Why AI-Driven Marketing Fails Without Fresh Identity Data, with Experian's Matt Tipperreiter and AtData's John Oronte

0:04John Oronte: Inside the Stack is a custom podcast series from AdExchanger where we interview leaders in ad tech.

0:16John Oronte: This episode of Inside the Stack is brought to you by AdData, a part of Experian. Ad Data helps businesses unlock intelligence about the person behind the email address with email validation, data enrichment, identity matching, and fraud protection.

0:43John Oronte: Welcome to another episode of Inside the Stack. I am Lindy Johnson, Senior Director of Content and Community at Ad Exchanger. Who is teaching AI who your customers are? AI is beginning to shape how brands understand consumers and decide what to do next. Yet many marketing organizations are automating decisions before they have established a trusted view of the people. those systems are meant to serve. The defining question is no longer whether marketers can connect more data. It is whether marketers control the consumer intelligence guiding their AI. In this episode, we have Matt Tipwriter and John Orante from Experian and AdData to examine why consumer intelligence is becoming one of the most important business assets for brand.

1:47John Oronte: They will explore how durable identity gives disconnected signals, meaning by connecting them to a coherent view of the consumer and why AI can only be as effective as the information it receives. Hi, Matt. Hi, John. Welcome to Inside the Stack. Hi, Lynn.

2:09Lynne d Johnson: Great to be here. Thank you.

2:11John Oronte: Yeah, it's great to have you guys. I'm very excited to dive into this conversation and understand how consumer data and email data help shape how marketers learn more about consumers, especially in this AI era where the AI is only as good as the data you feed it. And that's where you guys come in to give great data. So, Matt, I'm curious, you've been at Experian for over 25 years. So you've lived through identity going from cookies to graphs to whatever we're calling this AI moment. What's one of those, wait, that's not how this works anymore moment that has stuck out with you now?

3:06Matt Tipperreiter: Yeah.

3:06Lynne d Johnson: Yeah, it's been a journey over a pretty long arc. So when I first started, the main channels for direct targeting was direct mail and telemarketing. We had something called the white pages, which basically represented a pretty clear source of truth, names and postal addresses. And every CRM system was built around having a very complete profile. and having a profile of a consumer was truly necessary. You couldn't operate in direct mail, of course, without it. So you had all these systems and processes built around it. As time evolved and different channels started emerging, email, social, mobile, you name it, every environment really only needed a part of that consumer profile record in order to do the execution.

3:59So the amount of data available at the point of execution became less and not more, which presented a very real challenge for a lot of marketers as they adapted platforms and systems.

4:11Lynne d Johnson: So I think the realization for me along that journey was every time that happened, you had to get comfortable with no longer having that complete record available in every single system. What was important was maintaining the understanding of the consumer in those touch points. And certainly in today's environment, where AI is lending a hand and making a lot of the decisions, that underlying understanding and context is just as important, if not more than ever.

4:42John Oronte: Yeah, I could imagine, especially I remember, you know, for five, five years ago when we were going through this cookie list era and everyone was talking about hashed email. That's the answer. That's the answer.

4:55Matt Tipperreiter: Right. Yeah.

4:58John Oronte: So, John, before email intelligence, you spent years doing cross-cultural work in China and Beijing, Center for Chinese Studies, Asia Society, Give to Asia. Do you think there's a through line for you between figuring out who someone really is across a language and culture gap and figuring out who's actually behind an email address? us?

5:26Matt Tipperreiter: I think the short answer to that is not necessarily.

5:32John Oronte: A weird question, right?

5:34Matt Tipperreiter: Yeah, I mean, to be honest, that, yeah, it was a different time in my life. But I do think that there was, you know, some tangentially, you know, beneficial benefits to like, to that experience. The most important thing being that it sort of helped me build a network that put me back in the US in San Francisco at a time when And San Francisco was sort of going through this, call it a renaissance, of becoming this hub for international, you know, innovation and technology. And it was almost impossible to sort of go about your day to day without running into sort of these major moments within the industry, such as like the Twitter IPO and, you know, Salesforce, you know, joining the Global 500 and Dreamforce being like a massive event in downtown San Francisco.

6:21So ultimately, that whole trajectory put me in a place where I was networking frequently with people in the industry.

6:29Matt Tipperreiter: And I ultimately came across Experian. And I think the thing that's, you know, the rest is history. I joined Experian then. And the thing that I found surprising is actually how many people have a similar background like that, right? I don't think that people who have that time in China necessarily use their experience in the ad tech industry in America directly. But I do think, you know, going abroad and trying to, as you sort of put it, like understand that the gaps helps you sort of approach problems in a unique way, especially when you're not used to sort of those different experiences. So a long way of saying not really, but somewhat, you know, it's certainly helping.

7:11John Oronte: Right, right. I could imagine just like understanding different cultures and people. I could see it as a through line, like understanding identity is, you know, people are not constant, right? They change email addresses. They use one email address for shopping and another one for speaking to their friends, right? So all of that, how do you resolve all of that and really get a full picture of who the person is? So I think we're going to dig into that a little bit now. As I mentioned earlier, talking to Matt, how hashed email has been all the rave, I think, for the past five or six years. It's the default joint key everywhere.

8:05John Oronte: Publishers have been told to get those logins. You know, when we talk about CDPs, clean rooms, identity graphs, all of it, we're talking about hashed emails. But a hashed email doesn't always tell you if that inbox is still active or if it's a burner, right? Or even if someone's using Apple's infamous hide my email. So what's closing the gap on your end when it comes to that? How do you know when a record's gone stale before a model can even act on it?

8:40Matt Tipperreiter: Yeah. So I'll take a step at that. But, you know, so I'll just answer that directly first and I'll provide some context. Basically, AdData has a proprietary network of email activity metrics, which, you know, obviously was recently acquired by Experian. And this gives us line of sight into roughly 99 % of all email activity in circulation today. And, you know, it's constantly being updated with signals, real-time signals for various partners and proprietary services that we provide. And in essence, what that gives us is this heat map, the world's largest heat map of its kind, really, of insights into email activity.

9:25Matt Tipperreiter: And the exciting thing now, obviously, that we're part of experience, is that we get to pair that with industry-leading identity resolution capabilities. And kind of as you and Matt have already alluded to already, I think that really aligns with the shift we're seeing in the industry, which is we're going from this period of more meant more. Matt talked about the complete profile. But now we're moving to this idea where less could mean more, right? Meaning that quality is now sort of taking more precedent over scale to drive better outcomes. And I think you can kind of see evidence of that all over the place.

10:03Matt Tipperreiter: For me, what that means when these issues kind of hit my desk is like people will say like there's too much noise in our data. Right. Or like how do we how do we narrow down to the right audience or the right people within this within this data set? And that can infect everything from segmentation to measurement. And honestly, the thing that's probably hardest to quantify but has an enormous drag is just trust in the data altogether. So for email data specifically, I think the model of more is more is a little bit problematic because as a data point, email is pretty transient. You can go out and essentially create an email that you could use every single day, and that's difficult to replicate when you're talking about moving to a home or an apartment or sending up for a new phone.

10:53Matt Tipperreiter: And that's completely legitimate. it. So I think as we sort of shift and think about like quality in this context, you know, there's certainly utility in sort of knowing the linkages either within your own data set or in the form of match rates on outside platforms. But I'd argue that the way in which we want to think about quality in terms of an email is whether consumers are actively using it right now. and I think I can give you a real life example. So premise of this engagement was we were looking at a client's new to file records and they had worked with a third party provider to ensure that the emails that they were collecting were actually belonging to that person and so they were dealing with what they considered verified emails.

11:42Matt Tipperreiter: We ended up taking a look at that that file and what we found was that almost like 14 percent of those emails that were verified, whereas what you just described as burners are essentially temporary emails that were set to expire within under a month. And so this analysis, you know, kind of gave them some direction on, okay, well, here's what we want to remove from this population. But beyond that, we can take a look at what's in this population and say, these emails haven't been used in six months, whereas these emails that you've been provided, They're using these emails every day, and that gives them some direction to help make better decisions with that data.

12:23Matt Tipperreiter: I don't know if Matt wants to add anything to that, but that's from my perspective of how that kind of looks.

12:28Lynne d Johnson: Yeah, absolutely. Very much agree with John on his points there about the email address and how it's being used. From where I sit, I think about freshness in data at probably multiple layers of identity, and it starts with what everything John outlined. From there, you're ultimately going to be using identity graphs and platforms to do execution or add execution. And so the freshness of the connection of the email address to other digital IDs is also important. So when you're thinking about how graphs are built, understanding just how far back or not those connections go, because that is also a proxy for activity, a proxy for trust in the data and the connections being made in reaching the consumer that you actually intend.

13:21Lynne d Johnson: So I know we'll get into a little bit more of that, but the freshness of those connections goes hand in hand with the freshness in your own first party data.

13:31John Oronte: Yeah, I think that's important, right? Like for a long time, marketers, publishers, renting data, right? Or using third party data and being able to make the connections, as you guys said. But then also like, you know, how can they control it, right? How do they know the freshness, right? How recent it is to you guys point. And I guess this becomes an even bigger issue once we put AI into play, right? Because if AI is acting off of old data or, you know, data that you can't even act upon, that would be a problem. So how would a marketer, say, check whether the identity data they're feeding into these AI models or to make decisions is something they control versus something that they don't, they can't really control, you know?

14:38Lynne d Johnson: Yeah. Yeah. You're really, you're joining two worlds together in many cases. You're joining together everything that you truly have control over, which is your own data, the intelligence behind it, the insights that you've derived. And then depending on the platform and the channel, you're then depending on someone else's data. So I think being able to understand with transparency in how the underlying identity data that you're relying on for execution was created. And that comes with partnership with your platforms and the companies that you work with. And it also comes with a lot of testing and learning.

15:20Lynne d Johnson: Do the behaviors and the outcomes that I anticipate occurring, are they actually happening? And do I understand why? It's going to come through a lot of that test and learn. So not easy, but the more transparency you have from the tools you're using, the more insight you have about the data that you're tapping into. I won't say it's easy, but the easier it becomes to understand that full end to end. And there's certainly going to be cases where you aren't owning someone's graph, you aren't owning something, you're leveraging it. And it's an important partnership. So being able to have a framework that gives you that flexibility and transparency is going to be pretty important.

16:06Right.

16:07John Oronte: So I definitely got from this because you said the word like five times that transparency is key. A marketer wants to work with a partner that is providing transparency. That is totally key. John, anything to add to that from your perspective? I can add the word transparency.

16:27Matt Tipperreiter: I can say the word transparency a couple of times too. I think, well, so, I mean, I think that there's this, there's this debate where it's like, do you own or rent your identity graph? But I think that there's, I think Matt kind of framed it perfectly. Like, I don't know if it's so much that you're, you're living in a, in a, in a place rather than using a tool. You know, and, you know, I'm, I'm a big fan of building stuff myself, but I also can appreciate that the right tool makes very light work. And I think that the reality is that like a lot of brands just don't have the capacity or the ability to maintain like an infrastructure on their own.

17:06And I think that an identity infrastructure, I mean, and I think that working with partners doesn't necessarily like, you know, relinquish autonomy, especially if there's transparency within sort of the relationship.

17:20Matt Tipperreiter: But again, the most important thing I think is Matt Testapone is like there's the first party data that is included in the framework. And in addition to having that relationship with partners, you yourself maintain a persistent identity within your data, such as email, as an example, that can travel across platforms. It's sort of ubiquitous in the areas that you want to go and meet consumers. And if you throw an AI in the mix, that underlying identity data just becomes way more important because either it's empowering you to make very intelligent decisions or it's scaling your ability to make very bad decisions.

18:04We run tests like this all the time where we have a client where their platform is working

18:13Matt Tipperreiter: with brands and they're struggling to perform and they're looking at us and they're saying, well, can you give us more data? We need more data to solve this problem because the results aren't there and our customers aren't happy. But again, we ran the analysis similar to the example I provided earlier, 15, depending on the client, depending on the brand, 15 to 20 % of the data they were taking on to their platform were essentially dormant, meaning that no one had used those records for six months or no one had used those emails for six months specifically. And, you know, when we shared that with them and the way that allowed them to then go explain that to their clients or the customers that they work with, they were able to then make the adjustments to then say like, okay, we want to be able to use this population.

19:01Matt Tipperreiter: And, you know, that drove better match rates, better outcomes. And then at the end, the reporting made sense. So I just think that, like, you know, that's my two cents on that. I mean, Matt covered a lot of it, but the transparency and being able to partner with someone, I think, can really drive meaningful results without anything. Right.

19:24John Oronte: I don't remember your exact phrasing, but it was something like, it wasn't a good partner. I forgot the exact phrase. You used what you said. Something like a good partner, light work, is light work. Having the right partner is light work or something like that. Do you remember what you said?

19:39Matt Tipperreiter: Yeah, the right tool makes life work, right? It's the right tool. You can do the job much faster.

19:45John Oronte: Okay, because that's a bar. I don't know if you realized that was a bar. That and transparency, I think, made this all make sense for me. And I guess, you know, something I always think is like, explains me like I'm a five-year-old, right? So I'm a beginning marketer and I really want to understand what you're telling me. Like, I want to be able to reach these people. How can you ensure that I'm reaching people who are actually going to take action? Explain it to me like I'm a five-year-old. Does that make sense? Did I throw you off too much with that question?

20:31Matt Tipperreiter: A five-year-old? I've got an almost four-year-old and I know Matt has a almost four-year-old as well. Okay, Dan, this is a perfect question.

20:42John Oronte: I didn't even know. So this is perfect. So you're explaining to this kid, you know, toddlers making new friends at daycare or wherever, right? Like, how do you know it's the right group of friends or something? I don't know if that's comparable, but yeah. Now, explain to me like I'm a five-year-old. Five-year-old marketer, though, not your kids.

21:10John Oronte: Oh, did I stump you?

21:11Matt Tipperreiter: No, I mean, I'm trying to think about how I would say this legitimately. It doesn't have to be a five-year-old. Maybe I might ask you.

21:19John Oronte: Just identity 101. How about that? Let's take it to identity 101.

21:23Matt Tipperreiter: The core of it is, like, I think, and I'll just put it out there, that I think that you want to be able to identify the right person and deliver the right message at the right time, right? That's your objective. There's lots of different channels that you can do that in. And more often than not, those channels today are digital. So the way in which you need to sort of accomplish that task, it varies. So I think, you know, I can get you right now. My three-year-old's already lost the script on it. But that to me is kind of the basics, right? You need to be able to have the ability to deliver the message to the right audience at the right time.

22:04Matt Tipperreiter: And that requires a mix of different solutions to get there, depending on what it is you're trying to do and what industry and the channels you're working in. I don't know, that's vague enough, but Matt, I don't know if you know. No, it hits it.

22:17John Oronte: Go ahead, Matt. You have something to add.

22:19Lynne d Johnson: I'm picturing my daughter playing with either blocks or magnet tiles, right? Right. So and they're sort of in this era of like matching shapes and colors and images and putting things that belong together together. And I think identity is a lot like that, especially when you're talking about the foundational layer. Right. Everything is going to rest on a proper foundation. You got the bigger blocks at the bottom. You have to have the tiles that fit together the way that they are intended to. And then you have some hope of turning that into something useful or even more playful than it is. So I picture that as each one of those blocks or each one of those tiles represents maybe an identifier about a person.

23:05Lynne d Johnson: It represents an insight about a person. And you're trying to assemble those in a way that give you as complete a picture as you can because that's what's within your control. when you start playing with your friends, when you're in a room of people, they may have other ideas, but you start the communication of if I take what I have and join it to what you have, do we come up with something that's even better or works for both of us?

23:29John Oronte: Wow, you really hit that task. That was great. That was great. Wow.

23:38Lynne d Johnson: I'll go thank my daughter when I see you for waiting today.

23:42John Oronte: That was great. I love that. Thank you. So in the identity game, we have many players, right? So there's players like LiveRamp, Google, the Trade Desk, right? And many of those players are looking for their customers to lock in to their system. But you guys are talking about, you know, identity works across all of these systems. It's not locked into just one. So when you're trying to explain that to someone, where do you meet resistance? Is it? Yeah, yeah.

24:24Lynne d Johnson: Yeah, I'd start off by saying that these are important companies in the ecosystem. They were power advertising for a lot of different companies, and they all have things that they do that are highly important, and we work across all of them in various ways. So it truly, to me, is an ecosystem of technology, data, identity that makes this economy go. I think when I talk with specific companies about their challenges and their four walls, it has a lot to do with how identities already kind of pre-wired into their ecosystem. And they have a lot of different organizational teams that own different components.

25:11Lynne d Johnson: And they themselves wrestle with how do I ensure inside my four walls that I've got that complete single view that I that I know how to make connections for the use case at hand. So when you talk to them about something new or different, even if it's better, there's going to be resistance because better represents can represent change, even if it's a good change. So I see it as sort of a natural and understandable initial concern or fear that everything's going to change. And then as a provider into a lot of these companies, we have to work with them and show them the path. You can't just hand something to them.

25:55Lynne d Johnson: And so resistance may be a little more lowercase r in a lot of cases. But they are certainly thinking about the big picture. If I make a changeover on one side, how does that cascade through to where I have these big, important integrations and partnerships already? Does it help hurt the neutral? How does that affect me as the company?

26:16John Oronte: Right. I would think it would make more sense to them now, especially as more of these companies are looking to have a holistic view of their target or of their customer. right john anything on that like you know of you know helping customers to see that yeah it takes many players to get this quality that you talked about earlier yeah yeah i mean i i think matt

26:47Matt Tipperreiter: kind of nailed it on the head you know there's there's resistance to change but i think that there's, you know, there's an acceptance of being adding incrementally to what it is you're trying to accomplish with your data. And I think that, you know, you know, as far as those, those, those partnerships that you mentioned, the types of companies you mentioned, like, you know, there's, they serve a very important purpose, as Matt pointed out, because, you know, Now, there's structurally, like you could be a CMO and you could have very aggressive targets and it's impossible for you to establish partnerships with multiple different publishers at the same time because the onboarding and implementation timelines are completely out of your control.

27:36Matt Tipperreiter: And so there's reasons why you would want to work with a platform that could put your ads on multiple different publisher sites or something like that. There's all these different nuances that organizations are dealing with. And as Matt put it, this isn't a new idea. They have a lot of workflows or pipelines, whatever you want to call them, established within their business and their owners within them. So I think that the idea of coming in and saying, this is going to help you make better decisions is just naturally leads into a conversation about incrementally improving what it is they're doing.

28:19Matt Tipperreiter: I think there's also the sentiment like this is working well enough, right? So no one's going to rip and replace anything. It's very rare, I think, that that happens. But it's like, you know, it's about being able to help people make better decisions. And, you know, whatever the thresholds there are for, you know, this is working, you know, there's always ways that you can demonstrate, especially with data, that there's ways in which you can improve their outcomes by adding different signals or different identity elements or different, you know, attributes to their repertoire.

29:01John Oronte: right so i think the key here is i'm hearing quality i'm hearing transparency i'm hearing that um there was enough word there and i just lost it i'm sorry quality transparency and uh i lost it it'll come back to me at the end when i wrap it up i'm sure but i i think you guys covered a lot of the now, which the now is definitely about quality and transparency. And maybe even the future is about this as well. So there's a way I like to close out the Inside the Stack podcast. I like to have the guests give me three words that they think describes the future of their particular part of the advertising ecosystem.

29:52John Oronte: So for you two, I want to hear three words that you would use to describe where you think consumer identity, consumer intelligence are heading. And tell me why you you pick those three words and then maybe my three words will come back to me.

30:11Matt Tipperreiter: Yeah.

30:17Matt Tipperreiter: I guess mine would be. Three words, I'm going to go with way less static. It's going to be way less static. I think there's just the idea of having predefined audience segments isn't going to necessarily go away. But with AI, the decisioning layer, I think, can be a lot faster and basically operate in real time. Instead of deciding in advance, this person belongs in this bucket, and then bucket X or why, you know, and then having that refresh as, you know, you go through life events like having getting married or having kids or buying a car, et cetera, you know, then now you have the ability to basically in the future, I guess you'll have the ability, depending on how you're using AI to make these decisions faster.

31:10And so I'm thinking in the future, we're going to be

31:13Matt Tipperreiter: making using it, thinking about identity in a different way.

31:18John Oronte: Right. I have concerns about that, but because faster is not always better, right? So what would, you know, I think that goes back to your argument about quality, right?

31:35Matt Tipperreiter: Right. I mean, obviously the premise would be that you need, that premise will not change. If you put bad data in, you will always get bad data out. It doesn't matter what, you know, Play-Doh mold machine you're using to actually produce something, right? To stick with the five-year-old thing. But I think that AI can make decisions faster than previously we were able to. So I think that that's going to happen. And I think that people are going to try and it will get to a point where you're trying to use behavioral signals or deterministic signals to actually try and meet consumers where they are in a way that I think we previously weren't able to.

32:14Matt Tipperreiter: And yes, I think that's going to be in real time. but you know to your point like there's that could result in a lot of different new challenges so especially if the quality isn't good but i you know the decisioning i think is going to be faster and i think it's going to rely on a way more less way rely way less on predefined buckets than we

32:35John Oronte: previously have been okay all right sounds good matt all right um so we have three words picked

32:44Lynne d Johnson: out and where they come from is just the overall thought that identity is about maintaining the trusted consumer understanding across how things change. So continuity would be first word, identifiers change, emails change, devices change, channels, things are constantly changing. But what we want to strive for is continuing that accurate understanding of a consumer. um so that's number one number two is choice some people might refer to this as interoperability um it's a very frequently used word but if we think about it in the terms of companies for buying they want to have choice they want to know that they can adopt technologies and platforms and systems that will fit their need without having to change over everything So the complexity of this ecosystem probably isn't going to go away, regardless of efforts to standardize on one particular ID or one particular stack.

33:48Lynne d Johnson: But allowing companies to choose the best platform for the job is going to be very important, has been, will continue to be. And accountability. So as AI is making more of the decisions, I'll throw out that word transparency again, but just understanding what is driving those decisions, how current is the data, how accurate was the data that was fed into those decisions? And are you able to update those decisions as knowledge and information changes? wow yeah your answers your answers were like they really dovetailed john's really nicely i i like how

34:35John Oronte: we we went from um you know that the shape of yes always bad data in bad results out right and um So how important that continuity is, as well as the interoperability piece, where I love that you picked that up because, you know, you guys are talking about working across different platforms, right? So a marketer is going to be concerned about how does that fit into my stack, right? So, yeah, everything may not be plug and play. There will be some challenges, right? Exactly. Yeah. And then the accountability, because in the AI era, right, it's like people keep thinking, you know, and as John talked about it being faster, people keep thinking that means less human involvement.

35:31John Oronte: But I think human involvement becomes even more important because of the accountability piece that you talk about.

35:38Lynne d Johnson: A hundred percent. And this is something we even reinforce internally. It's great to use new tools to build products and features. It's the world I live in. But how much are we validating the work that it did? It's not hands off. It's just a new powerful tool in our toolkit. Okay. Yep.

35:58John Oronte: Yep. Well, I thank you guys for schooling me like a five-year-old about identity and intelligence, where we are now and where we're going. I think this is a really important conversation for marketers to hear right now as they're looking to solve targeting problems as they can't reach consumers the way they used to because of how the LLM models have come into, yeah, that discovery phase, right? and also as they're rethinking measurement so measurement is performance marketing if I hear that anymore I don't know I'd be rich I'd be getting a penny for every time I hear it so yeah you guys really gave me and are giving our audience a lesson in all of this and I thank you for your time Thank you, John and Matt, for your time today on Insights.

37:03Lynne d Johnson: Yeah, thank you, Lynn. Really appreciate the time here. Yeah, thanks for having us.

37:11John Oronte: This episode of Inside the Stack was brought to you by AdData, a part of Experian. AdData helps businesses unlock intelligence about the person behind the email address with email validation, data enrichment, identity matching, and fraud protection. Thank you.