Identity & Data Privacy

Privacy, the Death of the Cookie &’s Identity Solution

A Conversation with’s Chief Technology Officer, Paul Harrison

Transcript David McBee: Hello and welcome to the webinar series. I’m David McBee, director of training. I’m here with’s chief technology, officer, Paul Harrison, for an update on all things related to privacy and a deep dive into’s identity solution. Hi, Paul, thanks for joining me again to talk about privacy and identity. Paul Harrison: Hey, thanks for having me again. David McBee: Absolutely. We talked what? Back in April, so it’s been four/five months. We discussed Google’s plan to eliminate cookies from Chrome and how you thought it might impact the industry. We also discussed Apple’s iOS proposed changes and how it tracks location. I remember thinking after that interview with you, Paul, that you sounded very confident and prepared for the changes. It’s been a minute since then, and I thought it might make sense to check in, see what’s changed, see how you’re feeling about things. I really want to hear about’s identity solution. Paul Harrison: Sure. David McBee: Let’s start with Google’s announcement that based on mounting industry opposition and the feedback it received when testing possible replacements, they will delay the discontinuation of third-party cookies until the end of 2023. What is your take on that? Paul Harrison: Well, I think the emphasis is more on the feedback or what they saw when they were testing replacements, much more so than what the mounting industry opposition was. The fact is, is that anytime you’re interacting with a browser, you need some sort of ID, even for basic functions, whether it’s frequency capping or basic identification, so that you’re not serving too many ads or stepping on ads or whatnot with a particular browser. When they introduced the notion of the FLoC ID and the FLoC specification, what a lot of us were able to do as we were going through the spec was we could just use it as yet another identifier. Others that were testing it had found ways to pair those FLoC IDs with some purchase information and browser history, which gave a much more complete picture of the user than third-party cookies were giving at the moment. It sort of defeated the entire purpose of what the exercise was about in eliminating the third-party cookies and replacing them with this FLoC ID. There’s a lot of back to the drawing board to figure out what the right path is forward. Moving the date out has more to do with to see if they can salvage the project than it does with any mounting industry opposition. David McBee: I see. Speaking of going back to the drawing board, I know we have a bit of a drawing board. You put together a timeline of data privacy changes that have occurred since, I think, 2016 and’s response to those changes. Would you be willing to share that with folks and give us a highlight of what’s on the list? Paul Harrison: Yeah. First of all, there are always changes in our industry. I think our industry is famous for that. There’ve been waves of innovation and you get winners and losers in the space. The winners tend to be very prepared and recognize that change is a constant in what we do, but it really goes back even before 2016. I mean, some of the first inklings of this were Safari having third-party cookies turned off by default. That was a little bit of a shakeup for folks. Then there was the introduction in 2017 of ITP. Apple seems to try to lead the effort in some of these privacy efforts. We also see quite a few from Google as well, but one usually signals to the other and then they implement their own version of that. We’ve also seen the introduction of legislative privacy, CCPA, GDPR, CPRA, which was renamed from CPREA. I guess they didn’t like the enforcement part of that acronym. Every time one of these types of … I call them opportunities, some people call them obstacles, are put in place, folks that have prepared for these kinds of events tend to fare very well. We have been preparing for these since the beginning. A few things that we have always done is, is that we took a hard position on mobile identifiers and we would one way hash those from the very beginning so that they really couldn’t be used and paired with other types of data. I recognize some advertisers would like to use it that way, but that’s the kind of behavior that legislators really don’t like and so they try to legislate the use of data and then of course the hardware and software manufacturers will follow up with their own draconian implementations. Each time the winners in our space have been faced with these opportunities, they’ve risen to the occasion. We designed our identity graph. We’ve designed our entire privacy policy. We’ve been a completely transparent company from day one, and so I think that has really fared well for us. I think we’re in a great position as we move forward, even with some of the outstanding legislative and future operating system changes that may be coming in the future. David McBee: You mentioned the identity graph. It’s my understanding that the biggest challenge presented by the elimination of cookies is a company’s ability to identify prospects for targeting. Can you tell me a little bit about’s identity solution? Again, I’m going to share a slide while you talk about that. Paul Harrison: Sure.’s identity graph from the beginning has been a foundational graph that built, but we built it in such a way that we grafted on other graphs. We have a meta graph or a graph of graphs. You hear a lot of the same buzzwords. We focus on how devices and apps are attached to households, and so we both have the ability to identify individual devices, individual apps used, and roll them up to a household. That’s a pretty common theme you hear, but not everybody has actually built it that way. Many folks just a rebrand what others are doing or they’re running entirely on somebody’s graph. But by having a graph of graphs and using machine learning in order to graph these things together to get a better view of the marketplace and identifying users, is really a critical step for those who are going to bridge the gap from the change in the identifiers. In point of fact, we’ve seen a proliferation of identifiers. We’ve also seen some identifiers being better for certain types of markets than others. For example, some identifiers work better with CTV, others work better with mobile, or others will work better on browser. You really have to have a strategy of looking at all of the available identifiers in order to make the best decisioning when you’re both bidding and doing attribution. David McBee: Is it fair to say from this image that this is like a current peek at our identity graph and that’s probably something that changes all the time? Paul Harrison: Yeah. That’s absolutely right. There are a lot of identifiers. This of course is just a sampling of the identifiers that we utilize or have been testing. FLoC ID, for example, is shown on there. That’s an ID that I don’t think is going to be long for this world. With the proliferation of IDs that are happening out there, because it’s such large market and there’s so much money to be made in identification, there’ll be plenty of these IDs to test, which really happened here is, is that with the changes in the operating systems, with the pending changes in browsers, with legislative changes, it means that nobody has a complete view of the environment. Everyone has a fragmented view, and so you really have to bring various IDs together in order to identify users, using machine learning to cluster these users together to get a better identification as you’re both bidding and doing your attribution. It’s going to change over time. One ID may work well for a particular market segment like CTV, another one may work better for browsers. It just depends on what your use cases are as to what combination of IDs you’re going to end up using. David McBee: You kind of answered the question I was going to ask about everyone having these same challenges, right? No one can see the whole web, so each of these identifiers sees part of it, am I understanding that correctly? Paul Harrison: Yeah. That’s right. If you think about it, we’ve seen a lot of articles written about, in particular, IDFA, so a mobile identifier for the Apple operating system. Depending on what the SDK footprint is, they may have seen a dramatic drop in identifiers, or they may have seen a modest drop. It just depends on what their install base is. For example, you may have an install base where it utilizes location data in a much more useful way. It could be a weather app. It could be some sort of navigation app. It could be some sort of tool that tells you what’s near you or who’s near you. Those types of apps tend to have high geolocation utilization patterns. Apps that have much lower utilization patterns for geolocation data may be using geolocation as an ancillary feature within the app. Those tend to either have less precision turned on, or the geolocation information gets turned off. That’s why you see such a disparity when you read one article to the next and you see such a broad set of quotes on what the signal loss was on geolocation information. It’s largely because of the segment of the marketplace that they see. David McBee: Okay. Now I know you can’t speak to every DSP that’s out there, but is it fair to say that our solution is unique? For example, do larger DSPs have better optics? Paul Harrison: I don’t know of any DSP that has better optics than we do. There are a lot of different solutions out there. There are a lot of graphs available. There are a lot of IDs available. I think the difference between the winners and losers in the marketplace are going to be those who can take those disparate IDs and stitch them together and cluster them together and identify the households using machine learning. I think those who don’t have a strong background in that area are going to have a tough time bridging the gap. I mean, one of the unique ways that we manage the data is, is we’re able to map it back to the households in unique ways, other than just IP, which is the typical way that a lot of these graphs are put together. Everyone’s going to have a little bit of secret sauce in their Identity graphs, especially the meta graphs that many of the DSPs or the more prominent ones are putting together. The real winners are going to be those who are taking advantage of the machine learning to really drive that opportunity. David McBee: Okay. Something you mentioned last time when we spoke was, well, you said there were more signals than ever. Can you explain that further? Paul Harrison: Yeah. There were the very common signals everybody would talk about. They would be mobile identifiers, like an IDFA or an Android ID or other types of advertising IDs, third-party cookie ID, so on and so forth. I think that what we’re seeing now is, is we’re seeing a lot of companies get into the game of identity and identity in a way that helps power the demand side of the advertising space. You’ve seen Experian in some of the moves that they’ve made. You’ve seen TransUnion in some of the moves that they’ve made. Those types of moves are indicative of our industry. There are a lot of players here in our space that are working on various identifiers. Last check we are at about 80. I’m sure that number has grown since last I looked, and the ways that people are identifying users continues to change. David McBee: Do you want to speak to these match rate and data signals real quick? Paul Harrison: Yeah. We’ve actually seen an overall increase in match rates. I think that’s one of the better ways of measuring this. There are lots of different ways of measuring it. You can measure things by how many GEO signals you see, how many mobile identifiers you see, things of that nature. At the end of the day, since you’re trying to target users and serve ads, I think one of the better metrics is just, what is your overall match rate? We’ve seen a steady increase in our match rate, especially since we’ve been testing and grafting on various other identifiers, and that type of approach is not just one dimensional, meaning it’s focused on the entire ecosystem. We actually are looking at it from various standpoints of, how does this work against CTV versus how does it work against a browser? Or, how does this work in some sort of a video type ad? We think about the various media types and the various signals and how they play off of each other to help us actually achieve higher match rates, and at the end of the day, better target users for our customer. David McBee: All right. If it’s okay with you, I’d like to get into the weeds just a bit. Can you tell me specifically how each of our tactics might be impacted by these upcoming changes? For example, will our targeting tactics be impeded by the changes? If yes, what does that mean for our campaigns as a whole? Paul Harrison: None of our current offerings are going to be impacted in any kind of significant way by any of the changes that have taken place or even the third-party cookie elimination. However, the biggest threat really in our space would be legislative. Let’s say that at the end of the day, government legislation comes down and says, “You can’t do any kind of behavioral targeting.” That’s really the bigger threat to the industry because there will always be some sort of identifiers available for identifying users and using the various techniques for behavioral targeting, unless a legislative change comes down. If one does, we have an extremely diversified product offering. There are a lot of articles floating around out there trying to scare people into the fact that, “Oh, we’re going to go back in time and we’re only going to be able to do contextual targeting.” Well, we do contextual targeting. We have our own contextual engine. We have our own keyword contextual. We have our own geo-targeting type mechanisms that don’t always have to rely on the lat-long information coming across. There are various ways that you can still do very impactful advertising, even in an apocalyptic scenario. That being said, I don’t think we’re going to get there. I think we’re in a great position to continue to offer the world-class products that we offer our clients. David McBee: Just so I am sure I heard that right. You’re prepared for the apocalypse. Paul Harrison: Well, yes. Of sorts. David McBee: We already made a lot of these changes, right? That you’ve talked about. What kind of results has seen so far? Paul Harrison: Yeah. As I mentioned earlier, we actually built this thing from the very beginning to be very resilient and allowed us to be nimble as changes came in our industry, because there are always changes coming in our industry. I mentioned a little bit about some of our match rates. We’ve seen an increase in the match rates as a result of some of the newer IDs that we’ve been bringing into our meta graph, as we see even additional ask because these IDs have to really reach critical mass in order for them to have enough value to graft demand. We’re working with the exchanges to continue to add various IDs, talking to publishers about certain IDs that we think will add value, and as we see those and we see critical mass, they’ll be included in the process for our overall meta graph. David McBee: All right. Now I want to shift gears to something that’s pretty important to our clients, which is of course reporting. How will these proposed changes, impact campaign attribution? Paul Harrison: I don’t see much of an impact in general on the attribution. I think that the areas where you will see some nominal effect is really on the very small conversion fences that can be set up in our system. The reason for that is, is that if you start seeing a lot more users using in-precise location information, that means that the bubble, in essence, that surrounds their location, can intersect with a small geo-fence or conversion fence in a way that makes it look like more users are showing up for attribution than are actually showing up in that small fence. What we’re probably going to end up doing is advising clients to focus on the trends, be looking at the lift metrics, because we spend a lot of time in our lift reports trying to cut out as much of that noise as possible. Instead of just looking at the raw attribution numbers, looking at your lifts, your trends, that’s where you’re going to see the value. David McBee: Do we need to reset our key performance indicators based on the changes and the amount of data that we’re seeing? Paul Harrison: Yeah. I think realistically there could be a slight increase in cost per view. It could be because of the attribution. It could be because of the filtering. It could be because of some of the in-precision information, but it’s not going to be anything significant. If it were going to be significant, we would put out some information on it, but the numbers look good. You have to remember, most of these changes have been in place now for several months and we haven’t seen any major shifts in any of the key KPIs. David McBee: All right. Well, I’m nearly done. I want to say thank you. I think this is awesome, but I do have one last question for you. With so much noise on the web about privacy and cookies and updates, the end of the world, right? How do you recommend advertisers cut through this clutter of news and what updates and information should they really focus on? Paul Harrison: When you’re reading the articles out there, you really have to look at what the source of the articles are. There are a lot of people that are writing articles for various publications, for a couple of reasons. One, it may drive more business to them. Example would be a contextual player in the space. Another is, is that shock headlines draw eyeballs and eyeballs draw advertisers. It’s a little bit of a self-fulfilling prophecy there. I think for the most part, we’ve seen many of the changes implemented haven’t had the kinds of impacts or the apocalyptic impacts that we saw in the articles. I think for the most part, it was much to do about nothing. From our standpoint, it’s business as usual. We’re going to continue to focus on these areas. We participate in the W3C pretty extensively actually. I’m on the board of the NAI. We are working with other organizations that are impacting our space. I think our webinars, we try to give you all the latest information as to where things are headed and what’s going on. I think that just put on the right lens while you’re reading through the various articles on the web. David McBee: Perfect. Thank you very much, Paul. Appreciate all your insight today. Paul Harrison: Yeah. Thank you. David McBee: Thank you guys for attending. If you are a current client and you have additional questions, please reach out to your account manager. If you’re a future client, you can visit us at or send an email to Thank you so much for joining us. I’m David McBee and we’ll see you next time.

Relevant Resources

We’re always exploring new ways of making deeper connections between advertisers and audiences.


Let's Connect

Get in touch and one of our advertising experts will reach out.