain, from 1, from 1
- 1:07:54e-mail address identifier. And I'm like, that has got to be
- 1:07:56a bot. That's that's actually this is
- 1:07:58actually an anti fraud protection bought for the for
- 1:08:01the company, but that's that was the activity that we sort of
- 1:08:05pulled out. Yeah.
- 1:08:06So should we go to quick hits now that we're at the hour or
- 1:08:10just just above the hour? Really great.
- 1:08:13Didn't we already do that, which is like the the three
- 1:08:15recommendations? I think that's, I mean, unless
- 1:08:18there's something else you want to talk about.
- 1:08:20Well, we have quick hits where we give you, we give you kind of
- 1:08:22quick short answers to to very simple questions.
- 1:08:29So I'll give an example. So one metric you'd ban from
- 1:08:32marketing decks tomorrow? Click through rates or view
- 1:08:36through conversions. What's one place in the stack
- 1:08:39where a lot of the bad behavior happens that you could just nuke
- 1:08:41if you had were able to? It's going to be, if we're
- 1:08:46talking about programmatic media buying, it's going to be a few
- 1:08:48SPS that let the majority of fraud onto the ecosystem.
- 1:08:52And if you cut some of those off, it's going to be a lot
- 1:08:54better. And the final one, because I
- 1:08:57know you got to run, we're at the hour and we all have
- 1:08:58meetings, is one question procurement should ask every DSP
- 1:09:03from a contractual perspective immediately.
- 1:09:05What's percentage of the dollar that I'm spending actually goes
- 1:09:09to the publisher and I can guarantee you 100% of the DSPS
- 1:09:13will not tell you Wow. Well, that's.
- 1:09:16Great. Well, Doctor Fu, awesome having
- 1:09:18you again. Thanks for spending another hour
- 1:09:21with us. It's always super insightful.
- 1:09:23Love the research and, and, and just the, the things that you
- 1:09:26support and, and I think it's for the betterment of the
- 1:09:28industry, It's for the betterment of the open.
- 1:09:30Web Yeah, your previous episode continued.
- 1:09:32That's one of the ones people continue to come back to, and we
- 1:09:35see it continue, people watching it.
- 1:09:36So we hope this will have more people go back to that one,
- 1:09:38watch this as well as this one too.
- 1:09:40It's a great discussion. This is lived up to the
- 1:09:42building. Thank you.
- 1:09:43Thank you. And for everybody listening
- 1:09:45that's made it this far, follow us on YouTube, Spotify, Apple
- 1:09:51Podcast. You can find all of our content,
- 1:09:53editorial podcast and otherwise on Deb Deb, Deb dot signal and
- 1:09:57Noise dot AI. And we will see you next time.
- 1:10:00Thanks everybody. Thank you.