#059: Display Ads and Ad Fraud with John Nardone

March 28, 2017

Well into our third year of the show, we decided it was time to show that we’ve become hard-hitting audio journalists by bringing on a heavy-hitter in the world of display media and grilling him with tough questions. And then we found out we’re not hard-hitting audio journalists. And the heavy-hitter we brought on was informative, articulate, and willing to muse objectively about the challenges that face display advertising. So much for the original plan! John Nardone of Flashtalking was actually the first person every targeted by a DoubleClick ad, and that was almost 20 years ago! You will hear that story in this episode, as well as sage little gems about “the mythical allusion that the (media) agency is in fact an agent for the client,” as well as how “the technology is ahead of a lot of the advertisers’ ability to deploy it effectively.” It was a fun and informative discussion!

Episode Transcript

[music]

00:04 Announcer: Welcome to the Digital Analytics Power Hour. Tim, Michael, and the occasional guest discussing digital analytics issues of the day. Find them on Facebook at facebook.com/analyticshour and their website analyticshour.io. And now, the Digital Analytics Power Hour.

[music]

00:28 Michael Helbling: Hi everyone, welcome to the Digital Analytics Power Hour. This is Episode 59. You know, display advertising goes back as far as the web itself. If you put up a billboard along the information superhighway, just imagine the return on your investment. And so, display media was born, and lo, it was good. But then Loki, who was more cunning than all… Okay wait, someone’s been catching up on History Channel’s Vikings, but actually, on this show, we’re here to talk about that which has been causing concern amongst analysts for a long time. It’s about measuring display media, and the ad fraud that comes along with it. And so now, without further ado, please welcome my co-host and “Arrntrepreneur”, senior partner at Analytics Demystified, Tim Wilson. Hey Tim.

01:19 Tim Wilson: Arr, arr, arr, arr.

01:21 MH: Yeah, do you like that? [chuckle] And I am Michael Helbling, I lead the analytics practice at Search Discovery. So obviously, Tim and I needed to turn to an expert, someone who could help us ponder the mysteries of media and how to prevent malpractice. So, we managed to find a guest to help us sort this out. John Nardone is one of the first people to place a display ad on the web. He’s also a founding member of the IAB, currently he’s the CEO of Flashtalking. Prior to that, he lead X+1 to their successful acquisition by Rocket Fuel, and he was the President of Marketing Management Analytics or MMA, they’re the company that invented marketing mix modeling. Recently, he was named one of the 50 top private company CEOs in America. John, welcome to the show.

02:11 John Nardone: Thanks very much, I’m glad to be here.

02:14 MH: Likewise. So, we know that 50% of our money on advertising is wasted, which half is it?

02:22 TW: Is it the display half? [chuckle] This might be the first show where we have a guest hang up and walk off before… Right after the introduction.

02:33 MH: Yeah, let’s try to pick someone who… Never mind, anyways. Yeah, the question was to you, John.

02:40 JN: I think there’s waste all over the place in the display ecosystem, and you can define waste in so many different ways. You can look at the supply chain of how an ad gets from the publisher through the various ad tech platforms, through SSPs to DSPs to ad servers and look at the waste if you will, that’s accumulated along the way. You can look at fraud as waste or you could look at just ads that are mis-targeted, and targeted, that deliver to other than the audience to which they were intended. There’s waste all over the place.

03:14 JN: I think all of us who’ve been in the industry for the last few years at the, sort of the dawning days of programmatic, believed that technology, and programmatic technology in particular, would address these problems, and in some ways they have, but in some ways they’ve created more problems, more and different problems. While there’s a better ability than there’s ever been before to target an ad to its intended audience, now the programmatic environment has opened the opportunity for cheats to run all manner of scams and to create fraudulent ads, fraudulent clicks and, like all opportunities to cheat the system, it’s a game of cat and mouse between the good guys and the bad guys.

03:56 TW: So definitionally, just to make sure, when I think programmatic, I think that that is where I’ve got the machine that is trying to look at kind of everything I’m doing, and it’s looking at the response and it’s basically, should be pulling the levers on audience, potentially on creative with a feedback loop that is tuning that in a constant manner. Is that a fair definition of programmatic?

04:22 JN: Well, I think that’s the wishful definition of programmatic.

04:26 TW: That’s the… Okay.

[chuckle]

04:27 TW: That’s what gets sold as programmatic.

04:29 JN: Yeah.

04:31 TW: That’s kind of where I was heading.

04:31 MH: He just called your definition wishful, Tim.

[chuckle]

04:35 TW: Well I’ve watched media agencies sell exactly that, and I think I worked at an agency for a while that was selling that without even the technology to back it up. What is programmatic? What’s the state of programmatic today, then?

04:50 JN: I define it at a more basic level, it’s using data and automation to execute marketing transactions. The sophistication of that automation though is what’s always in question. I think as you say there’s a lot of over-promising in terms of the level of intelligence and auto-optimization and such that is really built into these systems.

05:13 TW: My impression is that as soon as an agency is saying, “We’re gonna do programmatic buying,” there winds up being a cost increase, because the idea is that because it’s being programmatic and not just kind of a one planner and one buyer saying, “This is what we’re gonna set it and forget it,” because it should be getting tuned it seems like, or my impression is that you wind up paying a premium for the promise of that more optimized experience rather than kind of the reverse, or what I would think would be the ideal which is, “Hey I don’t care how you do it, I’m gonna pay for the outcome.” Which I think works with direct response a little bit better, I can pay a little bit better for the transactions, although that’s still open to fraud as well. But I don’t know, is that fair?

06:06 JN: Yeah. I guess it depends on which angle you’re looking at it. Perhaps, it costs more in terms of the fees that you’re paying, ’cause now there’s a tech fee. And there’s some amount of labor that’s required to run the system and do the optimization, but the flip side is, you wind up actually, I think paying less in hard media costs most of the time because you’re able to buy the target audiences much more specifically that you want. And so you can buy that more inexpensively because you have more information about what’s important to you than the publisher has. So there’s an information advantage for the buyer, and then of course, you can avoid inventory that’s not useful to you. So while there is some higher fees, I think net, net, you do wind up with a more cost efficient buy if it’s run well.

07:02 TW: So, it’s gonna be hard for me to… Having just met you on this episode. I feel like I have been burned so many times by media, so I apologize if I’m sounding like the skeptic or the cynic. But when you say a well-run programmatic, do you have a sense in the industry of how many big media agencies are trying to jump on the programmatic bandwagon that don’t actually have either the technology, or the talent to do it effectively? And I’m coming from a deep and dark place in my recent history when I’m asking the question.

07:42 JN: Yeah. I think it’s like so many other things when you’re talking about big professional services organizations, it’s about what team do you get? Within a big agency, you can have one team that’s doing a great job for a client, and then the next client gets the C team, and it’s the same agency and one client’s getting great work and the other client’s getting shoddy work. And sometimes it’s about the client. There are good clients and bad clients. Good clients who have the right structure and manage things and give good direction, and then there’s bad clients that nobody wants to work for and it’s a drudgery and a chore every time you gotta work on their business and they get poor service. I think there’s a pretty broad range, but I don’t know that you can say one agency is good, one agency is bad, this trading desk is good or this trading desk is bad, I think there’s really good people and there’s also not so good people at all of them.

[laughter]

08:38 TW: That’s fair. It’s like you say, maybe it’s the client’s great and the agency’s great, but there’s some asshole analytics consultant who’s trying to help out and that’s the one who’s wrecking everything. Maybe that’s… Then I’ll go into an even darker space.

08:52 JN: Well it could be that, but I wasn’t gonna bring it up.

[laughter]

08:57 MH: That actually brings up what is probably something that is not well understood by the community of analytics people, which is when we look at some of this data, it is not immediately apparent a lot of times what value it’s driving. And yet companies are obviously spending millions of billions of dollars on it. So, maybe that’s a great place to start, is just sort of where does that break down? Or how do companies sort of say, “Okay, well, yeah, we’re not gonna see a direct response ROI to this.” Click through and things like that is not something that’s really thought of as useful in determining display media. So how do companies think about it? Or how do they feel the value’s flowing through, if it’s not flowing through in response?

09:48 JN: You start off with the acknowledgment that our measurement tools are pretty flawed. You start with the fact that most people are using the, sort of the last touch methodology of saying whatever the publisher or the ad that was the last thing before the action, the success action, whether that’s conversion or download, whatever you’re using for an online… Sticking to online for a minute ’cause that’s easier, that’s not even the hardest. And then you realize that, “Well, that totally makes no sense. Why would you give all of the credit to the last touch when the consumer has likely seen many ads and clicked on some and searched and maybe seen affiliate ads and looked at emails and gone to the website?”

10:37 JN: So, we all know that last touch is flawed and yet 90% of the decisions in the industry today are made on last touch. But then, you go back another layer and go, “Okay, last touch being what it is, that’s based on cookies. And cookies are even more flawed because 40% of the time, cookies aren’t actually tracking what’s going on because there’s cookie rejectors, people who’s browsers reject third-party cookies, there’s people who flush their cookies.” So even on a last… If you accept the absurdity of a last touch model, it’s based on cookies, so by definition you have a flawed data set before you start. So, there are ways to address it. There are technologies that allow you to go where cookies don’t go. We have a product that does it, there are others. And there are…

11:30 TW: Is that device fingerprinting type stuff or is that…

11:33 JN: Yeah, yeah.

11:34 TW: Okay.

11:35 JN: We don’t like that term. We like to say digital signatures. But yes.

11:39 TW: Okay. [chuckle]

11:40 JN: Fundamentally speaking, yeah, that’s what it is. So, you get to a more stable marker for measurement but then you still have to then apply more rigorous analytical tools, like fractional attribution, to try to get a better distribution of credit for the various touches that lead up to whatever the success action is. But then for a lot of categories, it doesn’t happen online, there is no success action online because they’re going into a grocery store and buying their Oreos. [chuckle] It’s a mess.

12:13 TW: I feel like a lot of times that stuff, it is measurable, but it’s expensive to measure it, and it’s still imperfect and it’s often also really slow. If you’re doing a list study, one, you’re not gonna get the granularity of which placement and which creative, necessarily, and you’re also maybe not gonna get the results until months after the campaign, which really kinda reduces the usefulness of it. So given that, what is the… I feel like I’m always envious of the online retailers that, even if they’re using, I guess, cookies, that’s flawed, that’s just another, another depressing part. But getting view through data, getting… What?

12:58 MH: But there are flaws, we understand that flaw. [chuckle] That one’s okay, we can work with that one.

13:08 TW: I guess, I just feel like… Looking at the CPG world, which is where I’ve had the most experience with display, where potentially, because digital, even pre-digital it was, “We’ve gotta drive trial, let’s give away coupons and samples for our product.” The industry then shifts to saying, “We’re gonna measure that as success.” And there’s some logic there. If you get people to try your product, then they’re more likely to wanna buy it. But when the math doesn’t work out, when you work towards… Well, at least they have to go online to get a coupon, and I try to measure and I’ll give every possible credit I can to display. And the numbers still don’t add up, and yet we keep running display because of the idea that if we’re carpet bombing the right audience with the provocative picture associated with our brand, then we believe, and I don’t think it’s necessarily wrong, that we are actually making an imprint of the brand on their psyche. But trying to measure awareness is another one that’s in just a tough, tough space. And then you layer on all that click fraud, and you’re like, “Oh my God, what’s the point?”

[chuckle]

14:23 JN: Well, so that’s where you get to marketing mix modeling, right? You get out of trying to measure digital in isolation and get a read in the context of all your other online and offline, for that offline sale. And you’re right, there has been and continues to be this frustration. And you even hear it from the marketing mix guys saying, “Well, digital doesn’t read well in marketing mix models.”

[chuckle]

14:47 TW: Yep. Well, it’s very hard to control, it’s very hard to isolate digital by… Or it can be hard to isolate some aspects of digital by, like a GO, that you want to, that would help people.

14:57 JN: Yeah. I get really… And some of my friends still work at those companies, but the fact is, the marketing mix…

15:06 TW: They probably don’t listen to this podcast, so…

[chuckle]

15:09 JN: And therein lies the problem. Because there’s this divide between the digital guys and the offline guys, the marketing mix guys are in the offline world, even when they’re incorporating digital, they don’t understand it. I’ve just finished a tour of the big five marketing mix boutiques, explaining to them why you can’t use raw log files as the basis of representing digital in your marketing mix models. And they all started out looking at me like I had three heads, like, “Well, what else would you use?” And walked through all of the reasons why log file is bad data, that by its nature is so muddy that it actually distorts what’s going on, rather than illuminates it.

15:55 JN: So, the first thing you gotta do is integrate your… This is hysterical. Nobody integrates their verification service data into the log files. So, you know there’s ad fraud going on. You buy Integral Ad Science or Double Verify or somebody like that ’cause you know what’s going on, you want a read on it. But then, you don’t integrate it into the log files, you just send the raw log files with all the fraud in it, and the unviewed impressions and everything to your marketing mix provider. And you scratch your head going, “What? You paid for it.” [chuckle] But, that’s the state of things.

16:36 TW: I know I’ve been in places, I’ve had cases where the mix modeling guys, they’re off somewhere else and I get the request three chains removed which is, “Can you give me the total page views for our website for the last year?” ‘Cause that’s gonna be kind of the… That’s the attempt to model the organic contribution of a website, which is a destination, not a… And they’re like, “Oh, we just need impressions.” And so that’s the equivalent of an impression. And you can sorta see the horrendously flawed logic with a litany of uninformed people on that topic putting it together. Now I’m sitting on the other side saying, “I don’t understand how you do all your mix modeling, that seems like black magic to me, and you’re just taking the big buckets of spend and trying… ” So, it’s definitely two parties that are both, I think, very knowledgeable and have very… Reasonably sophisticated ways of thinking about things in their area, but crossing that boundary is really tough, or at least I’ve never seen it successfully done.

17:37 JN: Yeah. I don’t know if this is a forum to talk about various providers, but Marketing Evolution, I think, does a good job of bringing the two together with some innovative approaches. But for the most part, there’s a digital divide in the attribution versus marketing mix world, and those two just do not come together very well.

17:58 TW: See, this was the… I’m not sure we can go down the path, you were gonna throw market share partners or something under the bus. That was the controversial stuff we were looking for on this, when you’re like, “We can talk about specific people,” but you’re just gonna say nice things about specific.

18:12 JN: Yeah, I like John and Wes. They’re… I’m not gonna throw their company under the bus.

18:17 MH: No, that’s definitely the higher road and that’s cool.

18:20 TW: Yeah. [laughter]

18:22 MH: So it probably… I can just see the… We have a Slack channel in our measurement community where sometimes these things get discussed and Tim and I will pay a price if we don’t try to come at you with hard questions.

18:36 JN: Sure.

18:37 MH: So there was a study last year that I think was spearheaded or commissioned by, I don’t remember who commissioned it, but I think Ubiquity and another company ran it.

18:46 JN: Yeah, it was the ANA, the Association of National Advertisers that did it on behalf of a bunch of the leading advertisers in the industry.

18:54 MH: Right, and they looked at kind of all of the big trade shops, the big global agencies. And so they came back with some pretty… Well, they didn’t really release the really bad stuff, but it basically insinuated there are some real bad practices going on, a lot of things that needed to change. What’s been the fallout from that from your perspective? ‘Cause I don’t think…

19:15 TW: That study was actually reinforcing stuff that goes back two, three years, right?

19:19 JN: Absolutely.

19:20 TW: That they were… So that wasn’t like that was the first time that’s bubbled up, but nothing had changed in the past.

19:27 JN: Yeah, I think advertisers have taken more responsibility now. I think the trust… That was the last nail in the coffin of trust between advertisers and agencies and this mythical illusion that the agency is in fact an agent for the client. I think that that just pretty much broke that illusion, and now it’s caveat emptor for the advertisers is, “Make no assumption that the agency’s actually working on your behalf and scrutinize everything, read the fine print, and more and more set up your relationships with your technology partners directly, and do not go through the agency because it gives them a chance to hide things, mark it up and otherwise take a bite that you thought that they weren’t doing or weren’t allowed to do.” So I think… I see a significant trend to advertisers contracting directly with their tech partners and taking a lot more responsibility for the nature of the relationship with the agency.

20:34 MH: Oh, that’s fascinating. I think from the concept of sort of overall marketing technology though, that complexity is pretty challenging. Are companies up to it? Are they able to actually manage this stack? Or is it just more stuff that they’re gonna struggle with?

20:52 JN: Yeah, the answer is always “It depends.” I think the bigger advertisers who are spending big dollars have the resources, and have decided that this is mission critical to the marketing future and therefore they better acquire the skills and bring them in-house. And for the big advertisers, yeah, they can do it. I think when you get to the mid-tier it becomes a lot more challenging for advertisers, but I think the answer for a lot of those mid-tiers is to find the mid-tier agencies that are a little bit more intimate and can work with them in a more open basis that works for both. I think it gets really hard for the mid-tier to work with the mega agencies and it’s equally as hard for the mid-tier to bring the capabilities in-house.

21:42 TW: So I have a much, much smaller data set than you do for my kind of anecdotal evidence. I almost feel like the… And it’s the chain that’s been kind of outlined by many people that if you… Even in that study that said, “Look, the issue is nobody wants to stand up and say the emperor has no clothes.” And so in my very limited data set with at least one larger client, former client at this point, where they’re spending so much on media, offline and online, that when that bubbled up and it hit, they kind of looked at it and then it seemed like kind of immediately sort of stuck their heads in the sand because they weren’t equipped to say, “We’ve gotta scale up and build up the internal capabilities to really do this.” That’s hard work.

22:37 TW: We need to find people who are hard to find, that understand technically how the bits and bytes flow, and it’s wound up being sort of a blip in the… Kind of in their rear view mirror. Just today, I sort of saw from other contacts in that organization that they kind of, again, just shove their heads as far in the sand as they could, and it’s depressing. I guess it’s good to hear that the large organizations are… That some of them at least have said, “We have got to tackle this head on and we’ve got to get ourselves on a path that in six months or a year or two years, we sort of shifted who’s pulling what levers and who’s actually responsible for measurement.” That’s part of my reason of being depressed, is that I haven’t run into those organizations.

23:28 JN: Well, the clients that I work with today… Flashtalk is a big client company, so I mean, it’s Walmart and Verizon and AT&T and companies like that, and they have made the investment and have got folks who really understand what’s going on. They still use agencies, but now as clients they’re much more educated buyers and can hold the agencies more accountable than they could even two or three years ago. And there’s a arc to these things. At some point the knowledge becomes more broadly distributed and everybody is able to sort of master the basics of what, three years ago, was brand new tech. But ESPs, not that new anymore, and in a couple of years pretty much the knowledge will have diffused throughout the industry and we’ll be on to the next thing.

24:19 MH: So what is the next thing?

24:22 JN: Yeah, what’s cutting edge and now is dynamic creative, and people will say, “Oh well, dynamic creative is not new. Dynamic creative has been around since Dapper,” and for those of you who are listening, who were younger than us, Dapper was sold to Yahoo! About five years ago. [laughter] What’s happening now is something entirely different than that sort of Dynamic 1.0, which was simple geo-versioning. Now, what we’re able to do is take data from DMPs, from weather feeds, from social feeds and build out very, very complex messaging architectures and decision matrices about what message and what creative should be delivered on each impression to each individual. And that’s not necessarily, it’s related, but it’s different than the actual buying strategy. Deciding that you’re buying a particular person ’cause they’re your target audience doesn’t necessarily tell you what you should say to them, or what you should say to them on the fifth impression or the 10th impression or the 20th impression. So I think that’s the area of where we’re just at the beginning and is gonna be the hot thing in our industry for the next three to five years.

25:37 TW: That sounds a lot like the AB testing and multivariate testing and targeting for the website side of things, or even for marketing automation. Take your nurturing programs, which both in my… Apparently, just negative outlook on the entire industry. The challenge there is like you said, complex architectures, that now you actually have to have the staff with the talent to actually build out those architectures and build modular creative or things that can be kind of combined. Is that… Are you seeing companies be able to tackle that with eyes wide open? That it’s not… We have the technology that can serve a thousand different variations of this ad? Now, we actually have to have the thought and the planning into what the pieces and parts are that can kind of fit together?

26:33 JN: Yeah, I think the technology is ahead right now of a lot of advertisers’ ability to deploy it effectively, but that’s where the service companies come in, where we help our clients through that and help them. And if you’re a carpenter, you can have the very best tools in the world. If you’re sloppy in your work, the work’s gonna be sloppy. And if you’re a really talented craftsman, you’ll make it work with an old saw, right? And so I think that analogy still exists. Right now, there’s all kinds of tools, but the experience around the deployment of that is in its infancy. People will learn. They will get experience and the smart companies will hire experts to teach them and help them get up the learning curve.

27:19 JN: And when you get down to it, I think, talk to any marketer and in a way this is the dream that they’ve always wanted, is to be able to put the right message in front of each of their consumers. What marketer doesn’t want that? To be able to speak to their customers versus their prospects differently. To be able to speak to an engaged prospect versus a top of funnel prospect differently. To understand that you’re interested in safety, but I’m interested in handling, if it’s an auto client, and to be able to tailor the message on that basis. In concept, it’s not that hard.

[laughter]

27:55 TW: I’d say 100% of marketers want that and somewhere between 0.2% and 0.3% actually acknowledge the effort to actually pull that off successfully. The bane of analytics and operations everywhere.

28:09 JN: Well, and right now, a lot of that is being held up by the agencies who just can’t pull it off. The environment’s helped in that last fall there was something that’s become known in the industry as the ‘Flashpocalypse,’ which was Google turning off Flash as a default service in the browser, which effectively forced agencies to learn HTML5. [laughter] And with learning HTML5, it sort of forced the adoption of some of the next generation tools. Now, for our business, that was a big leap forward because until then agencies were able to get away with just building ads in Flash as they had been doing for the last seven or eight years and not adopt any of the new stuff. When that sort of got taken away from them and they had to acquire some new skills, it was actually a big leap forward, but they had to be prodded.

29:01 JN: But now, I think there’s some real momentum if you read a lot of the industry pundits, dynamic creative is on everybody’s lips now. It’s the next natural progression. We spent a bunch of years building DMPs and then learning how to get your data in order and get it into the DMP, and then learning how to use it for media buying, and now the next natural thing is “Well, how do I use it to get the right message in front of people?”

29:25 MH: So, two things. First, a comment, which is that’s gonna be great to see all these awesome dynamic ads, all of which lead to the homepage of the website. [laughter] I’m joking. Hopefully, companies, if they’re gonna go that far, realize that changing the ad means changing where they send people if they actually click through on it.

29:45 JN: Indeed.

29:46 MH: But the other thing is that basically, is taking this concept of, on the site side of things, which is where I primarily have worked in my career, that’s what we’ve been trying to build up in companies is this ability to create a message that resonates at the time of visiting of the site. And so in a sense you’re extending that now out past the visit, out into the advertising marketplace and saying, “Okay, let me try to continue to sell my product or start selling my product to you based on what might be actually potentially quite a bit of information I have about you or people like you.” That is gonna drive how advertising looks to you.

30:27 JN: Right.

30:29 MH: And then you get into, “Well, if they ever see it, that’ll be great,” which is another question. Anyways, I don’t know if that was a question as much as just a comment or just me trying to restate what I think I heard you say.

30:40 JN: Yeah. I mean, look, we all know what retargeting is. Your grandma will comment about how annoyed she is that the things she looked at at some website keep following her around the internet. [chuckle] So everybody knows what retargeting is, I think. And, in fact, that is enabled by dynamic ads in its most simple form, I looked at this item on the website, now I see ads for it. That’s not much beyond what Dapper was doing five years ago when Yahoo! Bought them, though, and I think as consumers we have the reaction of, that’s annoying. Particularly, when that ad continues to follow us after we bought the item or we’re out of market for it and it just keeps coming back and back and back. That’s just poor execution. That’s not a fault of the technology, that’s the lack of skill and vigilance in the people who are using it.

31:29 TW: Well, yeah. But I feel like as much as people complain about it, they actually are also acknowledging that they saw it and that, but because it was somewhat in their mind because they had visited… I had a very specific example where I bought a kayak at REI and I had gone to their site, rei.com, and then I had ultimately went in and bought in-store, and it persisted for quite a while. The fact that it was serving up the actual products that I had been looking at, I thought, “Okay, that’s effective ’cause it’s relevant.” When I look at a DMP that’s trying to get much, much more probabilistic, and based on all of the browsing behavior that they’ve got on me and they’re trying to put me in a certain audience, it seems like it’s gotten to where… All too often, the promise of the audience specificity, “Middle-aged, white guy, multiple kids living in Ohio who is angry most of the time, we should sell him things to calm him down.” They missed that he is angry most of the time and they actually have me as a female or something, I guess.

[chuckle]

32:48 TW: I’m skeptical of the DMPs, when I’ve gotten the glimpses, when you get those rare moments where you can say, “Let me actually really inspect that audience,” and all of the sudden you realize that, “No, this is all probabilistic and it’s all fuzzy.” It all makes sense that you’re trying to look at behavior and you’re trying to stitch this stuff together, but the reality is the specificity of this audience that you think you want, that you serve the super targeted ad to, can you really? Or does all of that foundational stuff around the messiness of cookies and multiple devices actually throw a big old wrench in that, that then we just turn around and ignore?

33:27 JN: Well, I think the first thing you have to do is separate, and you just did something that all too many people do, which is mix the DMP with the data. You actually have to think of them very, very separately. The DMP is the software that allows you to manage the data. The data itself comes from a myriad of data providers, some of which are good, some of which are transparent, some of which are lousy and are black box. So if you think about Oracle for example, they’ve made a hard split, there’s the Oracle Marketing Cloud and that’s where you buy your DMP. And then there’s the Oracle Data Cloud and that’s where you buy third-party data to put into your DMP and they are…

34:10 TW: Is BlueKai, is the… That’s the DMP?

34:15 MH: That’s the DMP. The software.

34:16 JN: The DMP. That’s right.

34:17 TW: Okay. That’s the first one. Okay.

34:19 MH: Yeah, Tim. Basic stuff, Tim.

34:21 JN: Well, no, no.

34:22 MH: I’m just kidding. I’m just kidding. This could really help.

34:26 JN: They have created so much confusion because before Oracle bought them, it was all together. You bought the DMP, and the DMP had the data subscription built in, and a lot of clients haven’t caught on to Oracle’s change of the model. I’ve literally, I’ve had a couple of clients recently calling me frustrated and angry that they couldn’t use third-party data to version their creative. That they are only able to use first-party data. And I called up Oracle and said, “This is crazy, what’s the… ” and they answered, “Your client doesn’t actually have a license for the third party data, they didn’t buy it.” But the client didn’t know, they thought they had. [laughter] So it is confusing.

35:09 MH: See, these kinds of things never happen in the digital analytics space, so that’s why it is so funny, it’s ’cause…

35:19 TW: But even with that distinction, it’s like I can feed my absolute first-party, my site browsing behavior into my DMP. And I know that this was somebody who was looking at this particular product, and that’s super high quality. I feel like those other subscriptions get thought of as, “Oh, they have that same level of specificity.” And maybe that’s it, it’s just you get… If you want a ton of different… If you want a larger set of audiences, that’s inherently gonna be a noisier set of audiences. If you want a very clean set of audiences, it’s probably gonna be more expensive and a smaller set, I would assume, that’s just an assumption.

36:00 JN: Yeah. If you’re an experienced programmatic media buyer, you have learned from experience the good data from the bad data, because you see what performs and what doesn’t perform. For example, the auto data out there is pretty good. If you think about it, most of the auto data that you’re buying online is coming ultimately from edmunds.com, Kelly Blue Book and Polk. And they are really credible sources of buyer interest behaviour.

36:33 JN: So auto data in general is pretty good. If you’re looking for information about people who like basket weaving though, you’re maybe getting something that’s modeled and… ‘Cause, where do you get that? [chuckle] It’s always buyer beware ’cause there’s, data providers are in business like everybody else to sell their wares. And you as a buyer, you gotta ask the questions of, “Where does this data come from? Under what privacy policy was it collected? What’s the quality control? What’s the freshness date on it?” This is one of the things I just loved to do to the data providers is, “When do you expire your cookies?” They go, “Huh, what?” Yeah. “How long… ”

37:20 TW: If they expire, our audience will get smaller.

37:22 JN: Where do you throw it out and say, “That’s not relevant anymore.” For most of them the answer is never. [chuckle]

37:28 MH: Yeah. Five or six years.

37:31 TW: Does that data get also muddied up by all these bots? Are the bots that are causing all sorts of the click fraud issues, does that also muck up the audience data?

37:44 JN: Probably not so much.

37:46 TW: Not so much.

37:47 JN: ‘Cause most… At least the quality stuff is coming from publishers and they’ve got logins. That’s probably not that much of an issue for the data providers.

38:00 TW: Okay.

38:01 MH: So, yeah. As always, this is just a conversation I feel like we’ve… This is part one of our ongoing series. Now, I don’t know, but we do have to wrap. And John, thank you so much for giving some of your perspective on the industry, but one of the things we like to do on the show is called the last call. Where we go around the horn and just see what’s interesting. What have you seen recently that you thought was noteworthy? John, why don’t we start with you?

38:31 JN: I think there’s the ongoing efforts by Google to close off more and more of their inventory to outside eyes. I’m sure you’ve seen the latest announcements regarding YouTube not taking, or reducing the third party pixels that they’re gonna allow and so forth. It’s just another step in this sort of duopoly of Google and Facebook trying to be more and more insular, and it’s very, very anti-advertiser and very anti-competitive. We’ll see if Mr. Pritchard’s call for more openness from the publishers and so forth falls on deaf ears or whether Facebook and Google will listen to Proctor and other advertisers like them.

39:23 TW: Does the government regulation wind up coming in from a… Not regulation necessarily, but is there lobbying, the IAB or others that are able to exert pressure on that or is that…

39:34 JN: I don’t think so, because they are always able to hide behind their privacy policy and say, “Well, the reason they have to do this is to protect their user data.” Advertisers wanna… The power of their data and if they don’t restrict these pixels and stuff, people can essentially infer identity, and it’s a threat to their privacy of their customers. And at some level, there is a grain of truth to that, but it allows them to hide behind it as well.

40:02 TW: Yeah. And those two companies who are constantly in the headlines over in the EU for privacy concerns.

40:09 JN: Right.

40:10 MH: So, “Hello pot, I’m kettle.”

[laughter]

40:16 JN: Well, if you start talking about the European landscape, there’s a whole other depth of absurdity as the new data protection laws that they announced are gonna be putting into place, essentially put even more power in Facebook and Google’s corner because they have first-party login data. And by eliminating or even more strongly restricting third-party cookies, essentially, it’s taking any options away from the advertisers and handing the market to Google and Facebook, even as they’re saying, “Naughty, naughty, you guys are monopolistic.” It makes no sense.

40:52 MH: Yeah, there’s definitely some structural problems with how that’s all happening. And certainly, even on the digital analytics side, we’ve all sort of seen some fallout from that. What about you Tim, what’s your last call?

41:05 TW: This is just a little bit of a video candy, I think, but when I was in at Superweek, one of the presenters was Tahir Fayyaz who showed a number of interesting things, but he referenced a video, the Nat & Lo, A Journey To The Bottom Of The Internet, which I went and watched and it’s kind of fascinating. It’s sort of looking at the ins and outs of the modern equivalent of a Trans-Atlantic telegraph cable, and their whole experience of diving into that. It’s just kind of a fun and informative little bit of trivia video to watch.

41:44 MH: Nice. Well, my last call was ripped straight from the headlines. As you know, recently the Adobe Marketing Summit was held in Las Vegas, and it’s very exciting to see the launch of the next version of their Dynamic Tag Management Tool. So, yeah. We actually had a lot to say about that. It was a, “Check out our… ” Check out their stuff as you’re looking to upgrade. And it goes into beta soon, it will be very interesting to see where that goes. Anyway, so, no. It’s exciting. Alright. If you’ve been listening and you’ve got some questions or you’ve got some comments, we would love to hear from you. Th best way to do that is through our Facebook page, or on the Measure Slack, or on Twitter. John, is there a way that people can reach you on Twitter, or you stay away from that?

42:33 JN: I stay away from social media. So, email is the best way to reach me at john.nardone@flashtalking.com.

42:42 MH: There you have it.

42:43 TW: And how can they run it? Can they run a display ad to show up in front of your favorite news site that’s targeted directly at you?

42:49 JN: You know…

42:51 TW: If they can, they should also send you their resume? Is that good?

42:55 JN: I know we’re at the end here, but you guys will appreciate, the first targeted ad targeted to an individual on the internet was targeted to me by Doubleclick on May the 12th, 1997, when my twins were born. And Doubleclick sent me an ad saying, “Congratulations, John, on the birth of your twins.” It was targeted to the Modem Media domain, so all of my coworkers at Modem Media saw it. That was the first time they had ever targeted an ad to a specific domain. So there you have it.

43:31 MH: Wow.

43:31 TW: That’s cool.

43:33 MH: Yeah. Doubleclick doesn’t go in for that kind of personal touch anymore.

43:37 JN: It was a different administration in those days.

[chuckle]

43:40 MH: That’s right. I’m sure. No, that’s a cool story. Thanks for sharing that. Alright. Yeah, we’d love to hear from you. John Nardone, thank you so much for coming on the show, it is a real pleasure to speak with you and get your perspective. Tim, it’s always a pleasure. So for all of us here and to you out there, keep analyzing.

[music]

44:06 Announcer: Thanks for listening, and don’t forget to join the conversation on Facebook, Twitter, or Measure Slack group. We welcome your comments and questions. Visit us on the web at analyticshour.io, Facebook.com/analyticshour, or @analyticshour on Twitter.

[music]

[background conversation]

Leave a Reply

Your email address will not be published. Required fields are marked *

The Digital Analytics Power Hour © 2017