← Next in Media
Next in Media
Why CTV Ad Targeting is Much Harder Than You Think
00:24:24
Next in Media spoke with David Levy, CEO of OpenAP, about some of the misconceptions in the market when it comes to data-driven TV advertising, and how TV networks can balance collaboration and competition in the face of the growth of Big Tech in TV.
Loading summary
Transcript39 lines
- [00:01]
A
Accurate up to date customer information is critical to your business growth. It all starts with quality data. Epsilon has spent decades ethically sourcing the industry's top ranked consumer database covering every marketable U.S. household. With a commitment to transparency and privacy. See how you'll predict future buying behaviors and build a lifetime of customer loyalty. Discover the deepest customer insights available@epsilon.com data that's epsilon.com data this week on Nexting Media, I caught up with David Levy, the CEO of OpenAP. David and I got deep in the weeds on CTV ad targeting why it's so challenging from brands to match their audiences up across a dozen or so platforms and measurement services and really reach the audiences they want. Dave and I also talked about whether the big TV players are working against each other as they try and out walled guard in the walled gardens and why he's hopeful that the TV business can actually broaden its pool of advertisers. Let's get started. Hi everybody. Welcome to Next in Media. I'm Mike Shields. I'm here with David Levy. He is the CEO of Open ap. Hey Dave, thanks for being here.
- [01:09]
B
Hey Mike, nice to be here as well and thanks for having me. Appreciate it.
- [01:13]
A
Yeah, thanks. We, we you and I chatted at ces. This, this I will make you repeat yourself a few times, but that was a great session and this is kind of hoping to bring some of the same discussion to the to a hopefully broader audience that weren't able to make it. I think it's probably good. You're actually one of a few guests who has a returning guest which is exciting for you, I hope.
- [01:31]
B
I like that.
- [01:32]
A
But it has been a while and so it's probably worth of a just a basic catch up on Open AP because I know the mission has I'm sure evolved that we talked about this CS you guys launched when streaming was just a thing if anything. So catch us up on like where you are in today in the mission.
- [01:50]
B
Yeah, I mean I think the mission really started with this idea that all of the different TV networks were so focused on ad innovation. How can we make the experience of advertising within our great shows better for clients? Initially we were all kind of competing over who could have the best innovation. The challenge with that though, hard to scale. So you can only buy a new kind of ad innovation thing on Disney or on NBC. Hard for advertisers really scale that. Sophie's initial premise was this idea that hey, if we could take the innovations that are working really well across the networks and then build some standards around them so you could buy those new ad innovations everywhere that would be better for clients. And so that was kind of the initial incarnation of Open ap. And when we got around the table, you know, the biggest bucket that everyone was investing in was how can we allow brands who are pouring more and more money into their data infrastructure, into their first party data setup, allow them to take the first party data that they're investing in and actually transact on it across television.
- [02:58]
A
They want to use it, but they don't want to use it 12 different ways, right?
- [03:01]
B
Yeah, exactly. If you use it 12 different ways, you're actually targeting different people. And so could we, you know, set up infrastructure to make it standardized so that you could take a first party audience segment, let's say, if you're a brand and you could consistently target those consumers across multiple different media companies. And so Open AP initially started by building infrastructure around that problem and challenge. And the focus initially was very much so on the traditional linear space, which was where most of the dollars have been transacted. And we, you know, been focused on building infrastructure and pipes to make that more of a scalable proposition. Now I'd say over the past few years our emphasis has kind of shifted more towards how can we take that, that same concept and expand it into streaming? You would think it's much easier, but actually there are a lot more obstacles in that process of getting an audience all the way into a CTV ad serving environment that can change the audience along the way. And so ensuring that there's infrastructure in place to make sure that Warner Brothers is targeting the exact same audience that NBC is targeting was really important. So that's what we've been focused on over the past few years.
- [04:20]
A
I want to come back to that in a moment, but I wanted to ask you, what's the latest? Are there any, is there any, anything new coming down the pipe? Like what kind of things we should anticipate from you guys this year?
- [04:30]
B
Yeah, I mean, I think the big things that we're focused on this year is how do we bring more consistency and transparency into how an audience traverses from a client's environment all the way to a media company's ad server and ensuring that that audience is, as it traverses, as consistent and done transparently so the client understands how their audience is changing, if anything, from when they send it to when it's actually being targeted. And so we're making a lot of investments in clean room infrastructure and in quality identity data to ensure that clients can have a more of a transparent and consistent process when they're distributing audience. So that's the big investments that we're focused on. And then additional to that, a big gap in the market has been quality planning data within the streaming and CTD environment. And with the advent of clean room infrastructure, programmers are now able to make their data available for query without having to actually share it, which is a big thing because that will allow clients to actually have much better insights into okay, I've got my first party audience, right. I really care about this very specific audience. I want to know across all the different apps, where are they watching, what are they consuming, where are they? And historically you'd have to send that audience to each individual media company and you'd get varied responses.
- [06:02]
A
It's not like you can do that at your desk and just have a ranker and find that audience everywhere. Like you gotta actually, in your planning process you actually have to send it out and that could expose you to competitive challenges.
- [06:14]
B
Or I think today there's two ways you do it. Either you kind of work with each media company individually to get an answer and you're probably getting, you know, the most complete data set that you're getting back, but it's all fragmented or you're working with a measurement company that probably has incomplete information. And while it may be a streamlined, you're probably not getting a very good view of where that audience is and you're certainly not getting a good view of where that audience is actually available for advertising. So that, that's the challenge. So I think having more direct standardized way to query programmer streaming data is really important. So those are the two things we're focused on planning and identity consistency.
- [06:59]
A
I think you're right that people assume broadly well CTV is going to should be easier than, you know, old school data driven linear or more targeting on television because it's digital and you can find audiences match them up pretty easily. It's seen. Why do things break down? Like if you have a defined audience of your brand, why would that be hard to execute consistently? What happens?
- [07:22]
B
Yeah, so typically in streaming you are taking a audience, let's say of auto intenders. And as a client that audience list is going to be reflected as a list of individuals and then it'll probably be translated to a list of households that those individuals existent and then you're going to share that with all the different media companies that you want to target that audience against. Target those auto intenders. Well, the challenge is that what you don't really realize is that audience then goes through multiple different hops and translations before it actually ends up going into an ad server. And you don't necessarily have transparency or control over those hops. So what, what would happen in that process? As an example, if you have a list of households you don't target necessarily on that list of households, you actually would target based off of a set of device IDs or IP addresses, and that would be associated with a household. So you would use a DMP like a Libram or Cadence to take those list of households and translate that into targeting information. Well, if each of the programmers do that in a different way, then what you end up with is a different list of targeting information from the exact same auto and tender households. And what that really means is that when you go to actually target folks in your ad server, you are targeting different households. And so the challenge is, if you're using what we call different householding solutions, which is different ways to group identifiers to households, then ultimately you're going to have a lot of variation in the audience that you're actually targeting. And then when it gets to measurement, the same challenges are persistent. So in measurement, if you have a measurement company coming in and looking at all of the logs of where ads were delivered, and those logs are going to have identifiers with them, IPs or device IDs, if that measurement company is now going to take their own identity spine and map those identifiers back to a household, well, if those don't match what was used to create the audience, it's going to look like it was significantly out of target. But that's not necessarily true. It just means that they disagreed, the measurement company disagreed with the data company or DMP that you used to create the audience target. And so now you have a campaign that looks out of target. Maybe it was, maybe it wasn't. And you don't really have a clear picture of who was actually reached or did you reach your intended audience at all. It's a big challenge right now because there just isn't a lot of transparency to that problem that exists.
- [10:12]
A
Yeah, this idea that, oh, I've got Mike's email address. He's looking for a car. I'll just hit him with that ad. When he shows up in a place where he's logged in and they know his email address. It's not nearly that simple.
- [10:23]
B
Right.
- [10:23]
A
Is there a way to. Is it about transparency or is it about creating a standard way of doing things? I guess, I guess I'm asking is, could we use a single Identifier that everybody agreed with, is that even realistic? And could all this come together in a simpler way or is it always going to be kind of scattered?
- [10:43]
B
Yeah, I think there absolutely is a way to. I mean, I think the first piece of this is creating transparency and consistency. Ideally, the client can help decide how to take their audience and translate that into identifiers that can be used for targeting. And I think they can then enforce that that standard is upheld across all the different media companies. And then you can basically make sure that the exact same protocol is used for measurement at the end of the campaign. And I think when you do that, that at least solves for the consistency challenge. Now there are other challenges that exist and that would be just the quality of the data itself. So there's really two challenges. One is inconsistency and then the second one is quality data. So if you are relying on IP and device id, one of the challenges are that generally that data is, you know, purchased could be stale. Data usually is, IPs are rotating 3% per day.
- [11:49]
A
So there's a little bit of a, maybe too much reliance on IP addresses or belief that they are persistent and easy. That's not, that's not what you're seeing.
- [11:58]
B
I don't think, I actually think IP is a great signal. It's, it's very persistent in all the logs. You, you do get that in all the logs versus other identifiers. I think the challenge is it does rotate and unless you are committed to ensuring that you're keeping up to date with that rotation, then it's a hard signal to use. And so, you know, I think, look, that's why Trade Desk has moved to email, because email is a pretty stable identifier. It is relatively easy to keep consistent across the campaign process, but it's not prevalent everywhere. And the one challenge with email is that because it's not prevalent everywhere, you don't want to necessarily create a huge advantage for the kind of black box networks that are, you know, social networks that are out there that are the only ones that kind of have this kind of closed loop system. And so email is important to get as many authenticated, logged in consumers as possible. But when you don't have that, we still need a infrastructure in place to make sure there's consistency, transparency and ideally quality data within that as well.
- [13:08]
A
Plus there's always a possibility that IP address could be vulnerable to regulation. Right. Some of the, depending on who you talk to, some of the states could go after this usage. I haven't really Seen that yet, but that's, that's an unknown 100%.
- [13:21]
B
Yeah, I think all these identifiers and we don't really know what's going to happen from a privacy perspective, which is I think why. The important piece though is a process to ensure that consistency is happening from the targeting to measurement process. And then I think whatever the identifiers are that we were using, or it's ip, device id, email, et cetera, the highest quality data that we can use possible is going to be important. And on each of those I would say there's different levels of quality data that are out there. And most of what's used today is not quality data.
- [13:56]
A
You know, you've emphasized OpenAP has always been about this. The TV industry, let's collaborate on stuff that is not worth competing over. Let's get our acts together because we want to be able to compete with these big tech giants. Like you mentioned, I think most people in the industry would buy into and agree with. But then there's, there is still this instinct to go your own way. You see like a Disney talking up its own identity graph and identifier and you know, it's, it's a unique set of, of buying tools and stuff like that. And a lot, a lot of the media companies do that. Are those two things at odds in your mind? Is that, is that counterproductive or is it just like an inevitable?
- [14:30]
B
I think you have to go out and, and play up the advantages that you have as an organization and where you're investing and spending money and why that's going to be better for clients to work with you. You know, I think that as long as you're also kind of collaborating and cooperating in some of the bigger industry initiatives, I think both things can be true and, and both can benefit you. So you know, we talk a lot about creating some standards and quality data. The other piece is like a logged in consumer or sure is an authenticated consumer experience is going to be some of the highest quality deterministic data that you can get. Where you know a logged in consumer is watching something. And if all the programmers work together, they can grow their pie of authenticated streams that they can actually have access to. So I think the more we can all collaborate together, the larger we can kind of make this addressable footprint where we know a logged in consumer exists. And so I think there's opportunities for everyone to collaborate and also continue to invest in their own spine. There's going to be neat and it's necessary for you to have your own spine for other purposes too. Like, you know, Disney has a huge business. Right. Parks business. Then you know, out of home business. There's lots of different things that they do that they're going to need to have a good sense of identity for their customers.
- [15:50]
A
Yeah. And they don't want to give themselves like.
- [15:52]
B
Yeah, yeah. I mean that, that makes it, it's more about how can you on a campaign keep consistency and be interoperable with these different identity spines from the networks. And I think that's possible without. I don't think there needs to be one spine that rules them all. That's not, that's not at all what we're proposing. I think all the networks and media companies and agencies likely will all have their, their own investments in data and identity.
- [16:19]
A
So you, you want to make sure that you're moving forward, everyone's innovating, but also recognizing the challenge that's happening with the rise of the big streamers, the tech world. Do you think everybody in the, in the industry understands the gravity and is there anything that's frustrating to you or an obstacle that you'd like to make sure that people understand that they need to push forward on?
- [16:37]
B
Yeah, I mean I do think the longer that it takes for us to have more streamlined solutions where even smaller advertisers can, you know, execute a highly targeted campaign across premium content and television like content, the more that goes where that's not true. You know, there's a lot of risk that exists in the industry for us. So you know, I think the sooner that we can realize that vision, you can almost feel it yourself. Like if you were you open a local business today, would it be feasible to advertise across the big media companies? It probably isn't. And that's some place I think we need to get to and. Cause there's just not enough scale in just one media company on their own. But if we can make it easy for you to take that audience onboard it and buy like a $25,000 media campaign across five different networks, that's where I think we can all win. And CPMs go up. You know, we need to get from like that couple thousand advertisers to the hundreds of thousands of advertisers.
- [17:47]
A
Yeah, you're right. That small business, it's probably, yeah, it's easy to log into Meta and Google. Right. They know those probably it's, it's everybody in TV has a self serve option, but it's probably daunting for the, those, those folks to figure them all out at once. So that's. That's an area where you.
- [18:03]
B
I mean, you're never going to get to where Facebook and go. They don't. We don't have, like the stand process is much higher in television and the creative. Like, it's just a much higher threshold. But I do think with the opportunities that exist today where you could see things going with AI and creative, you know, we need to be set up to enable for more scalable ways to buy smaller campaigns that are highly targeted across all the, like tv, like inventory. And that's doable. And I think it's being worked on by the programs. It's not necessarily always a top priority. And I wish that was different. But I also get it. Like, this industry has a lot of pressure right now on it and resources are dwindling all the time. So it's hard. It is a very difficult environment for executives right now.
- [18:52]
A
I want to come back to something you said. You mentioned how it's just not that easy to. There isn't a great. A single planning tool that you can just plan a campaign across streaming in the way that probably brands would like. You might think that would be true today because there's been so much investment in alternative measurement and currencies. So that's a long way of me asking, like, where are we with measurement right now? Is it. Have we advanced as fast as we would hope to? And could one of those players emerge as a better way of planning this stuff?
- [19:20]
B
I mean, it could. I think we focused a lot on currency and on measurement over the past few years, as if that was going to be the panacea for this industry. I think the reality is that as more and more viewing moves into streaming. Yeah, you're changing the whole calculus of the conversation. So in linear, there's measurement companies who had a panel of a small number of deterministic households they knew, and then they're kind of modeling up everything else. Well, in streaming, we know we have signal on 70, 80% of the impressions. Where was that impression delivered? We know the IP address, likely, we know the device ID for that impression that was delivered. And it's not so much a modeling game as much it is how do we accurately link that IP or device ID back to the right household and how do we count impressions? So I think to me, as we move more and more to streaming, the counting game of like, how many households were reached? You know, I'm not sure that's that actually important for measurement companies to solve. I think what's much More important is how do we consistently resolve signal back to the right household and then be able to open up measurement towards a lot more measurement innovation further down the funnel. So we can understand how these ads are actually pushing people down the funnel all the way to becoming customers. And I think that's certainly doable. But that's where I think when I think about measurement innovation and where we are, what I would like to be in a place in a year or so is be able to say we have this great environment in tv, almost like a sandbox for measurement companies to now come in and do all this type of measurement innovation that can lead to better ROI for customers. And they can do that because data is available consistently, you know, with campaigns and they can kind of do, they can bring their own data into the, into the equation and say this ad moved this group of people down the funnel in this way and here's where you should invest further. I don't think the currency piece is as important moving forward. I think we'll be able to count and bill even without measurement companies. It's more to me understanding one who in the home saw the ad, I think that's important. But then also what did they do? Like how did that influence them? And you know, we continue to hear more of that from clients too, like getting further down the funnel.
- [22:01]
A
Dave, last thing, I keep asking all these questions about problems and obstacles as I tend to do. What's got you really excited, maybe what's got you pretty optimistic about the next say like six months where, where your team is headed or the, or the industry overall?
- [22:14]
B
Yeah, I mean I think I talked about it is part of the problem. But I was also where I'm excited, which is I do think we're going to be in a place where because of the advent of kind of clean room technology we can have a lot more of this streaming data be interoperable between multiple programmers and agencies without that data needing to be actually shared. And what that gives us the ability to do is be able to consistently apply more valuable data to resolve as many exposures and viewership data as possible to households and individuals and then open that up to measurement companies and innovation so that we can have a lot more insights into what's actually happening and give more actionable insights to customers into how to drive more roi. I think the more that we can be consulted around, you know, how certain messages are resonating with different groups, it's going to be really important and so the more kind of ROI driven that we can be I think the better. So what excites me is I, I do think within the next few years we're going to be in a place where we can kind of open up this sandbox, the endless possibilities for measurement companies to come in and innovate and, you know, really start to feed more insights to clients around how they can improve their marketing campaigns. Right.
- [23:35]
A
And that's, that's when CTV really gets to its promise that we've all been waiting for.
- [23:39]
B
Yeah, exactly. And I think we're close to the, we're never going to be the same as one closed loop system like a Facebook or Google across this industry, but I think we can get really close. And if we can get to the place where an advertiser can activate a campaign within, you know, a couple days, that would be, I think, the goal. And I, I think we're pretty close to that vision. It's just we got some, a little bit more work to do to get there.
- [24:03]
A
Well, Dave, awesome stuff. Thanks so much. And let's, let's do this again sometimes.
- [24:07]
B
Sounds good. Thanks, Mike. Appreciate it.
- [24:10]
A
Thanks again to my guest this week, Open AP's David Levy and my partners at Epsilon. If you like this week's episode, please take a moment to rate and leave a review. We have lots more to bring you, so please hit that subscribe button. We'll see you next time for more what's next in media. Thanks for listening.