Turn the Tide on Programmatic Fraud with Analytics
Turn the Tide on Programmatic Ad Fraud with Analytics
Dr Augustine Fou, Ad Fraud expert, joins Matt once again to talk about the increasing awareness of ad fraud. Nearly every day there are reports and articles detailing how ad fraud funnels millions of dollars away from publishers and supports everything from hate speech to international espionage.
A recent infographic presented data that claimed that there were 200 billion programmatic ads served every day. While it may feel like that to some of us without ad blockers, Dr Fou was quick to point out there there are only 7 billion people on this planet, and most do not have access to the internet. Similarly, a few years back, Facebook claimed it had more MAU’s in the US than the actual population of the country! The programmatic advertising industry loves to make big, audacious claims like this, but the numbers don’t add up.
Dr. Fou provides actionable steps that you can take right now – using your own analytics – to spot fraud and wasteful spending. Sit back as he provides a cold dose of reality that counters the wishful rainbows and unicorns world of programmatic advertising.
[00:00:00] Dr. Augustine Fou: You have to trust that their numbers are accurate and that they’re not cheating. Now, some of the bigger companies, obviously they’re not actively cheating, but you can imagine the fraudsters are deliberately cheating, right? So…
[00:00:17] Matt Bailey: Yeah.
[00:00:17] Dr. Augustine Fou: There are so many techniques, right? One of the vul-, key vulnerabilities of Google Analytics is that anyone with your UTM code can actually, uh, write bad data into it.
[00:00:32] Bumper Intro-Outro: Welcome to Endless Coffee Cup, a regular discussion of marketing news, culture, and media for our complex digital lifestyle. Join Matt Bailey as he engages in conversation to find insights beyond the latest headlines and deeper understanding for those involved in marketing. Grab a cup of coffee, have a seat, and thanks for joining.
[00:00:53] Matt Bailey: Well, hello, listener and welcome to another edition of the Endless Coffee Cup podcast. And I have a past guest with me again today, Dr. Augustine Fou. Dr. Fou, how are you doing today?
[00:01:06] Dr. Augustine Fou: Hey, Matt. Very good. Glad to be here with you again.
[00:01:09] Matt Bailey: Oh, thank you. And, you know, we’ve been going back and forth ever since, you know, your past visits on the podcast, and I love seeing your updates on LinkedIn about ad fraud, analytics, I, you’re doing a great job of keeping the industry on their toes about what’s going on in ad tech and ad fraud.
[00:01:31] Dr. Augustine Fou: Thank you. Well, I’m a, I’m a scientist so, you know, if I, if I can’t find the numbers to back it up, then it’s just a myth or theory or something. So, I always like to see if there’s any kind of numbers that make it concrete and real.
[00:01:45] Matt Bailey: You know, that’s, uh, a great statement right there because the other night I was reading something and, I think it was probably one of your posts and this was just going through numbers. And I just, it explained, I’m like, “Everything’s fake. Everything is fake.” Anytime someone gives you any type of number at all, or, you know, even looking at certain videos, things, I’m just overwhelmed sometimes with just how much is fake, and…
[00:02:14] Dr. Augustine Fou: Yeah.
[00:02:14] Matt Bailey: It used…
[00:02:15] Dr. Augustine Fou: I mean, in digital it’s just so easy. Right? So…
[00:02:18] Matt Bailey: Yeah.
[00:02:18] Dr. Augustine Fou: You know, everyone’s heard of the fake influencers where they’re buying likes on their Facebook page or they’re buying more views for the YouTube videos. Because I think people in the early days got a sense that, “Oh, if this video has more views, then it must be more popular,” right?
[00:02:34] Matt Bailey: Right.
[00:02:34] Dr. Augustine Fou: So, what do those people who cheat do? They just basically go on and buy more views, and those views are generated by bots, not by real people. So, if they can juice their numbers, it makes it look better, right?
[00:02:45] Matt Bailey: Yeah.
[00:02:45] Dr. Augustine Fou: So, there’s been a lot of that going on, and we now have probably, you know, bad guys have 20 years of experience doing that. So, in digital, it’s almost like you have to assume it’s fake first until proven real, right? And you know, you got to have that common sense, right? Sometimes in, in my line of work, when we’re talking about digital advertising, you know, if we’re talking about some of the numbers being thrown around by the ad tech companies, you know, the one company claims to be looking at 15 trillion bid requests per week. Just think about that with common sense. How many human would it take to generate that much? Right? It’s really hard to get humans to go to your website when you want them to.
[00:03:24] Matt Bailey: Right. Right.
[00:03:24] Dr. Augustine Fou: And it’s super easy to buy bot traffic, right? You can tell the botnet, “I need 10 million page views on my site by tomorrow,” and they’ll do it faithfully for you, right? So, it’s very easy to fake those metrics and that’s why you have to be constantly vigilant about fake numbers and fake everything online.
[00:03:41] Matt Bailey: Oh, absolutely. And, you know, it, it led to, I think, a great teachable moment with my kids in, you know, “Let’s talk about Instagram,” and yeah, there’s the fake follower counts, but then you throw on filters, then you throw on and, and, and there’s one account, I forget what it’s called. I think it’s called “Influencers in the Wild” and it’s people who have stumbled upon an influencer, either taking video of themselves or having people around them taking videos of themselves and they just look ridiculous because they’re…
[00:04:18] Dr. Augustine Fou: Yeah.
[00:04:18] Matt Bailey: …dancing or something in public, but they’re also doing four or five or six or more takes of the same exact thing.
[00:04:25] Dr. Augustine Fou: Yeah. That’s the reality, right? You don’t see that. You know, he, living here in New York City, you see that all over the place, right? And there’s actually almost every single day here at the New York public library, there’s people doing, uh, Instagram stuff on the steps because…
[00:04:38] Matt Bailey: Oh.
[00:04:38] Dr. Augustine Fou: …it’s a, it’s a landmark location.
[00:04:40] Matt Bailey: Yep.
[00:04:40] Dr. Augustine Fou: So yeah, absolutely. You know, you, you only see the final product and of course you only see the best shot that they want you to see, right? It doesn’t reflect the reality of the 16 takes they had to, uh, do to get that one shot that’s perfect. Right?
[00:04:54] Matt Bailey: Yeah.
[00:04:55] Dr. Augustine Fou: People are going to post, uh, good looking stuff. They’re not going to post realistic stuff. So, what you’re getting is a world that’s, uh, displayed to you through rose colored glasses. It’s not…
[00:05:05] Matt Bailey: Oh, absolutely.
[00:05:06] Dr. Augustine Fou: …really reflective of reality.
[00:05:07] Matt Bailey: Absolutely. You know, we used to have a saying didn’t we, back in the day that, you know, “I’ve got to see it to believe it.”
[00:05:15] Dr. Augustine Fou: Yeah.
[00:05:15] Matt Bailey: And that’s been turned on its head. You absolutely can’t believe what you’re seeing.
[00:05:20] Dr. Augustine Fou: Exactly. I, I think we’re waiting for the courts to no longer accept photographic evidence or videographic evidence…
[00:05:29] Matt Bailey: Wow.
[00:05:29] Dr. Augustine Fou: …because now, you know, we’ve known for years with Photoshop you can manipulate images, but now, you know, there’s more powerful tools and there’s AI, you’ve heard of deep fakes…
[00:05:39] Matt Bailey: Yeah.
[00:05:39] Dr. Augustine Fou: …where they can literally swap Obama’s face onto some other person talking and then make it look like he said something that he never actually said. So, that’s the world we live in, and these tools have been available to hackers and bad guys for a long, long time. And now they’re becoming more and more mainstream. They’re easy to use. In fact, a lot of those face swap apps, uh, in Google Play Store or whatever, allow just regular people to do that. And that’s extremely dangerous if you don’t know that that’s possible, right?
[00:06:11] Matt Bailey: Absolutely.
[00:06:11] Dr. Augustine Fou: ‘Cause you’ll think that some person said this, when they never literally said that.
[00:06:16] Matt Bailey: Wow. Yeah, it’s amazing what is happening with technology and, and all of these things and, you know, it, it, it used to be, it was just video out of context that, you, you know, you see it, you make a judgment, but then you see, “Oh, there’s another minute that preceded what I saw…”
[00:06:34] Dr. Augustine Fou: Yeah, exactly.
[00:06:34] Matt Bailey: …or “That was after what I saw,” and we had to struggle with context, but now we…
[00:06:40] Dr. Augustine Fou: They can make it up entirely.
[00:06:41] Matt Bailey: Do whatever, yeah, absolutely.
[00:06:43] Dr. Augustine Fou: Yeah. They can fabricate it. I mean, and that, that mirrors some of the stuff that I see, you know, so faking a video is actually harder, right? You do need a video processing software and all that kind of stuff. But faking data, for example, in digital marketing is super easy.
[00:06:57] Matt Bailey: Right.
[00:06:58] Dr. Augustine Fou: In fact, we’ve seen many cases where some of these ad tech vendors were literally fabricating the log files. So, in an effort to prove that ads ran somewhere, they literally made it all up, right? The ads never ran. They never paid any money. They literally made up all the log files to make it look like they ran ads. And, you know, it’s all just bits and bytes, and it’s all just computer processing power.
[00:07:21] Matt Bailey: Wow.
[00:07:22] Dr. Augustine Fou: So, you know, you should not trust any of that.
[00:07:25] Matt Bailey: You know, it’s funny because when you say that, years ago, I would have loved to have one of those log files because when I, when I’m looking at analytics companies or when I’m looking at analytics, you know, what analytics should I use, I want data to populate in order to see what it looks like.
[00:07:42] Dr. Augustine Fou: Yep.
[00:07:42] Matt Bailey: And one of those log files would be extremely valuable for a test case and for some sample data.
[00:07:48] Dr. Augustine Fou: Yeah. And kind of like you said, we should no longer trust videographic evidence or photographic evidence. Um, that principle holds true with even log files now because of the ease with which bad guys can just create them out of thin air, right? It’s just letters and numbers, uh, in a database somewhere. Uh, they can make it look pretty realistic.
[00:08:07] Matt Bailey: Absolutely.
[00:08:07] Dr. Augustine Fou: But that’s why I don’t take anyone else’s log files because the other principle you should remember is kind of like in a crime scene investigation, something called chain of custody, right? So, if you don’t know who touched the evidence or, you know, in the chain, you can’t rely on it anymore because someone could have tampered with it. Right?
So, chain of custody of the data is extremely important. And that’s why I don’t take historic log file data because I don’t know what’s provenance. I don’t know who touched it, who created it, how it was collected, whether there was any security mechanisms in place to prevent bad data being injected in there.
So, for some of the studies that I do, I have to put my code in for analytics, right? So, we gather the data real time ’cause I understand what the limitations of the data collection are and whether there are any vulnerabilities or security loopholes that could be exploited. And knowing that gives you more confidence in the accuracy and reliability of the data. And without that you can’t draw any reliable insights or conclusions from it because it could be entirely faked.
[00:09:12] Matt Bailey: Oh, absolutely. And that’s one, I, I know when I’m teaching analytics, it’s one of the areas that I start with is let’s just talk about accuracy and reliability…
[00:09:23] Dr. Augustine Fou: Yep.
[00:09:23] Matt Bailey: …just from the standpoint of collecting data. And when you think about, you know, bot traffic, but also when you think about just the accuracy of collecting data on humans, themselves, that you’re, you’re 100% going to capture all the bot traffic, but it’s humans that are confusing. It’s humans that are difficult to wrap your mind around, what they’re trying to do, why are they going here…
[00:09:47] Dr. Augustine Fou: Yeah.
[00:09:47] Matt Bailey: …there’s no sense to what they’re doing, whereas bots are very predictable. And I remember, I think it was early 2000’s I had a client that they paid for a link or an advertisement, and of course what happened, what, and what the trigger was for me, the red flag was they guaranteed at least 1,000 visitors a month.
[00:10:00] Dr. Augustine Fou: Yep.
[00:10:12] Matt Bailey: And I, and I said to the, the client, I’m like, “Wait a minute. You can’t guarantee that.”
[00:10:17] Dr. Augustine Fou: Yes.
[00:10:18] Matt Bailey: And what was interesting is when we started looking at their analytics, there was exactly 1,000 visitors a month.
[00:10:24] Dr. Augustine Fou: Exactly.
[00:10:26] Matt Bailey: And we’re, and we, and they all came on the same day and they all spit and they all, like, one page view. And, and this was, this was early 2000’s…
[00:10:34] Dr. Augustine Fou: Yeah.
[00:10:34] Matt Bailey: …and I’m trying to show the client like, “You just got rolled.”
[00:10:37] Dr. Augustine Fou: Yep.
[00:10:37] Matt Bailey: And it’s not difficult to do.
[00:10:39] Dr. Augustine Fou: And I think in just that one example, you illustrated a couple of really key points, right? So, first thing would be guarantee, right? Force 1,000 humans to go to your site when you want them to. Right?
[00:10:50] Matt Bailey: Right.
[00:10:50] Dr. Augustine Fou: So, first of all, common sense will tell you that that’s not possible, you know, but bot traffic, it’s very easy, right? You literally tell the botnet, “I need exactly 1,000, uh, page views on my site,” right? And you could even pass all the query strings to make it look like it came from that campaign.
So, you know, 20 years ago, 10 years ago, all of this stuff is just regular practice for the bad guys. And good guys don’t even realize that’s possible. And the other thing in, in your example is that you kind of, when you looked at the details, you kind of saw all of those thousand visits come on the same day. You can’t get, you know, all of it to occur at the same time, except, you know, with humans, but you can do that with botnets.
The other problem I’m going to highlight here is the problem of averages, right? So, if you look at just the entire month period, you got your thousand, but you know, you think it’s spread out over multiple days, right? You get a few per day and whatever, whatever, and by the end of the month, you’ve got your thousand.
But if you actually had the daily breakdown or the hourly breakdown, and you saw all thousand came in the same hour, then common sense will tell you, uh, that’s, you know, not possible. Right? So, one of the key principles that I teach is, you know, averages really are not good, meaning they hide the fraud, right?
If you are able to get more detailed breakdowns, like by day or by hour or in some of the place… …when we do programmatic campaigns, right? If you look at each individual domain, as opposed to a blended average across everything, you will see that some of these domains have 100% click through rates or they’ll have a volume of ad impressions served in the overnight hours when humans are sleeping. Right?
So, little things like that allow you to very easily pick out the fraud, just with your own analytics, right? You don’t need any kind of specialized tools, but if you just look more closely and get more details, you’ll be able to see that some of this stuff just doesn’t make any sense.
[00:12:45] Matt Bailey: Absolutely. And that gets to probably I think the number one thing and the number one need that I’ve seen in marketers, and I’ve heard this from numerous other guests on the show, other articles, and it’s the need for marketers to understand basic analytics.
[00:13:04] Dr. Augustine Fou: Yep.
[00:13:05] Matt Bailey: That they don’t have to be data scientists, but just like what you said to go look at, where did these visitors come from? What was the domain? What days did they come? Knowing where to go find basic information and understanding what you’re seeing is such a critical skill for marketers. No matter what, if you’re running a campaign, you need to understand the data that it’s producing.
[00:13:29] Dr. Augustine Fou: Yeah, and I think it’s, it’s in direct contrast to the way I’ve seen digital marketing being done in more recent years, right? With programmatic, everything just felt easy. Right? You can put some money in, and you’ve got all these results out. Right? The results we’re talking about are not actual sales. The results are typically reported as number of clicks, number of page views on your site, and that kind of stuff.
So, in that particular case, you know, I kind of call it, some marketers are, are doing digital marketing as if they were playing a video game, right? It’s almost like, “Oh, we got these shiny colors on a dashboard and more numbers are, you know, bigger numbers are better.” Kind of like a higher score is better.
[00:14:07] Matt Bailey: Right.
[00:14:08] Dr. Augustine Fou: But you can’t actually do that. You can’t set it and forget it. You actually have to look at the analytics yourself to make, to see if it even makes sense. Right? But a lot of the largest marketers have kind of outsourced that. So, whether it’s an analytics department, that’s a different department than your marketing people, right? That’s not a good idea. Or you hand it off to your agency and then what are they going to tell you?
They’re going to tell you everything’s working swimmingly, so don’t worry about it. Keep spending. So, you can’t actually set it and forget it and think it’s someone else’s responsibility, right? If you’re spending millions of dollars in digital marketing, you should be looking, uh, at the analytics. And again, you don’t have to be an expert. You don’t have to spend 12 hours a day looking at it like I do, right? Or other analytic people, but you have to at least look at it to see if any of it makes sense.
And better yet, you know, really bring in sales and outcomes data, if possible. Right? I realize in some of the bigger companies, those are again, different departments, right? It’s the CFO’s department that’s tracking sales and all that kind of stuff, and sometimes you don’t have ready access to that. But in any case, uh, you should be aware of the risks and limitations of using quantity data. And especially if those quantities are digital things like clicks or page views or a number of impressions or things like that. Alright, any kind of quantity metric is so easily faked by bots and other forms of fraud and digital, that again, you need to be sensitive to the fact that it may not be 100% reliable.
[00:15:39] Matt Bailey: Yeah, that’s absolutely true, and, and that’s what that understanding of analytics provides you, is that really the, the being equipped to do the job, to understand those things, and I, I’m seeing more and more marketing groups, companies asking for that basic digital…
[00:16:02] Dr. Augustine Fou: Yep.
[00:16:02] Matt Bailey: …understanding, the analytics, and it, it’s been such a refreshing, I, I think analytics has always been in demand, but I think more and more people are understanding that this is a key skill…
[00:16:12] Dr. Augustine Fou: Yep.
[00:16:13] Matt Bailey: …for the next 20 years for this industry and, and for this type of marketing.
[00:16:19] Dr. Augustine Fou: Yeah. And I think in digital, you know, the early days the theory was in digital, everything’s measurable. Okay? But what if the things you’re measuring are not the right things, right? And what if the numbers that you’re seeing in the analytics are, you know, unreliable, meaning they’ve been tampered with and stuff like that.
So, not only do you need to have a working knowledge of it, right? You don’t, again, you’re, you’re not an analytics person who has to do this 12 hours a day, but as a marketer, this is a core skill that you can’t just outsource. Right? And it, it’s not only understanding the basics of analytics, but also getting some kind of a, uh, gut feel for what is realistic and what’s not. Right?
So, in the early days, the reason I kind of got into the fraud side of things was I was just looking at click-through rates that were 70%, 100%, 50%, and having been in digital marketing long enough, we know that humans don’t click on ads that much. Right?
[00:17:12] Matt Bailey: Right.
[00:17:12] Dr. Augustine Fou: So, for banner ads, the, you know, rough ballpark is 0.1%, so 1 in 1,000. For search ads, it’s a little bit better, it’s probably, you know, 1 in a 100, so about 1%.
[00:17:23] Matt Bailey: Yeah.
[00:17:24] Dr. Augustine Fou: But if your orders of magnitude out of whack, right, when you’re seeing click-through rates on banner ads reported as 50%, 70%, or even 100%, something’s wrong with that. It’s not that your campaign is doing so awesome that that many people clicked on it.
It’s because it’s all bot activity. Right? So, that’s when I started looking into this. That doesn’t make any sense and I’m kind of encouraged because, you know, the younger marketers coming out, they’re not kind of fettered with old habits, bad habits, right? And some of the older marketers, they come out of TV and whatever. So, they think, “Oh, wow, it’s so magical in digital that you get all these numbers,” but they don’t even ask the question, “Are these even real?” So, I’m glad that the younger marketers coming out are actually starting to ask these harder questions. They’re looking to train themselves, right?
[00:18:11] Matt Bailey: Right.
[00:18:11] Dr. Augustine Fou: Take courses and things like that like your analytics course, to bolster their knowledge and, you know, common sense about what is realistic and what is not in analytics, ’cause that’s a core skill that they got to have.
[00:18:24] Matt Bailey: Absolutely. And one of the things that I’ve, I’ve changed in my training just to get the, the point across is I offer five different definitions of an impression and what is, and, and I ask them, “What is your organization’s definition of an impression?” And it’s very interesting that out of the five options, I will give an, a nearly even distribution of answers.
[00:18:52] Dr. Augustine Fou: Interesting. Yeah.
[00:18:53] Matt Bailey: Because they all sound good. It’s, it’s, you know, “It’s an ad that’s been seen,” people like that. And, and then I’ll use the Facebook definition of anything greater than zero pixels and zero seconds.
[00:19:08] Dr. Augustine Fou: Yes.
[00:19:08] Matt Bailey: And it, it, it, it really shocks them.
[00:19:11] Dr. Augustine Fou: Yeah.
[00:19:12] Matt Bailey: That, “Internally in our company, when we talk about an impression, I’m thinking of it from a human term.”
[00:19:18] Dr. Augustine Fou: Yep.
[00:19:18] Matt Bailey: “But we’re buying digitally from a company that defines it this way.”
[00:19:23] Dr. Augustine Fou: Yep.
[00:19:24] Matt Bailey: And…
[00:19:24] Dr. Augustine Fou: And you don’t check it, right?
[00:19:25] Matt Bailey: Right.
[00:19:25] Dr. Augustine Fou: You don’t go to check it, so it’s whatever metric they happen to say they like, and they’ll report it to you, and unless you have an independent way of verifying it or checking it, you’re beholden to them, right? You have to trust that their numbers are accurate and that they’re not cheating. Now, some of the bigger companies, obviously they’re not actively cheating, but you can imagine the fraudsters are deliberately cheating, right?
[00:19:47] Matt Bailey: Oh, yeah.
[00:19:47] Dr. Augustine Fou: So, there are so many techniques, right? One of the vul-, key vulnerabilities of Google Analytics is that anyone with your, U, UTM code can actually, uh, write bad data into it.
[00:19:58] Matt Bailey: Wow.
[00:20:00] Dr. Augustine Fou: So, I’ve seen cases where you go out and say, “Oh, I’m running a campaign. I want to get so much traffic.” They can literally write fake data into your Google Analytics to make it look like you got the traffic, when they didn’t even have to send a bot to load your webpage. Right? They even saved the time, they saved the effort.
[00:20:16] Matt Bailey: Well…
[00:20:16] Dr. Augustine Fou: They don’t even need bots to do it. They can just manipulate your GA to make it look like you got the traffic that you were promised. So again, back to your case of, “Oh, how can you guarantee that?” “Oh, I can certainly guarantee that because I’m using bot traffic, right?”
[00:20:32] Matt Bailey: Oh, that’s amazing. That is so amazing. And, and it gets to the point there that if my definition of this term, my definition of impression or view does not match internally within my company where I’m buying it from, then we’re speaking a different language. We’re reporting different things if my management has the perception that an impression is a physical human view, then we’re not even talking about the same thing.
And so, it starts within organizationally that shared understanding of, “What do these terms mean? What are we actually buying? And how do we understand that and apply that to our expectation or our outcome of the campaign?”
[00:21:19] Dr. Augustine Fou: Yep. And once you can narrow down those definitions and agree on which is the one you’re going to be using, then you can actually start measuring correctly.
[00:21:28] Matt Bailey: Right.
[00:21:28] Dr. Augustine Fou: And then you can actually see, uh, if your digital marketing is actually generating the outcomes you thought it was generating. Because a lot of times, you know, here’s another concept I like to talk about, which is correlation versus causation. So, there are many, many cases where there are sales happening when there are digital marketing campaigns happening. Those are just happening at the same time. It doesn’t mean that the digital marketing campaigns were causing the sales. Right?
[00:21:58] Matt Bailey: Right.
[00:21:58] Dr. Augustine Fou: So, for a lot of, a lot of the largest companies, this affects them disproportionately because they sell soup and soda in grocery stores. They don’t sell online, so they don’t usually have that feedback loop of, “Did these ads actually drive sales of soup in the grocery store?” So, they’re just assuming that it does, but at best it’s correlated, right? You’re running digital marketing at the same time that the sales were occurring. Right?
Remember the famous P&G example where they turned off $200 million of their digital spend and the sales just kept happening. So, you could clearly see that the digital marketing, right? The $200 million of digital, spent in digital did not cause any of the sales. The sales of the soup and soda or Tide or detergent or whatever was going to happen anyway.
[00:22:47] Matt Bailey: Right.
[00:22:48] Dr. Augustine Fou: So, for marketers, they also need to understand the difference between correlation and causation, and they also need to understand incrementality, right? Did any, every dollar that I spent in digital or any other form of marketing cause incremental sales, which means, say, the amount of sales that would have happened anyway? And I think too few marketers are actually taking that extra step and really looking at incrementality.
[00:23:13] Matt Bailey: Absolutely. And that’s, I remember the last time, one of the last times we talked, that was a challenge that you gave, you know, at the beginning of COVID, when everything’s up in the air, people were asking, “How are we going to respond to that?” and one of your pieces of advice was shut everything off. See, what’s working and, and see if sales are still happening there. And I’m wondering if that’s part of the, you know, one of the things we talked about is you’ve seen that things are starting to shift, that, that people…
[00:23:42] Dr. Augustine Fou: Yes.
[00:23:42] Matt Bailey: …we’re starting to respond to this, and, and I’m hoping that part of that was your advice to shut it off and get a sense of what’s your baseline. What’s…?
[00:23:51] Dr. Augustine Fou: Yeah, well it may, it may not have been my advice per se, but, you know, because of COVID, you know, big advertising pausing their stuff, right, in the height of it in middle of 2020, they did pause some of their digital spending because, you know, people were literally not going out, so they couldn’t go buy stuff, even if you wanted them to, right? So, they paused their campaigns in the middle of 2020. They started to realize, “Okay, well nothing’s changing,” right?
So, in there, kind of under, about to turn stuff back on, they’re being a little bit more cautious about it. So, an uptick in Q4 of 2020, and then throughout all of last year where the marketers are starting to be more vigilant, right? They’re looking more closely, right? The number of marketers that came to me to say, “Can you help us with some audits,” right? “We want to take a closer look.”
That really happened in 2021, last year. And I’m encouraged that that’s going to continue through this year and, and beyond. And as they’re looking more closely, they’re starting to realize some of the stuff that we assumed was working, uh, wasn’t working at all, or wasn’t, at least wasn’t working the way they expected. Right? So, now they’re being more cautious for more of those details, like we said earlier, right? So, they’re finally asking for police reports with domain level breakdown.
[00:25:13] Sponsor: As a regular listener of the Endless Coffee Cup, you know that I am passionate about learning and teaching others. With that, I’ve launched my own education portal for marketing professionals. It’s at learn.sitelogic.com. I’m creating and launching a series of courses on analytics. First course is how to create an analytic strategy that aligns to your business objectives.
The second course is six analytics skills that turn data into insights. The third course, launching in just a few weeks, is how to eliminate data overload and improve your presentations. This final course teaches you how to speak the language of the C-suite in your presentations, making you more persuasive, powerful, and effective in your marketing role. You’ll also find short topics on data, Google algorithms, and local SEO.
And finally, the centerpiece is the OMCA certification course. This is a 40-hour coached course that provides you with a solid digital marketing foundation. At the successful conclusion of this course, you are eligible to sit for the OMCA exam, and you’ll receive a free voucher for the exam, a $225 value that is included with your course enrollment. I’m an advocate of the OMCA certification, as OMCA certified marketers make 16% to 25% more than their non-certified coworkers. The OMCA is proof of your skills and knowledge in this industry. Go to learn.sitelogic.com right now to accelerate your marketing career.
[00:27:04] Dr. Augustine Fou: Right, previously they would just get an Excel spreadsheet at the end of the month that said, “Here’s the 10 billion ads that you bought. And here’s how many, you know, millions of clicks you got.” And none of it gave them enough details to do anything with.
[00:27:15] Matt Bailey: Right.
[00:27:16] Dr. Augustine Fou: So, getting back to our point about the averages, right? So, all the fraud was right there, but because averages and total numbers per month were, you know, reported to them, they couldn’t see any of it. Right? So, now that they’re getting more detailed reports and they’re looking more closely, they’re asking harder questions, that’s actually starting to gather steam and we’re starting to see more marketers realize, “Okay, well, this wasn’t working the way we, uh, thought it was working.” And I think a lot of that was because prior to the pandemic, no one wanted to be the one to rock the boat. Right?
[00:27:49] Matt Bailey: Right.
[00:27:49] Dr. Augustine Fou: Everything was just going on and we’re getting huge click through rates and lots of clicks, so no one wanted to be the one to rock the boat, but because the virus rocked the boat, now everyone has an opportunity to see it. And that’s kind of what led to my advice last year, which was, okay, here’s an opportunity for you to turn it off and see what happens, right? ‘Cause previously you wouldn’t be the one to say, recommend to your boss, “Let me go turn off my digital marketing.” Right? Right, now they, they have those kind of pauses where they could actually see the difference.
[00:28:19] Matt Bailey: Absolutely. And, and I, I’m hoping that part of this is there’s more emphasis on outcome. What is it that we are trying to drive, and is it actionable? And, and like you were saying, so much of this data that’s produced is not actionable data.
[00:28:36] Dr. Augustine Fou: Yep.
[00:28:36] Matt Bailey: It doesn’t, what do I do? That is, it’s the most basic question, and, and that’s the promise of the data, that analytics data, all the, “You’re going to get so much data you won’t know what to do with it.” Well, that’s exactly what’s happening.
[00:28:50] Dr. Augustine Fou: Yeah, exactly. I, I have a perfect example for you, right? So, a lot of the fraud detection companies are, are what I call black box, right? Black box means they simply report you a number and then they don’t tell you how they, you know, got that number, right? Why was something marked as fraud or why was something marked as not fraud? So, when they report a number like percent IVT, so say it’s 1%, say it’s 5%. What can you do with that, right?
[00:29:12] Matt Bailey: Right.
[00:29:12] Dr. Augustine Fou: You really can’t do anything except try to go get a 5% refund. Right? But then the other thing that happens is that the person on the opposite end will say, “Well, my vendor says it’s only 1%.”
[00:29:23] Matt Bailey: Right.
[00:29:23] Dr. Augustine Fou: And because that vendor is also black box, then you’re just stuck in a he said, she said, right? “My vendor says it’s 5%, I want 5% back.” The other person says, “Oh, my vendor says it’s 1%, so we’re not going to give you 5% back.” So, because it’s black box, because you can’t troubleshoot, because you can’t understand how they got the number, it’s not really actionable.
[00:30:00] So, the alternative is when you’re looking at your own analytics, right? You don’t have to use any specialized tools. If you’re looking at your own analytics, you can actually see the sites and the apps that are the most egregious cheaters. Right? So, you’ll see a site have a ton of clicks and all that kind of stuff, and then the click through rate’s 90%.
Okay. When you see that level of detail, you can say, “Well, something’s wrong with that,” and if you choose to, you could actually just add that domain or mobile app to your block lists, right? And what you’re doing then is cleaning up the campaign while it’s still running, rather than getting a fraud report at the end of the campaign to tell you what percentage was fraud. Right?
So, the difference between not actionable, right? Just getting a percent fraud from a vendor to knowing which domains and apps to turn off during the campaign, right? ‘Cause if you’re turning off those domains and apps during the campaign, you’re reducing the amount of money that flows to bad guys, so that you have more money left over to go to better sites and better apps, so then your campaign…
[00:30:44] Matt Bailey: Right.
[00:30:44] Dr. Augustine Fou: …can actually be more effective. So, that way you can monitor and manage, uh, your own campaigns. So again, you can do it with your own. So, my clients are for analytics to see more details than is available to them through GA, right, but that way they can monitor and manage themselves.
[00:31:01] Matt Bailey: Well, what you did is just outline a great, great comparison of fraud detection versus analytics. You know, just there are so many services out there for programmatic company or companies using programmatic to try and save something, but what you’re advocating is analytics is where you’re going to catch and get the data you need to react.
[00:31:23] Dr. Augustine Fou: Yes. And it’s common sense, and it goes back to an earlier point. You can’t set it and forget it. You can’t just buy one of these fraud detection services and think it’s done, right, that they prevented all of it. In fact, I’ve got lots of, lots of data that say they, they’re not catching most of it. Right? So, when they’re consistently reporting 1% IVT or invalid traffic, you shouldn’t assume that the other 99%’s good. It just means that they failed to detect the fraud in the other 99%.
So, what’s a better option, like you said, is looking at your own analytics, learning how to do that, and I’ve seen actual small business owners do a much better job than big corporate marketers, because for the small business, if they spend a hundred dollars in digital marketing and they don’t get any actual returns, they can’t afford to spend the next hundred dollars.
[00:32:12] Matt Bailey: Yeah.
[00:32:12] Dr. Augustine Fou: It’s very different than P&G or Unilever or any one of these large advertisers. They can spend a hundred million dollars in digital and not see any return, and they wouldn’t even know because they just assume it happens because the sales are happening while they’re doing digital marketing. Right? Again, it wasn’t caused by the digital marketing, but it just happened. So, they just assume that, uh, it’s related.
[00:32:36] Matt Bailey: Is, that’s amazing. One of the things you, you mentioned, and I, I think, I think this is part of what makes, especially the programmatic industry, it seems to run on headlines about spend.
[00:32:53] Dr. Augustine Fou: Yes.
[00:32:53] Matt Bailey: And, and you’ve talked about this a couple of times. There with the focus and, and again, we’re at the beginning of the year, and so it’s not uncommon to see headlines about how much is going to be spent…
[00:33:03] Dr. Augustine Fou: Yes.
[00:33:04] Matt Bailey: …on programmatic and, and P&G is going to spend this much on programmatic…
[00:33:08] Dr. Augustine Fou: Yes.
[00:33:08] Matt Bailey: …and, and it’s that emphasis constantly on spend, I, you know, you’ve said it…
[00:33:14] Dr. Augustine Fou: Yeah, I think as well…
[00:33:15] Matt Bailey: …and I said it, it’s the wrong thing.
[00:33:17] Dr. Augustine Fou: It’s the wrong thing. And I can actually, uh, offer an explanation for why people are so focused on spend. Okay, so this goes…
[00:33:24] Matt Bailey: Yeah.
[00:33:24] Dr. Augustine Fou: …all the way back to TV advertising. So, in the early days, when you put Tide on TV, you got more sales because you made more people aware of your product that weren’t formally aware of your product, right? So, in the good old days of TV advertising, when you spend more, right, you could reach more people, and the more people are aware of your product, the more people that will end up buying your product. So, that’s perfectly good.
In digital, however, you know, when you increase spend, it doesn’t necessarily translate into increased awareness because of the fraud problem. And most of this actually started, um, occurring in the last 10 years, right? So, let me again, let me take a step back. In the good old days of internet advertising, that’s meaning mid 90’s, you know, early 2000, whatever, when advertisers were buying from publishers. Right?
So, the buyers of the ads would go to a New York Times or Hearst or Conde Nast and actually buy ads from them. Then the ads would actually show on their sites. Then, because those sites had… and yes, that’s fine. Right? The ads were shown to humans. But when programmatic took off, in, you know, 2011, 2012, 2013, that’s kind of when I started, uh, focusing on the… that’s when the fraud increased and that’s because now, instead of the advertiser buying from the publisher, the advertisers now buying from a middleman, the ad exchange.
Ad exchange had its purpose because now you have hundreds of thousands of sites and the advertiser was not going to go negotiate ad deals with thousands of sites, right? So, they just wanted to go one place, and the ad exchange would basically aggregate all those sites. But because we’ve now dissociated the buyer, uh, from the seller and inserted a middleman called the ad exchange, it actually opened the flood gates to fraud.
[00:35:19] Matt Bailey: Yeah.
[00:35:19] Dr. Augustine Fou: And that’s, in the early days, you could see, I won’t name the exchange, but you could see this early exchange pretty much let everyone in. So, that means, you know, fraudsters could add tens of thousands of sites into that particular exchange, and start selling ads to the largest of advertisers, right? That wasn’t possible before, because a P&G would never buy from this no-name guy, you know, sitting in Brooklyn with 10,000 sites.
But now, because it’s all mixed together in an exchange, the buyer now says, “Here’s a chunk of money. Go spend it for me.” Right? And that exchange is going to find every possible place to put their ads, including fake sites. And those fake sites obviously have no human visitors, so all they’re doing is they’re buying 100% bot traffic. So, we’ve seen…
[00:36:05] Matt Bailey: Well, you’ve published, you’ve published…
[00:36:07] Dr. Augustine Fou: Yeah.
[00:36:07] Matt Bailey: …lists of some of these fake sites. And…
[00:36:10] Dr. Augustine Fou: Yeah. Just letters and numbers.
[00:36:11] Matt Bailey: …honestly, it’s ridiculous.
[00:36:12] Dr. Augustine Fou: Yeah.
[00:36:13] Matt Bailey: They’re ridiculous.
[00:36:13] Dr. Augustine Fou: No human would ever go to those, right? So, so I think, you know, just to round out the story of, because of the programmatic exchanges and because of programmatic advertising on open exchanges, right, in the last 10 years, that’s when the fraud is skyrocketed and that’s where a lot of the dollars end up flowing to fake sites, fraudulent sites, or the ones that are bent on spreading disinformation, right? So, we’ve seen over the years, piracy sites, you make a lot of money through digital ads, right? Those piracy sites steal the content, they don’t charge the users for listening to pirated music or watching pirated movies, they make it through advertising.
So, if these ad dollars are flowing to the piracy sites, or if the ad dollars are flowing to disinformation sites, it allows them to make money and it allows them to keep going. Right? So, now we’ve seen a proliferation of hate speech sites, disinformation sites…
[00:37:05] Matt Bailey: Yeah.
[00:37:05] Dr. Augustine Fou: …and so on and so forth. They now have a revenue source. So, that’s why the brand safety issue is now tied in with ad fraud, because again, these sites, not only are they spreading disinformation, they can also juice their own numbers by using bot traffic. Right? So, it all ties together and just making things worse and worse.
[00:37:24] Matt Bailey: Well, and to, to put another layer on that, when you bring in brand protection, again, we have services out there that are saying, “We’ll protect your brand from showing on these sites,” and they, they’re not doing it. They, they can’t do it.
[00:37:41] Dr. Augustine Fou: Yeah, you can just assume that the bad guys’ bots and the bad guys’ techniques are always better than the good guys’ techniques. And it’s not a knock against the good guys. I mean, they’re trying hard. They’re really desperate, but here’s the, here’s the reality of the situation. Someone working at a fraud detection company, right? Assuming they’re tuning the algorithms and they’re looking at the fraud.
If they say, “Oh, we just found this new kind of fraud we want to build into the algorithm.” Right? How long do you think it’s going to take to get corporate approval right after testing and all that kind of stuff to roll that out into production, right? So, for the good guys, you can think of that as a three to six to nine month process before they can roll out a new detection into the official production algorithm, right? Whereas for the bad guys, if they see that their bot is not making money, by the next minute they’re on it. Right?
[00:38:31] Matt Bailey: Yes.
[00:38:31] Dr. Augustine Fou: And they’ll fix it within five minutes so that their bot can continue to make money. So, for the bad guys, again, these are hackers. They make the bots, uh, they’re very, very quick at, uh, troubleshooting and fixing the problem, right, so they can actually get around the defenses so that they’re bot can continue making money. So, it’s really not a fair fight. The bad guys always have the advantage of, of only, of not only the skill, but also they can change things very quickly and they don’t have to play by the rules.
[00:38:58] Matt Bailey: Right.
[00:38:58] Dr. Augustine Fou: Whereas the good guys have all these constraints like corporate, you know, processes and approval processes and whatever. They’re always at a disadvantage. So again, if you can find a way to not have to rely solely on these black box fraud detection vendors, whether it’s fraud detection or brand safety detection, whatever, and you have a way to look at your own analytics, I mean, for the brand safety stuff, it’s like, okay, well, did your ads end up on a hate speech site? Or did your ads end up on a coronavirus disinformation site? If you had the list of domains where your ads actually ran and you had enough detail and you just took a few minutes to look through those place reports, you can solve this yourself.
[00:40:00] You don’t have to pay extra money to a brand safety detection company and think that they’re actually catching it all, because they’re not. Right? And we’ve seen many examples where they’re blocking the front page of New York Times and Wall Street Journal because of their crappy detection. Right? They make it out to be oh, machine learning and AI and all that kind of stuff, but it all boiled down to the front page contained the word COVID-19. So, the ad was… right? So, it’s basic keyword list. It’s not any of the fancy stuff that they said it was. It’s just basic, dumb keyword list, right?
[00:40:15] Matt Bailey: Amazing.
[00:40:16] Dr. Augustine Fou: So again, just use your common sense. Look at the analytics. Do what small businesses do. Like actually manage your own marketing, digital marketing campaigns, and you’ll be better off than paying for all these additional services that don’t work the way you think they work.
[00:40:30] Matt Bailey: Yeah. And, and like you describe, just looking simply at the domain list could probably, you know, remove like 80% of what’s going on and focus your budget on the 20% of domains that are legitimate…
[00:40:45] Dr. Augustine Fou: Yes.
[00:40:45] Matt Bailey: …that are known that you can go look at…
[00:40:48] Dr. Augustine Fou: Yes.
[00:40:48] Matt Bailey: …and verify that this is made for a human and it’s where people that I would target would go. Um…
[00:40:55] Dr. Augustine Fou: Exactly. And, you know, there’s, there’s one thing that you might find interesting. I use this in class. I ask my students to name off as quickly as possible 10 sites you visit every day, right? So, they’ll start rattling off a few, but once they get to 4 or 5 or 6, you know, they, they can’t get past 7 or 8 because humans don’t visit more than 10 sites regularly every day. Right?
They might visit a recipe site once in a while when they need to look up a recipe, right? But they’ve, they rarely repeatedly visit sites, uh, that many sites. So, it’s usually the same, same set of sites. If you do the same exercise, right? Name off mobile apps that you use every day as quickly as possible. They will kind of slow down when they get to 7 or 8. So, common sense will tell you humans visit a very finite number of sites and use a very finite number of apps every single day. And those are the places you want to be, even if the CPMs are higher, because those are legitimate places, right?
The CPMs are higher for sure, but you don’t have to buy that much quantity, right? So, the favorite pastime of a lot of marketers in last 10 years is just buying billions upon billions of ads because they think more is better, right? If we spray our message out to more people, but again, they’re not people. They’re bots, right? So, if you just bought smaller quantity of ads on a higher quality site, even if the CPMs higher, you’re still going to end up spending less money than you do now and getting more outcomes.
[00:42:24] Matt Bailey: Amazing. You know, it’s funny. So, my background’s in journalism and I, and I’m, like I said, I’m fascinated with the headlines, the headlines that focus on spend. But I also wonder if from a marketing perspective, the narrative, the modern narrative of modern marketing, I don’t think helps, because I, I think there is some assumption if you’re drinking the, the branding Kool-Aid so to speak…
[00:42:51] Dr. Augustine Fou: Yep.
[00:42:52] Matt Bailey: …that there’s this assumption that people want to engage with your brand story or they, they want…
[00:42:58] Dr. Augustine Fou: Yeah.
[00:42:58] Matt Bailey: …you know, there’s this infection in marketing that…
[00:43:02] Dr. Augustine Fou: Yep.
[00:43:02] Matt Bailey: …consumers actually want to engage with the brand and I, I, I’m sorry, but like you, you, like you said, banner ads less than 0.1%, email 1% if you’re lucky, you know, click, uh, paid search, I’ve seen 4% or 5%.
[00:43:19] Dr. Augustine Fou: Yep.
[00:43:19] Matt Bailey: But that’s just the click through you, you know?
[00:43:21] Dr. Augustine Fou: So, let me kind of tie this together, you know?
[00:43:24] Matt Bailey: Sure.
[00:43:24] Dr. Augustine Fou: Any tall tale is rooted in some, uh, element of truth, right? Some nugget of truth. So, these are tall tales, right? The fact that, uh, branding and that people will want to engage with your brand. So, there are certain cases like Apple or certain cases like JetBlue in their early days, uh, people do want to engage with those brands because they earn the right to have a dialogue with those people.
[00:43:47] Matt Bailey: Right.
[00:43:47] Dr. Augustine Fou: But right now, a lot of marketers just think, “Oh, if we send out more banner ads and we target them to people,” that they would want to engage with you. No, that’s not the same thing, right? You don’t have any relationship with them. And this ties in with also the, the, the data on ad blocking, right? The amount of ads has gotten so bad over the last 10 years, that humans are actively blocking the ads.
[00:44:10] Matt Bailey: Yes.
[00:44:10] Dr. Augustine Fou: So, some data that I pulled recently, you know, on the desktop side, it’s reaching 30% ad blocking, right?
[00:44:17] Matt Bailey: Wow.
[00:44:17] Dr. Augustine Fou: So, humans are blocking your ads and bots are not because it’s their job to cause the ads to load. So, if you’re still doing ads and programmatic and not buying it from legitimate publishers directly, you’re disproportionately serving it to the bots that don’t block ads, because a larger and larger portion of the humans in that mix are actually blocking your ad.
So, you know, it just doesn’t make any sense if you’re not looking at outcomes, right? The quantity metrics are all completely out whack, and there’s a lot of hidden things like the fact that a third of the humans are blocking your ads. So, they’re not, they’re never going to see it anyway. Right? So, now you’re marketing disproportionally to bot traffic in programmatic channels.
[00:45:00] Matt Bailey: Amazing. Amazing. There’s a couple of things, so, you know, like you said, the tide’s turning and you’re putting out some really good information about what marketers can follow and some trends that you’re seeing, and one of the things that struck me the other day, I, it was what I was trying to find earlier. It’s iPhone users targeting iPhone users, turns out that you’re actually getting humans…
[00:45:22] Dr. Augustine Fou: Yeah.
[00:45:23] Matt Bailey: …when you’re doing that.
[00:45:24] Dr. Augustine Fou: Yep. So, I think the problem recently, you know, in the discussion around privacy is that Apple has made these moves where they’re doing away with the third-party cookies, or they’re actively deleting it in safari. Right? So, a lot of marketers have been told by their ad tech vendors that, “Oh, well, if you don’t have those third-party cookies, you can’t track, and therefore you can’t target anymore.” So, that’s fine. So, there’s an issue of the cookies, third-party cookies going away in the, in the reduction of targeting.
But, you know, I’ve also written about the topic of the targeting not being very accurate anyway, and I won’t get into that for, for right now, but even if you don’t have the targeting for iOS devices, if you just look around, right? A lot of humans are using iPhones and iOS devices, right? And that study that I posted was saying that teenagers disproportionately use iPhones, right?
[00:46:17] Matt Bailey: Right.
[00:46:17] Dr. Augustine Fou: And I can ask my kids and you know, your kids, you could see their, all their peers have iPhones, not Android devices. So, for the marketers, because of their fear of the reduction of targeting, they’re now not targeting iOS devices or not even bidding on it, right? So, now we have a huge, you know, amount of inventory that the advertisers are simply not bidding on it, or just deliberately bidding less on it. So, if you get your ad on an iPhone, even if you don’t have 15 targeting parameters, at a very minimum you’re getting your ad in front of a human.
And to me, that’s almost like the prerequisite. That’s the first step before any of those other targeting parameters, right? ‘Cause if you can’t get this first step right, show your ad to a human, none of those other targeting parameters are gonna matter.
So, for the, for the advertisers, it’s almost like an insider tip here, you know, set up a campaign line. Target iOS only, with nothing else, right? Maybe with some basic age or, you know, demographic information, like one or two parameters, you’re going to be much better off than trying to buy 15 targeting parameters, trying to target Android devices. And oh, by the way, from my research, bots love to pretend to be Android devices…
[00:47:29] Matt Bailey: Yes.
[00:47:29] Dr. Augustine Fou: …because they have cookies, they can earn more from the retargeting, the targeting, uh, parameters, the audience segments, all that kind of stuff. Whereas it’s much harder for bad guys to pretend to be iOS devices. So, you’re getting like a double bonus if you will, you know, getting your ad in front of humans and you’re much lower, at much lower risk of ad fraud when you’re advertising on iOS devices.
[00:47:52] Matt Bailey: That’s great. That’s great. And, and it’s one of those things that might be, you know, slightly counter-intuitive, but it makes complete sense that, you know, it’s harder to fake these are people, and, and yeah, I saw that same article. I think it’s like 70% of teens or something like that.
[00:48:08] Dr. Augustine Fou: Yeah, and, you know, I don’t put a lot of weight into the absolute numbers. So, just say…
[00:48:11] Matt Bailey: Right.
[00:48:12] Dr. Augustine Fou: …just say majority. Right? And if you just look around yourself with just common, everyday experience, yes, most, a lot of people like, you know, our doctor friends, you know, most doctors are on iPhones. They’re not on Android devices, right? So, you know, for some of my pharma clients, you know, if they just targeted iOS devices, they can get their ads to the doctors that they intend to get it to.
[00:48:33] Matt Bailey: That’s awesome. Okay. Hey, let’s, uh, try, I’m going to start wrapping this up…
[00:48:38] Dr. Augustine Fou: Ok.
[00:48:39] Matt Bailey: …but you brought up cookies, and I think that would be a great way to kind of finish off is what’s, what are you seeing what’s happening with cookies? I mean, it, the news has been a little quiet about this. What’s happening with cookies that we should be aware of and, and plan for in the next few months or years?
[00:48:55] Dr. Augustine Fou: Yeah, I think the, you know, the large companies like Google, they had previously announced plans to do away with third-party cookies in Chrome. And because Chrome represents two thirds of the browser share, browser market share, and that was a very significant move. But they pulled back on that, right? And so far, Apple has the, has been the only one to start enforcing some of these privacy things in iOS 15 and beyond, right? So, you should assume that cookies are going to go away. Right? We don’t know exactly when they’re going to go away, and we’re talking specifically third-party cookies, not first-party cookies.
[00:49:27] Matt Bailey: Right.
[00:49:27] Dr. Augustine Fou: And again, the difference is first party is set by the publisher, right? So, when you go to New York Times, you get a New York Times cookie. That’s totally fine. Third-party cookies are all the other ad tech vendors that are installed on the page and setting cookies and tracking people across sites without their knowledge. That’s the part that’s going away.
[00:49:46] Matt Bailey: Yep.
[00:49:46] Dr. Augustine Fou: So, you should assume that cookies are going away, but you should not panic that cookies are going away, because like I said before, there’s other studies. If people want to see that they can look on my LinkedIn, but the cookies were used for targeting, but the targeting was based on inferred data or derived data. Right? And they derive those audience segments from website visitation patterns. So, they’re trying to figure out who these people are, what they like based on the collection of websites or pages that they visit. And as you can imagine, some of that is actually terribly inaccurate.
[00:50:00] So, over the years, as a lot of marketers bought these audience segments or targeting parameters, they paid extra for it, what they didn’t realize is how crappy those segments were, how crappy or inaccurate the, the targeting was. So, when cookies go away, I actually think of it as an inadvertent reason for the marketers end up doing better digital marketing, because they’re no longer wasting as much money on targeting parameters and audience segments that didn’t work anyway.
[00:50:55] Matt Bailey: Yeah, absolutely. And, and one way I show my students this is we go look at the ad settings in their Google account, and I, I’ll post the link there, but you can go see all the segments that Google has placed you based on your browsing history. And one example I, I showed them is it says that I like cats, or I have an interest in cats. And I explained the only reason that’s there is because my, my oldest daughter had a cat, she moved away, and she couldn’t take the cat with her. And so, I’m posting the cat on places to get rid of it. And…
[00:51:35] Dr. Augustine Fou: Yeah.
[00:51:35] Matt Bailey: “Come take this cat,” get rid of it. I don’t want it, but because I was visiting those sites, Google now thinks that I have an interest in cats.
[00:51:43] Dr. Augustine Fou: Exactly. Right.
[00:51:44] Matt Bailey: And…
[00:51:44] Dr. Augustine Fou: And a lot of times we see that retargeting stuff, right? So, you visited…
[00:51:47] Matt Bailey: Yeah.
[00:51:47] Dr. Augustine Fou: …sites once, or you bought a, a baby gift for a friend who had a baby. Now all the algorithms think you had a baby, right? So, they keep retargeting you with, uh, the ads that are completely irrelevant to you. So, yeah, all of those things are basically algorithms making assumptions, and so, that’s not terribly accurate. I will say that on Google, a lot of users are logged in all day long, right? They’re logged into Gmail. They’re logged into YouTube. Google has better information about those users. And then similarly on Facebook, people are logged into Facebook or Instagram all day long. So, those two platforms actually have more accurate data based on historic usage of the humans.
Whereas a lot of the ad tech companies, because humans are not log-in to the content sites that they’re visiting, they are only looking at or collecting website visitation patterns, and then now trying to infer who they are and what they like based on website visitation. And that’s the part that’s not very accurate.
[00:52:45] Matt Bailey: Yeah.
[00:52:45] Dr. Augustine Fou: So, that’s what leads to the segments and targeting parameters sold through programmatic channels to be not very accurate.
[00:52:53] Matt Bailey: Absolutely. It’s always a, and, and that’s the thing, being in this industry and getting served an ad, especially when I go to YouTube, I’m always interested to see who’s targeting me and what kind of ad it is, because that tells me who people think, think I am.
[00:53:09] Dr. Augustine Fou: Yes.
[00:53:10] Matt Bailey: And, uh, there’s always a celebration when we get, uh, a foreign language ad because…
[00:53:14] Dr. Augustine Fou: Yeah.
[00:53:14] Matt Bailey: …we’ve been doing something else and, “Yay, they mistargeted us.”
[00:53:18] Dr. Augustine Fou: Yep.
[00:53:18] Matt Bailey: So, it’s, it’s always a fun day. So, what you’re saying, don’t worry about cookies, it can actually make our marketing better, and I love that message because it’s, it’s, it gets back to basic, “Let’s brand on a large scale instead of branding to 20 people who we think…”
[00:53:38] Dr. Augustine Fou: Yeah.
[00:53:38] Matt Bailey: “…our brand is, is perfect for.”
[00:53:41] Dr. Augustine Fou: So…
[00:53:42] Matt Bailey: Well Dr. Fou,
[00:53:42] Dr. Augustine Fou: …so if I were to wrap…
[00:53:43] Matt Bailey: Yeah.
[00:53:43] Dr. Augustine Fou: So, if I were to wrap this up, I would just remind people, you know, marketing is not a set it, forget it kind of video game, right? You really have to roll up your sleeves a little bit, right? Uh, and dig into some of the analytics. And when you have detailed enough analytics, your common sense and your gut feeling will tell you when something looks wrong, right? So, you can use those analytics to not only find the stuff that’s not good, but also corroborate that things are actually working well.
And I would encourage more marketers to kind of be like small business owners and focus on outcomes, right? To the extent that you can. I understand large companies have departments, departmental silos where it’s really hard to get all the data you need, but focus as much as you can on outcomes, so that you can actually see that your digital marketing is working, uh, well for you. So, we can now get back to real digital marketing after a decade of fancy looking metrics that don’t mean anything.
[00:54:40] Matt Bailey: Wonderful. That is a perfect, perfect wrap-up and summary of the discussion. Thank you so much, Dr. Fou for your time today. I really appreciate it, and again, just, you know, a solid hour of great, great information.
[00:54:53] Dr. Augustine Fou: Thank you, Matt.
[00:54:54] Matt Bailey: Thank you for your time.
[00:54:54] Dr. Augustine Fou: Glad to be here with you. Thank you.
[00:54:56] Matt Bailey: Alright. Thank you. And thank you, dear listener, for tuning into another episode of the Endless Coffee Cup. I look forward to seeing you again on future episodes. Thanks again.
[00:55:07] Bumper Intro-Outro: You’ve been listening to the Endless Coffee Cup. If you enjoyed this episode, share it with somebody else. And of course, please take just a moment and rate or review us at your favorite podcast service. If you need more information, contact me at sitelogicmarketing.com. Thanks again for being such a great listener.
Dr. Augustine Fou, Ad Fraud Investigator
Marketing Science Consulting Group, Inc.
LinkedIn profile: Dr. Augustine Fou | LinkedIn
Twitter Profile: Dr. Augustine Fou | Twitter
Listen to Dr Fou on previous episodes:
ECC Podcast #50: The Myth of Microtargeting
Upgrade Your Digital Marketing Career
Get a Digital Marketing foundation that will advance your career. A 40-hour coached course that prepares you for OMCA certification. (Exam fee included in course registration cost)