← AdExchanger
AdExchanger
Breaking Away From Low-Quality Data
00:59:26
Scott McKinley shares his journey from professional cyclist and captain of the 1988 US Olympic Road Cycling Team to CEO and founder of data validation provider Truthset. The road was more linear than one might think.
Loading summary
Transcript54 lines
- [00:00]
Scott McKinlay
Foreign.
- [00:13]
Alison Schiff
Welcome to Ad Exchanger Talks, the podcast devoted to examining the issues and trends in advertising and marketing technology that matter most to you. This episode is sponsored by the trade desk Edge Academy, the online learning platform designed to help marketers stay ahead in the evolving world of digital advertising. Check out their certifications, expanded course curriculum, and more, all led by the people who are shaping the world of ad tech. Enroll today for free at edgeacademy.thetradedesk.com this is Alison Schiff and you're listening to Ad Exchanger Talks, the last episode of 2024. Thanks to everyone who listened this year, including those of you who listened at 1.5x speed. I know how busy you are, and I'm very grateful for your time. Every week we do our best to bring you quality conversations with quality people. This week, my guest is a quality person, Scott McKinlay, CEO and founder of Truthset. But our topic is the lack of data quality on the open web, which Scott says is no better than a coin toss. It's bad out there. Demographic targeting data is in some cases worse than useless. It doesn't have to be that way, though. After we finish shaking our damn heads, we talk about solutions and how to fix the problem. You know the line this is why we can't have nice things. Let's make it no longer apply to data quality in 2025. But before we get started, if you're not listening to our other podcast, the Big Story, you're missing out. It's a weekly roundtable discussion of the biggest news of the week with a rotating cast of members of the Ad Exchanger editorial team. And we're an amusing crew. So let us entertain you and inform you during your morning run or on your commute. And while you're at it, please also save the date for next year's CTV Connect March 12th and 13th in New York City, Ad Exchanger is joining forces again with Synopsys, Ad Monsters and Chief Marketer to host a can't miss summit on all the key issues and opportunities in Connected TV. Learn more and register@ctvconnect.com hey Scott, welcome to the podcast.
- [02:32]
Scott McKinlay
Hi Alison, thanks for having me.
- [02:34]
Alison Schiff
So I have a question for you. Was Breaking Away your favorite movie as a kid? And I asked this not to steal your thunder. There'll be lots of time to tell to tell a story. But you were a professional cyclist. You were Captain of the 1988 U.S. olympic road cycling team in Seoul, Korea, if I am correct. And you were pro in Europe and the US with a whole bunch of different cycling teams. So. Yeah, tell me a story.
- [03:01]
Scott McKinlay
Well, you're making me blush a little bit. I'm glad you asked if it was Breaking Away as my favorite movie and not American Flyers, because in the cycling community, American Flyers, if you remember that one with Kevin Costner, was. Was broadly ridiculed for its lack of integrity in representing the sport. However, Breaking Away was spot on. And really, it really was a. I mean, you could watch that movie. And that was me as a freshman in high school, you know, excited about, you know, learning how to read French so I could read the reports from races in, in Europe while my, my, my peers were, you know, playing football and, and trying to get girls and stuff. So it was a very different approach to sports and life for me. And yeah, I was, I was, I was in the Olympics. It was feels like several lifetimes ago. Yeah. And it actually has relevance to what I'm doing now. I mean, sports is, if you take out the cheaters, the ultimate test of truth. It's simply, especially in cycling, you start a race, you race 120 or 130 miles, and the first one to the other side across the line is the winner. It's very pure, it's very accountable, it's very individual, and it's great in that regard. And in fact, I. Why I quit, it's because the cheaters arrived and the drugs arrived and I didn't want to participate in that. And I've always been looking for ways to be truthful and tell the truth, and I found a way to do that in business.
- [04:26]
Alison Schiff
Well, you've just handed me on a silver platter. The perfect segue. But I'm not going to take it because I'm just curious, was cycling not a good way to get girls in high school?
- [04:37]
Scott McKinlay
No. No. This is 1983. 84. Oh, my God. I'm dating myself. But, you know, I would, I would go to school in a Lacoque Sportif like cycling jersey thinking that was cool. And it was cool in my own head, and that was fine with me.
- [04:52]
Alison Schiff
It's the most important place to be cool is in your own, your own head. Um, so to go through a little bit more of your career trajectory, it's 1996, you retire from being a professional cyclist and you started a job at Cox, working on advertising and, and monetization for some of their TV and radio stat. You co founded a company in 99. Sweet, smart digital media measurement for brands. Spent a whole bunch of years at Nielsen. And we'll get Back to that and then some other stuff. Da, da, da. Fast forward. You found Truthset in 2019, a company that validates the accuracy of consumer data. But we'll get back to that, too. Tell me a little bit more, though, about how you transitioned from professional road cycling to the consumer data a business. Because I like your answer, but I feel like it's a story you've told before, that it felt a little pet. How did you actually find yourself in ad tech?
- [05:55]
Scott McKinlay
It's actually true. So I retired and I came back from Europe in like 92, when again, when the. When the real drugs started to hit the cycling world. Raced for a few years in the States and then decided I needed to get an actual job. Remember, I didn't go to school. I didn't do that track. I was on the athlete track. I was in the Olympic program since the time I was 15 and did the Olympics when I was 19. And so I was like, well, what am I going to do now? And this is 1995, 96. I'm beginning to figure out what I got to do next. And the Internet had sort of arrived, right? This is the years of AOL and Prodigy and the. Well, I don't know if you go back that far, but news groups and it was exploding. It's funny, because I started. I taught myself how to code HTML and make websites for small companies. And it's funny, I had a partner, and one time I looked at him in 1997 and said, I think we missed the boat. I think we're too late for this. Which was hilarious in retrospect. But that's. That's how it started. I started coding websites and I immediately got a job with ktvu, a Cox station in San Francisco, running what was. Essentially, I didn't start running it. I was writing stories, actually, about where to ride your bike in the Bay Area. But the website was like a city search kind of website, like what to do in San Francisco. Cox had 27 of these across their major markets in the country, where they had a dominant traditional media position, a radio station, a newspaper, or a television station, they would leverage the assets and the sort of promotional power of the traditional media network to support the new website. And that's what we did. And we were trying to sell advertising, and it just struck me how bad we were at it. I mean, the sites, we didn't know who was coming, so we didn't know how. What kind of brands might be appropriate for those kinds of audiences. So I suggested in my naivete, having not learned how to not ask questions because I didn't do the traditional path. I just said simply, why the heck are we not trying to figure out who our users are so that we can do a better job putting the right ads in front of the right people in this amazing new medium? And it evidently was a good idea. And I was given budget to make essentially survey applications for the whole, the whole network of 27 sites. And we got good at building distributed apps for the website. The website itself wasn't that successful, but what was successful was the development of these tools to help websites and owners of websites understand who their users were. So I created a company that just did that in late 99. We got our friends and family paychecks or our investment, rather four of us in the company in March, one week before the headline hit the Wall Street Journal, the bubble has burst. Yeah, so terrible timing. And I went back to the investors, I said, you guys want your money back? I don't know how we're going to navigate this. And they said, no, go for it. And so we made a bunch of survey applications that would help website and various tools to help websites better understand their customers and users and do a better job monetizing those audiences. I raised institutional funding for that company in 2005. We sold it in 2008 to Symphony, the holding company. And then I went to work for Nielsen after that. So basically, to answer the question directly, it was a necessity that I find something that I could do that didn't require the traditional path that was the Internet. I could learn how to code on my own. I could start a career path there. And it took me into the world of consumer research and data. And then at Nielsen really understood how that worked at working at the largest market research company in the world. And that was that. And then after that, by that time, I'm 18 years into this career and I learned that the one thing that I felt was an endemic problem in the whole advertising, digital advertising ecosystem was the quality of data. And I decided basically, I'm either going to leave the industry because I was so tired of the snake oil salesman and the, you know, the obfuscation and bs, or I was going to create a company that tried to clean up the mess a little bit and give everybody a better shot at using data to predict who someone might be on the other end of a device and give a better ad experience, support publishers with higher CPMs, and give give advertisers what the Internet always promised, which was, you know, precise, you know, highly profitable. Advertising investment.
- [10:23]
Alison Schiff
It's amazing that today we're still asking that question, who are our customers and is the data that we have about them true or not a question that you were asking yourself in your first job?
- [10:38]
Scott McKinlay
Yeah, I mean, look, I have a little theory that I'm going to bore you with. I think that what's happening here is if you think about the golden age of advertising before there was person level addressability, it was all about the media which was television, print and radio, broad, broad targeted. You can't precisely target these media. And it was all about the creative. Right? This is the Mad Men, Don Draper days and the big agencies and all that. All the power shifted towards the creative and to the media with no thought really at all about audience. Audience was just age and gender, if you even cared and that was it. Right. And I think with digital really that whole ecosystem has lost the plot. I think that the value now in the three legged stool that makes advertising work, which is the audience, the medium that the audience is on and the creative you put in front of them, I think the power has shifted from what used to be like whatever, 50% medium and 40% creative and 10% audience to more like 50 or 60% is the value of the audience and how well you can target and creative and media is not that important. I mean the statement of maybe five, six years ago that a user is less interested or less receptive to advertising while watching a TikTok scrolling session versus a full engagement Paramount's what is that Yellowstone is just false. I mean the data shows clearly, the market caps show clearly that it's not the medium, it's not the message, it's the audience. It's the audience. So we've forgotten like the fact that audience data only gets around 20% of the dollar and the rest goes to media and creative I think is a huge, huge miss that the tech media giants have, have accidentally backed into and that the traditional media networks and properties which still drive a lot of the website creation and consumption, everything else, I think they've just missed the plot. So it's a long way of answering your question around why are we still struggling to understand who the audience is? Well frankly the big tech media giants aren't struggling. They have authenticated identity. They've earned the right to get email addresses and build lasting profiles and attach really, really accurate third party or first party behavioral data to those profiles and then deliver great ad experiences to those users. While the rest of the open Internet and the laggard television networks are still playing A game that is. It's just woefully, woefully out of date. And I think that's what we're seeing here. When, you know, you look at the walled gardens walking away with 80% of the money over the last 15 years in digital, why is this not going to happen in television as well? And I think the punchline here is we've got to get the audience right. You know, it's more important to be able to predict who's watching and what kind of message might be relevant than to focus on the creative or the media. At this point, I think the power has shifted.
- [13:53]
Alison Schiff
Facebook. I haven't used Facebook in years and years, but when I signed up, I told Facebook what my birthday is. And every year, the only activity on my wall, when people I knew from high school and past jobs and when I lived abroad, like almost 20 years ago, they all wish me happy birthday. But Facebook still knows that about me and still has my email address. And those are incredibly strong signals, even though I don't engage with it at all.
- [14:21]
Scott McKinlay
Yep, absolutely. I try not to be Chicken Little. You know, we're sitting on this data collective with over 20 of the largest data providers. I see how good everybody's data is. I see what happens when you move data from place to place, and I see how, how quickly it falls apart from what might have been good data at source, all the way to the endpoint where it's really a coin flip again. And it's hard for me not to just shout, you know, hey, everybody, authenticate your users or die. Like, that is the relatively extreme position. But I really think it's an existential crisis that the open Internet and traditional media is in. If we don't. If they can't figure out how to authenticate their users, they're going to end up just licensing their content to people who can monetize it better, which are these walled gardens. And then they're going to go out of business. I think that's what's going to happen. I think we can stop it if we, if we work together to, to solve the problem.
- [15:11]
Alison Schiff
And some of the. The stats are kind of. They're gobsmacking in terms of the lack of data quality out there. You guys raised your Series a in October. $5 million. Congratulations. And I was reading an overview of the deck that you gave to investors, and it references some research you've done into data quality, or really, more accurately, the lack of data quality. And something like, what 60% of consumer data is wrong and if that's the case, like, what are we even doing here, right? Like data driven advertising, precision marketing, personalization, all of that, that feels like further down the line. I mean, the emperor is kind of nude, right?
- [15:58]
Scott McKinlay
Yeah, I mean, it's, it's bad. But I think the mistake is thinking that that audience data doesn't work. It only doesn't work when it's wrong. So, you know, you don't throw the baby out with the bathwater. You don't rewind all the way back to mass market. You know, advertising, that's, that world is behind us. But you do pay attention. You know, you stop covering your eyes and holding your nose and just buying off the shelf data where you don't really know where it came from. You don't know how heavily it's been modeled. You don't know how many times it's been matched through sketchy, obfuscated IDR practices of matching one data set to another. And that is what has to stop. I think people have to realize if they want to benefit from the efficiencies of digital advertising and precise targeting, which can, by the way, benefit everybody. Everybody wins when the data is better. The consumer gets a much better experience, the publisher gets higher CPM yields, it can monetize their audiences more efficiently, and brands get better. Pure ROI or roas. Whatever your metric is, everybody is better. So I think the answer is like, pay attention, pay attention to the numbers we're putting out there. We're an independent company that has no horse in this race. We support everybody. We have measurement companies in our client portfolio. We have very, very large television networks. We have brands at some of the biggest brands in the world. We want to benefit everybody by shining a light on this stuff and not saying, hey, it's so bad, you shouldn't even bother using it, but rather say, hey, you know, now we have a way to independently measure how much you can trust this identity. Is it or is it not a Hispanic New mom and should receive a Pampers or Huggies ad in their language. Like, that's the amount of precision that we can bring if we finally pay attention to the value and the, and the accuracy of this audience data. And again, if you, if you're on the traditional side thinking like, I don't really trust it, it doesn't really work. There's evidence that it works in the market caps of the big tech media companies who do just that. They get identity, they maintain an ongoing relationship with every single user on the Meta platform or Google or Netflix or Amazon, and That's the bar against which media properties and brands should be measuring themselves against.
- [18:11]
Alison Schiff
Well, in the spirit of shining a light and maybe scaring people straight, there are a few other stats from your analysis of basic demographics across public data marketplaces that I just want to mention because they're wild. So the average accuracy of an age segment across 132 providers is 32%, meaning that 70% roughly of age related data is wrong. And the average accuracy within a gender segment is 61% across 75 providers. So gender data is basically not better than a coin toss at this point. And then this is just like a shake my head 1. So the number of unique US based people based IDs in an age gender segment that you analyzed was greater than 1 billion, which means it included more people than actually exist in the entire US Nearly three times over. So I mean, I don't actually have a question. It's just, huh, that's kind of wild.
- [19:18]
Scott McKinlay
But you have to, you have to unpack this. So what you want to get into now is like, why is it like that and why is it staying like that? Like what are the incentives that, that, that, that keep this mess from being cleaned up when stats like this come out? And everybody, nobody challenges these stats, by the way. Like, and we're happy to show methodology and show how we did it. And we do this all the time. We're very, very completely transparent. So nobody even questions it. So look, why is it allowed? And why are companies still rolling up to a data marketplace and just buying off the shelf segments without any curiosity about how accurate they might be? And I think it's really, there's a bunch of things, right? One is just the incentives around scale over anything else. And it's kind of forgivable because there was no way to know how accurate segment A was over segment B. Like, which one should I pick? Which data provider should I pick? There was no, there was no two set. There was no independent way to say, okay, well I can get a reading on like a quality rating from J.D. power or something. Or you know, like at least you can get an independent understanding, objective understanding of how, you know, one source might compare to another. So that exists now. So that excuse is no longer. And I think what's left is just this addiction to scale and match rate at the expense of precision. You know, And I think that again, here's where the walled gardens have ignored that entirely, right? They don't deal with, they don't even like mass reach campaigns. They like precision that delivers results because you put a dog coat ad in front of a dog owner. Right, I mentioned that because we're doing a case study right now with a company that sells those. But I think it's number one is the incentives. And then the other thing is people don't. Very few people really fully understand how the data ecosystem works and what happens when you take what might be great data at source and move it through the ecosystem to the point where you can try to target devices against that data, for example. So if you have like a. I won't name any single data provider in our data collective, but there are many who have 88, 90% accurate for some very, very valuable segments. And they even come to us saying, hey, you got to help us diagnosis. Because when we load this data up into the ecosystem, it goes through an onboarding process, maybe it goes into a householding process, maybe it goes over to a DSP where they apply across device graph to it and then it gets expanded and they're like, by our own tests, we see our 90% accuracy at source, which Truthset is validating, turn into 21, 22% accuracy by the time it hits activation, which is trying to predict, is that actually a Hispanic mom behind the screen or is it something else? And so the degradation of data is actually a bigger problem than the accuracy of the data itself. And the degradation comes from all these absolutely ridiculous hops, skips and jumps as we try to match one data set to another. Now we're trying to match, you know, email addresses to cookies or maids or IP addresses. I mean, it basically, I'll tell you this. Every match process Generally, generally every ID sync or match process, you're going to lose 40 to 60% of your records and you're going to introduce 40 to 60% error. And if you do that twice, your potential for compound error is greater than 100%. So that's the simple way of looking at it. So what needs to get fixed and diagnosed, diagnosed and fixed is, is the. It's basically a form of supply chain optimization focused on data. What happens when we match your email address, Alison, to a cookie on a device or onto a maid or a tv IP address or a ctv? How much can we trust that? How much error is there? And are we degrading it so much that it's not even worth doing targeting in the first place, which is actually often the case?
- [23:10]
Alison Schiff
All right, well, we're going to take a quick break and when we're back, I'd love your unvarnished point of view about id Bridging, because I feel like you'll have something to say about that. So stick with us. I'm Alison Schiff, managing editor of Ad Exchanger. And today I have Stephanie Patteric with me, GM of Editorial and Editor in chief at the trade desk, where she leads the content strategy and execution. And she also has more than 20 years of experience, experience in media and journalism. Stephanie, welcome.
- [23:55]
Stephanie Patteric
Allison. I admire your work. So happy to be here.
- [23:58]
Alison Schiff
Thank you very much. So what shifts or forces are driving the evolution of the ad tech industry? This is such a big question.
- [24:09]
Stephanie Patteric
It is a big one. And I think we're in a remarkable moment in time because we're really witnessing the maturation of the ad tech industry. And by that I mean, you know, just in my career as a, as an advertising journalist, I've seen programmatic go from this niche part of the ecosystem that transacted primarily in remnant inventory to a vital player that's supporting premium content. And I think that three forces are sparking this. It comes down to consumers, regulators, and innovators. And first, I think it's important to understand that more people are spending time on the open Internet than in walled gardens. So they're spending more time streaming shows, listening to podcasts, reading news. And second, the Department of Justice has really held Google's feet to the fire this year, and that has exposed the danger of black boxes. So everyone in the supply chain is being scrutinized and there's a pressure to add value, not just extract it. And all of this really represents a massive opportunity for innovation and for ad tech to keep evolving with quality content, supply path, efficiency, and really transparency for all parties at the fore.
- [25:22]
Alison Schiff
What is one piece of advice that you'd give to someone looking to take their career to the next level?
- [25:30]
Stephanie Patteric
I love this question so much. Get outside of your own expertise to understand the big picture. I think one of the biggest mistakes that people make in their careers is to silo themselves within one area of expertise. And I want to be clear, specialization is valuable, but the more that you can get curious about what your stakeholders know, and the deeper you understand how the entire media ecosystem works, I think the more brilliant you'll be at what you do. So don't be afraid to play the role of a reporter and ask the subject matter experts in your organization what they know. I think you'll be surprised at how generous people are with their knowledge.
- [26:08]
Alison Schiff
Advice for your career, advice for life. And how can the trade Desk Edge Academy help a marketer with their career?
- [26:17]
Stephanie Patteric
We talk about transparency a lot in our industry, and one place where it's vital is education. At Edge, we have a strong belief that the more everybody understands programmatic, the healthier the whole ecosystem becomes. And we really believe that this knowledge is an advantage for anybody who's looking to grow their career or their business. And this goes for the cmo, the planner, and the trader. Edge has served the advertising community for 11 years, and it's evolving to stay ahead of the curve. So we've got new classes on timely topics like Omnichannel Advertising, Connected tv, as well as foundational certifications like Trading Essentials and Marketing Foundations. So it's just a fun challenge to meet learners where they're at and help them navigate this space that changes, as you well know, literally every day.
- [27:10]
Alison Schiff
Stephanie, thank you for the insights and for the advice.
- [27:13]
Stephanie Patteric
It's been so fun to be here. Thanks for having me.
- [27:31]
Alison Schiff
All right, we're back and ID bridging. So I think everyone who listens to this podcast knows what ID bridging is. It's a way for publishers to do identity matching without relying on third party cookies. But it's also not the most transparent thing in the world because you're making a lot of assumptions about who a user is. And the IAB, they have added ID provenance to the latest OpenRTB spec. So there is that. And there's more attention on bringing transparency to ID bridging. But yeah, what's your take on it? I mean, do you hate it? I'm assuming you hate it, but maybe I'm wrong.
- [28:11]
Scott McKinlay
Well, look, first of all, I respect every publisher's attempt to try to make their data better and try to understand users and try to deliver a better experience. I mean, everybody's just trying to stay in business and make money and be profitable and grow. So you can understand the incentives for doing this. However, how can we not respect the oncoming regulation around privacy? How can we not pay attention to the problems that have that have arisen from violating people's privacy? We can't do that. We shouldn't be trying to do things that are blatantly prohibited. Even if federal regulation hasn't hit yet, why would we not want to future proof ourselves at least a little bit by not wandering into the third rail of hey, here's an IP address. Can I send this over to somebody who maybe kind of sort of can attach it to an email address? And now I can say I've got first party data like that is the most egregious abuse of the technology and data That I, that I can imagine, followed closely by what's still called fingerprinting. I mean people hate to use it, but that's what it is, where they're using as many indicators of who this person's device is as they possibly can to create a lasting ID that they can use to bypass asking the user for the right to do that basically. So even if you do it formally and above board, how challenging it is to match any two ID spaces together. We did a study last year that was sponsored by SIM and a bunch of publishers that showed that the average accuracy of householding data across some of the largest providers of that Data was around 50% correct. Which means they got it right half the time and wrong half the time. Right. So those are three ways of trying to get, trying to get a better sense of who the customer or the user might be without explicitly asking them. And here's why I once again get on my soapbox and say why don't we just explicitly ask them? Because you know, who doesn't have these problems of wandering into privacy territory or trying to rely on opaque ID spaces and keys. It may or may not be considered pii. You know, doesn't worry about that. The companies that have the quality content that deserves a question. Would you provide your email address? And so I can give you personalized experiences and build you a better advertising experience. So you know, I just can't help but say, you know, if we measure ourselves against the history of completely untargeted mass media, which is the pre Internet, we're doing pretty good. We're getting a little bit better all the time. Maybe we're half right, maybe we're half wrong. But isn't that better than mass reach targeting? Well, sure it is, but you shouldn't be measuring yourselves, everybody against where you were. You should be measuring yourself against the people who are running away with the advertising dollar and delivering value and efficiency and performance to advertisers. And those are the Walt Gardens who don't monkey around with any of this stuff. They just simply ask their users to authenticate.
- [31:23]
Alison Schiff
Not that they don't have their own privacy problems, but I do take your point. So before you were saying that, I guess the game of broken telephone is a big part of the issue. But putting that aside, I mean, are there any certain types of marketing data that are more prone to be inaccurate than others or is it more that broken telephone game?
- [31:50]
Scott McKinlay
I mean the, it's kind of one and the same. I mean the biggest problem isn't actually the data itself. It's how you move data through the ecosystem from one point to another. You know, it's like we built these enormous factories to harvest, you know, billions and billions of signals and we're trying to translate them into a person. Basically, that's what we're doing. You know, it's like it's noise until you can hook it to a human being who's got, you know, ears and eyeballs and wallets and kids and needs. Right? That's the whole point of marketing. So we take this, this massive exhaust of digital noise, we squeeze it through these incredibly complex multibillion dollar factories that venture capital has given us over the last 20 years, and you try to squeeze it down so it attaches to some signal that represents a human who might have those needs and might have attention and might tune in and watch something and actually take an action off an ad. So, you know, the problem again is the process of taking all that noise and squeezing it down to represent a person. And there's a lot of pretending going on in that supply chain. There's a lot of guessing. There's a lot of what we love to call probabilistic modeling. Right where you are taking.
- [33:03]
Alison Schiff
I love a euphemism.
- [33:04]
Scott McKinlay
Yeah, totally. It's basically, it's euphemism for pure guessing. Right. With an incentive to maximize scale at any cost. Once you have those two variables, like, how can you possibly trust what comes out the other end?
- [33:16]
Alison Schiff
Is data quality in the eye of the beholder, though, would a brand maybe be less particular than another brand about what quality even means?
- [33:26]
Scott McKinlay
Yeah, that's this. I love this question. So there's two. You asked two different questions. The first question is, is data quality in the eye of the beholder? The answer is no. I'm objectively male, I'm objectively married, I'm objective dad. Like, I'm. All those things are objective truths about me. And we shouldn't be arguing from vendor to vendor and provider to provider over who got it, right. Like, we should just level up and collaborate and do the best job we can while respecting privacy, of understanding who these consumers are so that we can serve them better and help everybody serve them better. So the reason why we exist is to create an objective truth as close as we can get to these attributes about people and use them in a responsible way to drive better performance for everybody in the ecosystem. On the question of does one brand care more or less? Absolutely. So that's why when we designed our system, we didn't say data is good or bad. We say, okay, how likely is this data to be inaccurate or accurate on a spectrum? Because you just don't know for sure. I mean, we may be in an AI right now and you may not be a real person, I don't know for sure, but there's a good chance, et cetera. So we create probability, I hate to use that word, but likelihood scores for identities and attributes that allow a brand or two brands to decide, you know, do they want super, super, super accurate AAA rated data because they're doing a model, they're creating a seed file, they're doing customer analytics where scale is not important, or do they want, you know, to relax the accuracy a little bit in order to get more scale? As long as it's still human and there's still a chance it's still in my, in my target. So we actually allow every one of our customers, whether it's publishers or brands, to choose how to draw that line between accuracy and scale and settle on what's best for them and their business case.
- [35:17]
Alison Schiff
So do you know the nutrition labels, of course, that the IIB introduced back in 2019? They set a data transparency standard for minimum disclosures for audience data providers. So then a buyer has some idea what they're buying. So like, who provided this segment, how it was constructed, the audience description, where the data was sourced, but the purpose of these labels, and I don't know how much people actually use them, maybe, you know, but it's to create like a baseline for transparency about data collection, processing and modeling. Those are all things that inform data quality. But explicitly on its website, where they talk about the nutrition labels, the IAB says the standards are not intended to be a qualitative grade about performance or efficacy or a grade for quality and accuracy.
- [36:13]
Scott McKinlay
Right.
- [36:13]
Alison Schiff
So a few questions pop to mind based on that. Like one, is there any arbiter for what counts as quality? And I mean, should there be actually? Because like we were just saying, there is maybe not a question as to, you know, whether you are married and a dad. And those are very specific points. But I mean, if certain data is perhaps not labeled as quality, but a brand would be interested in buying it, then it's on them.
- [36:46]
Scott McKinlay
Yeah, so what part of the iab? I was in some of those conversations early on and it was funny because the origin of the label was to explain quality is to give a buyer a chance at sorting out one provider from another and understand which, which one they could trust more for their practices. And they decided after some, after some scoping it was impossible and so they backed off the specific to be everything but quality, right? These are the things that the data label. I'm not saying it's useless. I'm just saying that things that IT records and stamps are sort of rough proxies that may or may not correlate to the actual quality of the data, which may or may not correlate to actual performance or efficacy. So the shame was we weren't up and running enough at the time when this was all set in motion. And so we sort of didn't have a. A card to play. But that's what we do. We actually address the actual quality of the data down to the record level so that selections can be made literally on a record by record basis for whatever use case you want. And it's, it's. I mean it's no different than the best analogy to use for true set is really like the octane rating on a gas pump. Like you can imagine the days, which are the days of today in digital where you've got, you know, two competing gas stations on corners and nobody knows if one guy's putting water in his gasoline to stretch his profits and then the guy doesn't. You just wait to see if your car breaks, if it even breaks at all. You just don't know. And then the, the benefit of the octane rating is it's independent and it gives buyers choice, right? So if I'm driving a rental car, I can put 87 octane gas in. If I'm driving my own expensive car, I can put 91 octane gas in. I get choice. The sellers of the commodity, in this case the gas station gets to earn margin because the higher price stuff generally makes more profit. This is good for everybody. And this is essentially what we're trying to do for the data ecosystem. I would love to team up with the IAB and offer data ratings, which is an objective scoring sorting of data into easy to use grades that are appropriate for different use cases. AAA and B is how we're sort. We're gonna launch this in Q1 of, of 2025. So you're getting a little bit of a scoop here, Allison, but we're basically launching data ratings that anybody can use to understand the accuracy of data from provider to provider and also to choose the right grade of data for their use case from again, consumer analytics all the way down to broad targeting where accuracy is not as important.
- [39:25]
Alison Schiff
Let's get into what happens when a marketer has bad or low quality data in the mix because there are so many downstream Bad effects. I mean, if you think you're targeting a certain segment and it's basically just fiction or no better than a coin toss, then you don't really have a good sense of how your media should even be performing at all with the people it's intended for. And then that messes up your optimization and then it messes up your planning and then the whole thing happens all over again.
- [39:58]
Scott McKinlay
Yeah, I mean, I would start with the consumer perspective. I mean, just think of yourself in your own media consumption. Wherever you decide to accept ads instead of paying for content. The ad experience is really, really important. You know, and I like, I could put you in front of it. I could put anyone, anyone who's listening to this, I could put them in front of the television set and put, put them programming on and they could tell you within 30 seconds what kind of, what kind of, what they're watching. Are they watching streaming? Are they watching linear? Are they watching. You know what? How is there authentication here or not? Because the quality of the ad experience is directly related to the quality of the underlying data used to target those ads. And so the first casualty is just absolute irritation from interruptive advertising that isn't relevant, doesn't fit, and I have to sit through and suffer through in order to get my free content. And you compare that again to good experiences where, I mean, I actually think advertising can be so good that it's considered a service by consumers. It's actually meeting my needs. Exposing me to products that I care about will make my life better. That's what we should be shooting for. So first of all, let's start with the consumer. The user who gets a bad experience doesn't want to watch fast TV channels anymore because the ad experience is so bad. For example, I don't know what the example would be. And then for brands and publishers, let's just walk up the line. The next would be publishers. It's just low yield. Like why, why is inventory going for 25 cents and 50 cents in display and programmatic or, you know, $5 and $7 in legitimate streaming environments? It's ridiculous. It should be much, much higher. And it's all dependent on the ability of somebody to get the user right so that performance is there and that the brand and the publisher can agree that CPMs can be higher. Right now it's just absolutely crushing the open Internet because it's considered a commodity. Everybody's using mixed models. Those pay attention to efficiency, not effectiveness, which drives down the commodity pricing of all this inventory. And it's Just bad for everybody. And then finally, let's get to the brands. Of course, I'll tell you, here's a sort of mini case study. We started working with a alcoholic beverage company that was targeting Hispanics. We started by just measuring how accurate they were when ads hit the glass. And they were on average across all their portfolio for that target, about 29% on target at reaching that particular Hispanic audience. And we started working with them. The first thing we do is analyze their sources and by just pruning out a couple of low performing sources among the 45 or 50 segments they were buying off the shelves, we were able to bump that on target percentage from 29% to over 50%. So right away, phase one, they're throwing 71 cents of every dollar thereabouts on the ground by putting in language Spanish, you know, Spanish ads in front of people who can't speak the language. That's really, really bad for brands. We were able to bump it to 51%. And then by just selecting high grade identity from multiple providers, basically multi sourcing and just picking the best for every haystack we could find, we're able to drive that over 80%. So you look at almost a triple improvement in the ability to reach the intended audience, which of course is going to translate into performance because those people care about the product and they can read it in the language that's primarily spoken in their home. So there's an example of all three constituents. The, the users of media, the consumers of media, the publishers who are trying to monetize those audiences, and the brands who are trying to deliver products to those people. So they buy them in stores. They all suffer, you know, direct pain from poor quality data and they all can benefit from better data.
- [43:48]
Alison Schiff
So yes, it is, it is time to stop crying into our beer and do something. So we already talked about the Truthset Data collective. Data providers, they upload their data anonymously to you guys and they can get a quarterly benchmark and a sense of how they stack up against others in terms of quality. And the idea is that the tide will raise all boats. But I guess why should data providers trust you with this?
- [44:21]
Scott McKinlay
Well, I mean, it's a great question. We didn't know if this is going to work when we started the thing. It's a, it's a very audacious ask to go to these very, very large, successful data providers and say give us all your data, we're going to score it and it might hurt you because we're going to share that information some of it. Not, not explicitly, but we're going to allow people to come to you and say, hey, I only want to buy your great data or your, you know, your whatever. And they all did it. And the way that we provide value back is a myriad ways. We offer a free tier. First of all, it's not a pay to play. You can just participate as a data provider. You can get a score for your data, you can understand how you stack up against the average. But we do offer a range of really valuable diagnostic tools that are aimed at helping data providers diagnose and understand their own sources. What drives improvements in accuracy? Where should they just give up if their data is really, really far behind the competition and they should focus on areas where they're stronger or where they should heavy up strategically against attributes and categories that are high value and they're competitive in those spaces. So we offer again a range of dashboards and diagnostics. And to give you one example, one of our data collective members is very strong in the identity space and we were able to help them improve. Almost double, about an 85% improvement. Maybe that's a little too high. Maybe that's 65% improvement in their ability to put the right people in the right households, which underpins an awful lot of marketing and measurement. So a lot of benefits to the data providers. And it's worth noting, we have zero attrition. Everybody who has joined has stayed in and they generally moved up the tiers. And it does call out the very few large providers who haven't joined. And you have to wonder like, why haven't you? If there's benefit here, you get marketing benefit, you get badges, you get to say we're transparent. You get to allow your buyers to have choice over the grades of data that you can, you can acquire from these companies. That'll be coming in 2025. You know, why wouldn't you want to participate? And it's like, what, you know, basically, what do you have to hide? Like, we really think it's like, you know, again, three gas stations on those four corners. One of them doesn't have an octane rating and three of them do. Which, you know, why doesn't he have an octane rating on his pumps? What is he, what is he trying to hide? You know, and we're not trying to be strong armed about it, but it really is a benefit for everybody. It allows data providers to give their buyers choice and really diagnose and improve the fidelity of their own core data assets.
- [46:50]
Alison Schiff
I do have to ask though, you invited this question, who hasn't joined yet?
- [46:56]
Scott McKinlay
I'm not going to out anybody because we want them all to join and I don't want to make anybody feel bad. I would say it's the onus. Here's where responsibility can shift back to the consumers of this data. Like if I were to recommend the brands, we do this all the time. If you're looking at, if you're running an RFP for consumer data or audience data or whatever it is you're trying to do, ask them are they part of the True Set data collective? You know, can they share their scores like this is, this is available at the option of the data provider. We don't, we don't, we don't disclose anything about anybody. It's on the data providers who participate to do that. But they have all that data and they frequently use it to win deals. So, you know, if, if, if you're a buyer again, just like the gas station, would you put gas in your car from a gas pump that didn't have an octane rating on it today? There's no way you would, there's no way you would. You don't know how quality, what the quality is. You don't know anything about it. It might break your engine. So it's the same thing with your, with your data driven marketing. Why on earth would you buy unvalidated data from someone who is unable to tell you the sourcing, you know, the methods used to produce it, or give you any truly objective assessment of how accurate that data might be and how it might perform for you?
- [48:10]
Alison Schiff
How exactly though do the companies in the collective make what they learn through you actionable? How do they do something with that information?
- [48:21]
Scott McKinlay
Yeah, so there's, there's lots of, again, lots of diagnostics we offer, including in the indices that allow them to talk about their scores against the average in public. So we index everybody against 100. And so if you're 135 on multicultural data or you're a 111 on gender data, you're 11% better than the average. You're 35% better than the average. And we even in some cases do rankings where you're the best in this, you're the best in that we don't do it. But we allow the data providers to uncover those insights and surface them. And we also offer a badging program. So every data provider in the data collective can put badges in their market materials and their sales decks to show that they're transparent and accountable. And in 2025, what the badge will mean is that as a buyer, you can come to anyone who has a true set badge and buy different grades of data because sort of unbundling the data ecosystem. Just think about cable. I don't need 444 channels, I only want 12. Eventually I'm going to give up on the package and just get the 12 I want. It's the same thing that we're going to do for the data ecosystem among the data providers in the collective by allowing them to basically parse their data into these data ratings tiers and give buyers choice. If someone wants again, super accurate data to push into an LLM model or some sort of a seed file creation or analytics, they can buy AAA or AA data from anybody using the data collective who's optimistic into that. Or you can just buy the whole pile and still know you know what you're getting from provider to provider.
- [49:54]
Alison Schiff
So I've got to ask you something. We're nearing the, the end of the show. Going all the way back to the time you spent at Nielsen. You spent a bunch of years there, just over seven years between 2010 and 2017. A bunch of different executive roles, EVP of client services and EVP of global product leadership and eventually just EVP in charge of product innovation globally, including person level data and analytics, the data co op and identity management. So Nielsen, I mean, they've faced a bunch of challenges over the past few years. Kind of, you know, heavy is the head that wears the crown. The incumbent often feels like nipping at its heels. But you know, they've had an issue with undercounting TV audiences during the pandemic. There's been underrepresentation of certain communities in their data. I know you've been gone for a while, but you were right there like in the guts of the product at one point. So did and does Nielsen have good enough data to keep doing what it's doing while the world changes so rapidly around it?
- [51:08]
Scott McKinlay
Oh, man. I mean, you got another hour. Look, when I came to Nielsen, I was the digital transformation guy, right? I came in pretty high level after, after selling the company to Symphony previous startup. And I was supposed to go and shake things up and agitate and lead and I was just, I was a super skeptic. I was like the guy, well, I can't believe you're in business. What is this panel? We have got big data. What are you guys even doing? Oh my God. And it took me a couple years, especially in a global role, to really appreciate what Nielsen was and what it had built. And I got to say it's a tragedy that it's lost some of its sheen and position in the marketplace as the arbiter of truth for a thing that was very important to a lot of people, which is the size of audiences on linear television. So I think that they. And by the way, the critique around lack of innovation was not for lack of trying. What people don't fully understand about Nielsen is there was a whole financial engineering thing happening the whole time with taking it public and then back private and then public again. And the ownership being pe, where really, really smart people with really, really good intentions in that company were just unable to make the changes that, that we all knew could happen and that our clients were asking us for. So it wasn't for lack of trying or pure stupidity. It was just sort of the leadership was not supporting because they had other taskmasters that were financial oriented, the innovation that needed to happen. So it's no surprise that Nielsen has failed to innovate. That said, I don't think there's any doubt they're going to hold on to reach and frequency measurement. The bigger question is, and I think the pressure, by the way, from alternative currencies was very healthy for the industry. I think they're going to hold on to reach and frequency measurement because they've been forced to do innovations and they pull it off and they're doing, I think, a pretty good job with Nielsen one as far as I can tell. The question is how valuable is reach and frequency measurement going forward? That's the real question because we're moving much towards a world where it's not that important. We're going to get more and more precise with, with advertising. You know, it'll be move away from linear. Linear is aging out. Streaming will hopefully will not make the mistake that programmatic did and will actually act like I've been advocating this whole time and try to act more like the walled gardens than just, you know, digital property where you can take guesses. But that's my, that's my take on Nielsen. I have immense respect for that company. I really hope for its success. I'm sympathetic to what I struggled through is trying to be a change agent there. But the big question is how important is reaching frequency measurement for the next five or 10 years?
- [54:01]
Alison Schiff
It's a fair answer to a question I just sprang on you. So penultimate question. What are too many people in ad tech obsessing over right now? That to your mind is just fluff or BS or a nothing burger and you would love not to hear anyone talk about it anymore in 2025, which is like two minutes from now.
- [54:26]
Scott McKinlay
I mean it's back to what I said. I mean we. I'll use the royal we to encompass the collective legacy media that's trying to figure out how to survive, including television networks and open publishers. This whole fascination and obsession with a lot of the things we talked about today, id, syncing, guessing, probabilistic matching, quality, guess who doesn't have any of these problems? None of them. The walled gardens. Once again, who ask for email address and are responsible in the use of it as much as they can be. Of course there were issues there in the past. I think everyone's on the right side of that. Hopefully that's where we need to focus. And I go to these TV conferences, you know who's not in the room? Google or Meta or Amazon, like what are you talking about? And you know, sometimes I get on these panels and I ask the audience like, what are you guys talking about? Like, look at your own media behavior. 90% of my media consumption, mine personally, is authenticated because I get a better experience. Like God forbid you click on the, you know, Britney Spears, you know, tragedy link and you, you're like, oh my. Hold your breath. It's like leaving AOL in 1995. Like who knows what's going to happen, you know what I mean? So you know, I think the, the whole it's not a nothing burger, but the obsession with continuing to make guesses because we don't have the courage to ask users to identify themselves and develop that direct relationship that would, that would facilitate all this growth and profitability. That's the big miss. We're focusing on the wrong frickin thing. And the second part of I'd say we need to have an obsession around data quality. Back to this, the statements I was making about how this is not the 70s and 60s and 50s anymore, whereas all media and creative, it's all about the audience data. And the ecosystem has got to get that through its head so that it begin to operate at par in terms of monetizing audiences and surviving against these big tech media companies who are doing it right.
- [56:36]
Alison Schiff
You answered the last question I was going to ask you, but I'm going to put a twist on it. So what do you think is the most important topic in ad tech today other than the lamentable state of data quality that not enough people are paying attention to?
- [56:53]
Scott McKinlay
Again, you have to parse the media ecosystem between the tech media giants who have demonstrated that they have what it takes to earn the dollar from the advertiser and those who do not. Why I think that is and I think it's a existential crisis for legacy media. We keep measuring ourselves against where we were. And I can imagine the boardroom meetings and the, you know, the strategy meetings at a Paramount or a CBS or NBC or a and you know, the tech guys are in there bragging about how they're building an identity graph and they're now we've authenticated 12% of our users and isn't that better than 10 years ago when we never knew a single viewer of our programming? But that's a mistake because it gives a false positive like, hey, we're doing pretty good. We can identify 10% of our users. That's a mistake because the bar that everyone should be measuring themselves against is where the tech media giants are, right? It's they have 100% or 90% authentication. They have the ability to monetize five to seven times better than these legacy media properties with their amazing full sight, sound, emotion, content. So that's the Ms. I think Allison is we, in this case, the open ecosystem and the television ecosystem that's desperately trying to figure out how to survive in a world with Netflix's and Amazons and Google's and metas, need to stop pretending that we're making progress fast enough, because we're not. And if we don't hurry up, everyone's just going to end up licensing their program to the folks that can monetize it more accurately with accurate consumer data and eventually consolidate and probably go out of business. So that's not a future we want. We want to support the open ecosystem and fair and balanced journalism and all that. So that would be my answer to that question.
- [58:43]
Alison Schiff
Fair and balanced journalism and all that. I'm into it. This episode is sponsored by the Trade Desk Edge Academy. Whether you're a seasoned programmatic pro or are just starting out in digital advertising, the Trade Desk Edge Academy's expert LED courses and certifications are here to help you stay relevant, ready and ahead of the curve. Enroll today for free at edgeacademy thetradesk dot.