[music] [Lory Mishra] Hi, everybody. Good morning.

Thanks for being here bright and early. We appreciate it. Welcome to session 506, 5 Strategies to Execute Cookieless Marketing Now. Right now. My name is Lory Mishra. I'm a Product Marketer for Adobe Real-Time CDP. And I'm joined by an Adobe celebrity winner of the Experience Makers Award, David Barnes of US Bank.

- [David Barnes] Okay. - And his wonderful team that's in the front row. Thank you all for being here. Yeah, we all kind of won the award. So let's get this party started.

This is the year, at least we're being told, that third-party cookie apocalypse happens. We won't have cookies at the end of the year. We're being told. And actually, I'm really excited about this. Everyone in the room who's actually happy that we're going to be rid of third-party cookies in a few months, raise your hand. Who's excited? Yeah, I'm with you because I hate cookies. They get in the way. They're terrible identities. And I hate things lingering for years. Really annoying. Right? But there are other possible emotional responses to this. One is skepticism. I talked to a couple people on Monday who said it ain't happening. I mean, they're professionals in the industry. They live it just like I do. And they don't trust Chrome to finish this. That's one response. Another response is anxiety. And I'm guessing a lot of you have that to be up at 08:00 AM in the morning and here. I'm assuming there's some nervousness around this, that it doesn't make you happy.

So I'm going to ask you a question. So that's your personal response, but what's your organization's response? Is your organization ready for third-party cookie deprecations? I'm going to do another hand-raising exercise. No, not at all. You are in total denial, and there are some of you there, and I respect that. Somewhat. So you're talking about it, but you don't feel that everyone else in the org really cares enough. Anybody in the somewhat category? Okay. We got more hands. We're making progress. We're talking about it. We're not sure we're going to be ready when it happens, but we're talking about it. A few less hands. Now, who in here is confident they've got cookie mastered, that's under control, and they will be ready whenever Chrome gets its act together? All right. You need to be up on stage. Yeah. You can come up here and talk with us. So my experience. Sorry. My experience at US Bank was that we were more in category A and a half until one of the paid media platforms came to us, talked to one of our marketing leaders and said, your ad budget will decline in value by 45% if you don't figure out how to do conversion APIs using something other than cookie. And you have six months.

And all of a sudden we panicked and we entered B. So right now, I'd say we're B moving towards C as an organization. Now, I've been obsessing about this for years, but it's hard to get people to listen. And I know a lot of you probably had that cookie denialism challenge, but it sometimes helps to have an external influence. Some company comes to you and says the way you're doing it today, not moving forward with post-cookie strategies is going to cost you money. And that helps a lot actually. So let's go over some assumptions. And I already mentioned there's skepticism that this is actually going to happen this year. There are other assumptions that we're basing our position on, and I want to be transparent because some of you may disagree, you may have different assumptions. So here are our assumptions. First of all, the one I mentioned, Chrome is going to actually finish the job. They've started their 1% test. We think they're going to be done by the end of the year. But we could be wrong. We could be wrong. That's an assumption. The second one, no role for third-party cookies at all. And I've had an intense debate with people internally who believe you can mask a third-party domain, make it look like a first-party domain, still use third-party cookies. I personally don't believe that. But again, we won't know till it's done, which is another reason I want it done. Rip the band-aid off and I'll quit having this argument. Third, IP address is the new cookie. The freak out that happened like three years ago was, hey, we can't do remarketing anymore for unauthenticated site visitors after cookies are deprecated. So what are we going to do? And I think the industry coalesced pretty quickly around, we'll use IP address.

There are a lot of issues with that, but for now, we're going to assume IP address is the new cookie until Apple and Chrome do what they threatened to do, which is to finish masking IP address for this purpose. They've threatened that for years, particularly Apple. But they haven't gotten very far with it, as far as we can tell. And the final one is there's no industry consensus. And I'll mention, I'll talk about this later. I had hoped back in 2021, the IAB, somebody would force us all to come to an agreement and say, here's how we collectively, as an advertising industry, as an ad tech collective, are going to respond to post cookie. And in my opinion, that hadn't happened. It hadn't happened legally. It hadn't happened in terms of privacy. It hadn't happened in terms of even operation. Like, what id are you targeting on when you target on hashed email? Are you using RampID from LiveRamp, UID2 from Trade Desk, PAIR from Chrome, Google? It's your choice. And there's other IDs. So we don't have consensus. We're kind of still making it up.

So why does this matter? So cookies end. Who cares? The reason we think you should care is because there's an economic cost. There's a cost to your business of getting this wrong. In a study that Forrester did, they found that twenty-one cents of every media dollar is wasted because of poor data quality. And honestly, if you think about all the mistargeted ads that you get in a week, 20% probably feels kind of low. I'll tell you a quick story about something that happened to me just a couple days ago. So after work, I booked a nail appointment, and I booked a ride share ride to go to the salon because, you know, you got to look good for Adobe Summit. And I noticed that in the confirmation page for the Lyft ride, it now had an ad for me. And the ad wanted me to read a white paper from a marketing automation software company.

And I found myself feeling within a couple of seconds of seeing that, that this has nothing to do with me. I don't use marketing automation software. Why would I want to read a white paper at 07:00 PM after work? So I just made a snap judgment about that brand whose app I was using and also the brand that was trying to reach me. And the thing is, this wasn't cookie-based. It's a rideshare app. So they have my email address, they have my phone number. They even know where I live, and they got it wrong. So the point really isn't about identities. The point is about how do you make use of the data and the technology that you have available so that you can adapt to the world in front of us. And the world in front of us is really the high expectations we all have are the same high expectations our consumers have when we're trying to reach out to them. We literally have a couple seconds to get it right or risk getting written off. So in our session today, what David and I will talk you through is hopefully some strategies and execution tactics to answer this big question. And we have a lot of you and I imagine a lot of people who do different kind of jobs and different responsibilities at your organization. So what we'll try to do today is give something to all of you. So when I talk, what you'll hear from me is the Adobe perspective and the industry perspective and lessons that we've picked up from our customers as they have evolved into cookieless. Now, David, Experience Maker Award winner, has actually made progress on this. David has expertly adopted technology and worked inside his organization to bring this to life. So he's where the rubber meets the road. So David will tell you how to get it done. And at the end of the presentation, we actually have compiled documentation and resources. If you're really hyped up about this and you want to go into a room and start implementing, so hopefully we'll have something for everybody out of this session.

Okay. And to be totally fair, we haven't mastered a lot of this. We have implemented some and we are in the process of implementing or trying to implement some. So just to be fully transparent, we haven't solved everything either. So we're going to talk about five strategies. And these are really categories of strategies. None of these are definitive, meaning there are other possible use cases. There are other things you probably want to do, and we'll talk about that at the end too. But the first one is I mentioned earlier, we used to be very reliant on cookies for unauthenticated visitor marketing. Whether that's remarketing, personalizing on the website using like an audience manager third-party data connection. So how do I personalize content in a post-cookie world? Secondly, particularly for first-time visitors, the second category, we're moving to hashed email is the currency for paid media targeting. There are other options, but it's to me it's the dominant one. And we can execute today for known using AEP integrations to Facebook, whoever. We're going to talk about though, how to optimize that in paid media without cookies. How do you target your known customers in paid media? Some of the best ways possible. We'll give you an enhancement there. The third one is prospecting. Now I'll say, and I put it in quotes, a Customer Data Platform is not inherently for prospects, right? It's for customers. And so how do we use the CDP for getting new people in the door as customers, new customers. So those are three different targeting categories, types of audiences we would target. The fourth strategy we're going to talk about is one of the elephants in the room, which is consent. And I have a very North American point of view, and I'll quit touching my microphone, I have a very North American point of view. So I know some of you, many of you are not North American, but I'm going to talk through a North American use case and I'll mention an international use case, but my experience is North American. But how do you manage trust, how do you manage consent, particularly right now in California for us North Americans? And we'll go through that. The last one is this is a lot of change. We're trying to use the CDP to do a lot of really interesting things and how do we do that quickly and we're going to have some buzzword, I know this was actually clickbait so you guys would come here, but we threw AI in here and we're going to talk about how AI helps the other four strategies.

But it's real. Yeah, it's not fake. But we thought some of you would bite so maybe some of you did. - It worked. - Yeah, it worked. It worked. So engaging first-time visitors, this is a soapbox of mine. Whenever we talk about authenticating unauthenticated visitors and giving them a personalized experience, most of my business and marketing partners say, oh good, I want to know if they're in market for a credit card. I want to know if they're market for a loan, what their issues and needs are. So I can sell, sell, sell. But personally, I think personalizing for a pseudonymous unknown visitor, it's better just to give them something that feels relevant. How do I put a different experience in front of them so they feel like I kind of know them even if I don't know them. And so we'll give you an example here. An unauthenticated visitor shows up on your website and we're using, we're going to use the default Adobe Luma experience. And I assume many of you have seen this page in some kind of demo or things and this is the default experience. But let's say when that person shows up we take their browser characteristics, IP address and we reach out to an identity partner and we pull back some information about them from that third-party identity partner. And in this case, we're asking for sports and leisure interests and pet ownership. And again, that's not for sales, it's for engagement. So let's say it's returned that they like hiking, or probably should have said climbing or photography. And I changed the page so it's got this really dramatic, beautiful outdoor activity image, and that will feel more relevant to someone who is into hiking, climbing or photography. Or I ask, we get back that they're walking, running, and they're a dog owner. So we show them an image. They're all selling the same thing, they all lead to the same experience, but the image is targeted to that person, at least what our identity partner knows about that person's IP address.

So now Lory's going to tell you a bit more about how it can be implemented. And again, there's a connection in the back that gives you a lot more detail. Awesome. Thanks, David. So let me walk you through two high-level options you have to personalize to somebody who you don't know and is coming to your site for the very first time. The first option is pretty basic, and I imagine many of you are doing this today, which is you collect information when they land on your site from the browser signals, right? Maybe you now know a little bit about where they live and you use that with a tool like Adobe Target or some other personalization engine to change up their experience a little bit. If they're coming from California, maybe you don't want to show them snow or vice versa. The other option, which is the one that we're really excited about, is working with partners and Real-Time CDP in that mix. So here, like David was just describing, what you would do is in real time, call back to your partner, gather some information, and use that to personalize the experience a little bit more than the GEO or whatever browser signal you're getting. Where this really gets awesome is because you have a CDP profile in the mix, you can stitch together their unauthenticated and authenticated states and really start to understand them throughout their journey and what it took for you to convert them from somebody who just landed on your site that you didn't know to somebody that you now know. So let me show you how that would actually work with a little illustration I have. So in this example, we have an anonymous visitor, first-time visitor, land on our site. We don't know anything about them. We show them our homepage and we give them a generic experience. We then take the signals that we're picking up from that visit and ping it to a data partner of our choice. And when I say data partner, I mean companies like Acxiom, Merkle, Epsilon, et cetera. What the data partner then does is sends some info back. In this scenario, it tells us this is likely a millennial woman, and she's actually in market to buy a new home. And we combine those two pieces of information together in one bundle, and we start to understand who she might be a little bit better. Then we ping that to Real-Time CDP, where hopefully you have created your audience definitions and your segment definitions, and you have a segment created that says women in market for new homes, millennial, urban, et cetera. So this visitor qualifies into that segment. And what you're then able to do is push that to a personalization engine like Adobe Target, or whatever else that you're using, and that engine is able to change up the experience. So now we give her an experience that makes a little bit more sense to her, and the story has a happy ending. She loves it and she authenticates. So you learn a little bit more about her, and you collect information about her, maybe her email address, maybe her phone number or whatever else that you're asking for. And the cool thing, because you have the Real-Time CDP profile in the middle, that all gets stitched together from all of the other data points that you had collected previously, and now you have a full person that you know versus somebody random who landed on your website.

The awesome thing about this, as it says in my headline, all of this happens in 350 milliseconds or less. And I had to Google this, but that's one-third of one second. That's pretty awesome. And the stitching of the unknown to the known profile, just another couple of minutes. So the point of doing all of this is, of course, to see her as a person and start to understand the life stage she's in and the things that she cares about so you can engage more meaningfully with her outside of your website. Now, you know who she is, and maybe you can give her a consistent experience on social, email, whatever other channel that you want to reach out to. So if you think back to my nail appointment story, the tragic nail appointment story, if the data was used well, in that instance, maybe I would have gotten an ad for a beauty product, because that's sort of the mindset I was in. And they know how often I tend to go to a nail salon. They know what else I could be interested in by working with a data partner. So now that you have collected this valuable first-party data, which is hard to get, the second strategy that we want to talk to you about is how do you make the most out of that data, and how do you reach these customers that you know at a greater scale. So I'm going to start with another illustration here.

And we're going to start with the person that we met in the previous example. So here we have a profile that we've created in Real-Time CDP of our visitor that we now know. Let's call her Jane. And Jane has given us her email address, her phone number and a few other things. She's opted into sharing that with us. We have appropriate consent on that data. But like most of us, Jane probably has multiple devices. Maybe she has a work email address, maybe there's a shared device with her family that they all use. So what we can now do is work with an identity partner of your choice, whoever you like to work with, and bring them into the mix and augment the identities that you have on Jane. So now we have a little bit more information on Jane, like her secondary email, maybe some offline data about her, and maybe a household device ID. And again, now you know a little bit more about her and you can maximize the value of your first-party data by reaching her via more identities and more channels. And David actually did this at US Bank, so I'm going to hand it off to him to walk us through it. Okay, so like most of you, we have integrations from AEP to Facebook, Google Customer Match, Pinterest. And we've been doing that for a couple of years. And that's just an email or hashed email, depending on what your sequence is, integration. And it works. It works fine. But last year, Adobe approached us and said, we've got this LiveRamp Connect thing, and we know you're a LiveRamp partner, we think you'll be interested. And so we met with them, worked with them, and US Bank loads our entire marketable customer database to LiveRamp every month. And we've been doing this for years. Years. And we've been loading static lists to go with that, to then market to your Facebooks, your Googles, and your Pinterest, et cetera.

But that list contains name, address, phone number, email, and our internal key. All that's already built. Some group, other than the CDP team, other than our wonderful team here, is managing all of that because we've been doing it for years. The media team owns that. All we did was we're using the CDP to then send that same internal key, which is lower risk, by the way, not standard PII, but that key is sent to LiveRamp from the CDP. That's all we're sending. And then LiveRamp handles all those integrations, which is what it does best. So it's communicating with Facebook, Google Customer Match, Pinterest, LinkedIn and Trade Desk.

And that was a huge improvement for us. So what are the advantages of doing that? So the first advantage is, like I said, enhanced identity matching. You've got name, address, phone number and email, which is a much better match. Anyone will tell you that versus email, I don't know the exact number, but I think it's 30% to 40% lift by having everything versus just email. So that's huge. So there's your starting point. That's huge. But the second, the two things that really helped us, one is every time we had an integration between AEP and a bidding platform, we, the CDP team, owned the actual authentication credentials. And when the credentials changed, they often didn't communicate with us because the paid media team wasn't used to communicating with us. They owned it in LiveRamp, but they forgot to tell us. So it would break and then we'd go find them and we'd get the credentials fixed. And that happened more often than it should. And by using an identity partner, we don't have to worry about that, we, the CDP team. That's handled by the media team that does this every day. Then you've got privacy. We're a bank. And privacy information security, all those related risk compliance assessments is really important. And when you're sending. We're having debates right now, is hashed email and okay identity to send to anyone outside the bank. I think we're going to win that battle. We better or we're going to quit doing a lot of paid media.

Anything more than that? Absolutely not. But you don't even have to have that debate if you've already got this identity partner set up. All we're sending is an internal key which does not set all these-- The privacy information security doesn't set them off. So it makes it much safer and easier to get approvals if you're going through an identity partner that already has all that data through another means that's managed by another group in another way. The only real disadvantage of this is speed. Adding the identity partner adds an extra hop. It slows things down a bit. There's actually a little bit more to this if you want to wait a second for your snapshots.

So it's a little slower, it's an extra hop With LiveRamp, I think it typically takes us an extra day. So if we were using a direct integration, if you really need that paid media in front of the person within a day, you might want to do a direct integration, LiveRamp adds a day, typically, but that's the only real disadvantage for us. Everything else is so much easier if we go through the identity partner. But the biggest thing is money, right? So if we had a million people in an audience using LiveRamp, we get 2 to 3 million. And that's for two reasons, obviously. One is matching improvement, you just get better matches. Secondly, as Lory was saying earlier, we also know that this is the email they use to talk to us, but they have these other two emails that they use on a regular basis. And LiveRamp knows that we didn't know that. So LiveRamp will then send all three of those to the Facebook, your Google Customer Matches, your Pinterest, et cetera. So we found it extremely valuable, efficient and just kind of a major, really simple but major step forward.

Expanding audiences with partnerships. - Yeah. - Yeah, this one's yours. Sorry. All good. Well, awesome. That's great. I think the first two strategies we talked to you about, we're really grounded in using your centralized profile and audience that you have in Real-Time CDP, right. Like you've spent all this time and effort to bring the right data sources together and start building your actionable profile, the stuff that you want to market with. This third strategy here is how do you expand the reach and the impact of your audiences by working with strategic data partners.

And I'm going to talk to you a little bit about our roadmap today. So obligatory roadmap disclosure slide. Things may change, but regardless, we're really excited about our movement for this strategy. So let's again start with an example. You probably saw this coming. We have our same person that we engaged with when she landed on our website, we didn't know who she was. Now we know who she is. And in this example, she has booked travel with a travel company. So in that process, you collected some zero and first-party data. You know where she's going, what time her flight is, maybe you know her seat preferences and because of how often she flies, now you know what loyalty status she is for your airline, but you want to go beyond that because that information is useful for maybe a notification to remind her when her flight is. But it's not really great if you want to talk to her and engage her in other offers and other things that you have for her. So you decide to take that information and work with a strategic partner here. In this example, it's a connected TV publisher that you have. This connected TV publisher has information that she has shared with them with consent and you can use privacy-safe mechanisms to stitch your data together and really understand what she likes to watch and what time of day she tends to watch TV. Is she a sports fan? Does she like comedy? Whatever. The third piece is, you want to learn a little bit more about her, so you bring in your data partner into the mix, and you learn that she's in market for a new car. This person's having a big year. She's buying a house and a car all at once. My next example is her getting a puppy. So what the point is of this scenario is, again, you spent all this time creating your first-party profiles and audiences in CDP. How do you make the most out of that without recreating those audiences in those silos? Because today, maybe if you're working with a connected TV publisher and you're running advertising, maybe that's a totally separate team that's running that, and maybe that team is creating their own audiences, right? Maybe they're pinging the same data warehouse wherever you keep your golden record and starting brand new every time. The problem with that is that your consumer gets a totally disjointed experience because your teams and your technology are not connected. So to solve that problem, yesterday on mainstage, we announced Real-Time CDP Collaboration. This is our privacy-centric and marketer-friendly data collaboration application to help you get the most out of your data with useful second and eventually third-party data partnerships. Today, like I was saying, you might be working with publishers and other brands to get this done, and that's happening in different systems. Maybe you have separate clean rooms set up for every publisher that you're working with. Maybe you have separate mechanisms because everyone has their own preference for their technology. That's a lot of burden on your teams. That's a lot of burden on your business. And again, it's still kind of a poor experience for your customers. So what we want to do with collaboration is radically simplify how that happens today by centralizing all of this in one place. So with collaboration, we're bringing to market an agnostic technology. Regardless of the data, measurement, identity, and agency partners that you're working with, it'll be compatible with this tool. And the best thing is, if you're already an Adobe customer, especially a Real-Time CDP customer, you've created your audiences, right? Whether you're an advertiser or a publisher, you can tap into those audiences that you've already spent the time creating and use them to collaborate at an audience level in a privacy, safe, neutral environment with other parties that can augment your understanding, if you've been an Adobe customer, especially a Real-Time CDP customer, you might know that we've offered similar capabilities since 2021, called segment match. What we learned with segment match is that was a really good first step, and we were actually the first CDP in market to bring collaboration, because we've believed for a long time that it's important to augment your first-party data and that it's not sufficient. But what we learned with segment match is the world was changing, and we needed to think bigger and bring to market something more agnostic. That actually doesn't even require you to be a CDP customer to take advantage of.

So I'm going to hand it off to David because he has actually been exploring this and thinking about this for a while. So we implemented, US Bank implemented the CDP in the fall of 2011, and as soon as January 2022, our management wanted us to start acquiring new customers, brand new, not cross-sell, but brand new customers using the CDP. At the time, the CDP wasn't ready. AEP wasn't ready. The Adobe, I even had an Adobe consultant. One of the product managers say, we don't do prospecting use cases. That was Adobe's official position. We do not do prospecting use cases. But over the last several years, I think Adobe's gotten the message that that's pretty, pretty critical, that we all need prospecting use cases.

So first, we really need to, we all need to collectively move the CDP beyond just customer, the Customer Data Platform beyond customer.

I see, and if I'm missing any, you know, love for people to tell me, but I see three ways that you can use the AEP for prospecting today. One is they've added that prospect audience capability, where you can buy third-party data from your identity provider we've been talking about constantly, and load those records. You can load the audiences of third-party records, and it counts differently against your license, usage, and things like that. They protect it. It doesn't link automatically, so it's a brand-new way to load the data. I personally don't see the value of that. I'd rather target from the third party, and then once they show up or engage, then bring them into the CDP, I think that's just more effective. But Adobe has created this capability. The second one is what we've been doing for 20 years or however long audience manager's been around, which is an unauthenticated visitor shows up and we talk to them. Some of those are prospects. So that is a method of prospecting. The one we're talking about right now, which I've been really excited about for two years until we realized segment match had limits and didn't work for us, is to connect our data to a partner, a publisher, and I'm going to mention NBCUniversal in a second because that's the one that we were most interested in. But connect to that partner with our first-party data, connect to their first-party data, come up with a unified list of these are the US Bank seeds, those of us who do lookalike models, kind of a seed file that's connected into their ecosystem, and then they would use that data either with an Adobe look alike model, which is not enabled yet, or a NBCUniversal look-alike model. They would then find known to them, but unknown to us, individuals and market on our behalf. So we are working. So I kind of skipped through this one. So we are working, want to work directly with a partner to accomplish that. And we are putting together, Adobe has helped us even with segment match. They've come up with some solutions, I would call them rigs, but they have found ways to get past some of the issues we had. And we are talking about testing segment match now. Lory has told me that collaboration will be ready, what, June? Targeting H2? Yes. Okay, in H2. So if it happens fast enough, we may jump straight to collaboration and test this out. But right now we can test segment match with NBCUniversal. So we are very interested in doing that. And the initial test will just be, can we connect with segment match in this little. We cannot use email as an identity namespace because emails can be on multiple customers. Like a father and son might both have the same email on their account, but they're two different accounts and we can't link them. So we cannot use email. And that's been a big roadblock for us. Adobe has solved that. So they have a way to use even segment match until collaboration's ready so that we can use email as the link without making it an identity namespace and collapsing our profiles in our instance of AEP. So we're very excited about that and see a lot of value in it.

So those are the three customer targeting sets, right? We've talked about unauthenticated, we've talked about known, we've talked about potential, pure prospects unknown to us. So let's move into, and I actually really like this topic. I think some people are annoyed by it because it feels limiting to have a regulator tell us what to do. And I have a session this afternoon, by the way. Shameless plug. There's a session with me in one trust this afternoon. I think it's S730. We'll list it later. We'd love for anybody to come if they want to hear more about this particular issue and how post cookie impacts consent. But cookieless marketing requires post-cookie paid media consent management.

And this actually bit me really bad. And I'll tell you why in a second. But just to be clear, the AdTech industry, like I said before, has not reached consensus. We've been reacting, right? And this is one of the reasons this is such an annoying topic. We're not driving it. Instead, what has happened is we've had deceptive practices and we've had bad actors, Facebook, Cambridge Analytica, who have driven the conversation. And then Apple, Safari, Firefox and Chrome just started doing stuff on their own. They didn't consult us, the advertisers, right? They didn't ask us how we wanted to manage this. They just did it. And then finally, government started protecting their citizens. And again, they didn't ask us how we wanted them to do it. Well, maybe they did a little bit, but they didn't listen all that much. But you got GDPR, CCPA, and CPRA. And again, North American person here, GDPR is, I'll tell you why it impacted even me or us, but CPRA is where the rubber hit the road for us. And David, like, what was it like for you and your team at US Bank when all of these things were happening? Like one-off? Well, again, North American point of view here, January 1, 2023, CPRA became the law, right? You could not target paid media audiences in California unless you had an opt-out option. So fall of 2022, our legal privacy paid media teams were trying, were scrambling because, again, we're not way ahead of the curve on this one, but we were scrambling to be ready for that date. And we were focused on cookie. And the reason for that was most of our compliance legal teams were GDPR focused. We have bank, we have a subsidiary, Elavon Merchant, in Europe. So they were trained in GDPR. We have people who knew GDPR. Right, Theresa? And so, you know, it's a nail. They're a hammer. Bang, bang, bang. And they solved cookie. We solved cookie by January 1st, 2023. The problem was on January 2nd, 2023, we couldn't target anything in social in California because we didn't have an email opt-out. And to make that even worse, we bought a bank in California and merged with them in May and could not talk to these brand new customers in social because we didn't have an email option, opt-out option for CPRA. They did. The bank we bought did, but we couldn't integrate it fast enough. So we quit communicating with them in social and, you know, not ideal. So one of the things, so we can't wait for everyone else to get their act together and tell us how to do this, we've got to make sure that we're ready. And I've got four dimensions on here. First of all, you got to move away from thinking about cookie. Again, cookieless, that's our conversation here, is cookieless. But just imagine there are people in your institution that don't think cookieless. They're still in a GPR state of mind. They're still obsessed with cookie. You got to help them move on to email or PII because that's where the world's going.

And then because of that, you've got multiple identities here. So let's say IP address is the new cookie. Well, how do you do consent with IP address? If I want to do remarketing on IP address and I want to target someone based on their IP address shown up on my website, what's the consent? If I'm targeting in California, and I'm sure there are international equivalents, but again, North American point of view here, how do I manage consent? One idea is that I translate the IP address to an email and then I give them email opt-out. But we haven't gotten our legal and privacy to agree to that. But that's the talk track I'm using internally. Third, make sure that everyone in your organization is on the same page. So educate them. One of the things I learned a year and a half ago was most of my legal compliance privacy team did not respect email as a targeting mechanism and paid media. Again, they were cookie-obsessed. And we had a year of conversation before I even got them to pay attention. Make sure all your support teams, the teams you need to be on the same page with you to do your job, understand what you do, and understand this conversation. And finally, be ready for disruption. Be proactive. Don't wait. We waited. Not good for us. Don't do what we did.

Okay, so here we were. We had bought this bank. We had an extra million or so customers. We couldn't talk to them in social because we were behind the curve. So we talked to Adobe and we found out there was this one trust native integration, out-of-the-box integration. We then we already had a modal out there on our website that allowed cookie opt-outs. We updated it, so you can enter an email and you would enter the email and click opt-out. That would send the email through an API call to OneTrust. OneTrust would store that email and maintain the audit log so that legal and everybody's happy. Audit's happy. Then OneTrust, through a out-of-the-box integration, sent that data to AEP. We put it in the data lake. We didn't use dual. For whatever reason, we have not invested in mastering dual. So we just put it in the data lake. We ran some computed attributes, loaded an attribute to AEP that was, this person has opted out. This profile has an email that is opted out. So again, I said earlier, email could be on multiple profiles. If you opted out on an email, it would set both those profiles opted out. And once we had that, we created a suppression file. We sent that suppression file to every one of our bidding partners. And this is the end of this slide, so fire away. So we sent, and by the way, the presentation is going to be posted online, isn't it? It is, yeah. So you guys don't actually have to do that, but that's fine. - Yeah. - Yeah. But you know what, if you want to put this wherever. Yeah. Put it on LinkedIn, we're happy. That's fine. Except for the disclosure bit, the public-looking stuff. - That's right. - Don't do her bit for LinkedIn. Send it to your family. But, so we would send the suppression to the bidding platform. Now that had two advantages. One is we didn't have to control every single delivery to the site. And two, in an abundance of caution, if we actually built an audience in LinkedIn using rules in LinkedIn, we could still apply the opt-out. You know, at this point, we're treating the opt. There's not that many of them. We're not getting tens of thousands of people. We've probably gotten a few hundred in four months. So there's not that many people opting out on email, but if they do, we just won't send them any paid media advertising. It's just so much simpler. So we send a suppression file to the bidding platform and it's blocked by the media team, but we also remove them from all CDP audiences that are solely for paid media. And this, by the way, this took months, but entirely because we were debating what's the language, getting approvals. The actual implementation took a day to two days for OneTrust to AEP. Honest to God, we had it working within, I think a couple of hours, but within two days it was 100% done. And then the API to OneTrust took about the same amount of time. So the actual technical work here took us less than a week. It was all the stuff around it that took months. So we were very excited to be able to get over our delay problem. Yeah, and we were really glad. And this was sort of the reason to have an option like this for our customers who needed to move fast. So if you're already working with a partner like OneTrust where you're doing your consent and preference management, good news. There's out-of-the-box integration that David and team used to bring that in. The other thing David mentioned a couple of seconds ago was dual, and that is Adobe's patented data usage and labeling framework. And if you really want to kind of dig in deeper and if you have the time, I know you all had to move very quickly here, that framework lets you go down to the user level and the data set level channels audiences and set limitations on who can see the data that's in your system and who can access it and what channels is it allowed to be activated to. So say you have a data set that is only meant for social, you can set a policy in AEP, in Real-Time CDP, and it's automatically enforced if someone is trying to activate it to email, which they shouldn't be doing. So there's a deep framework here, and we're really, I think, aligned here because we take privacy very seriously because we're dealing here with PII, right. We're dealing here with customer data, and we're talking about doing things like working with partners with that customer data. So it's important to have those controls built in, and we have those controls available out of the box. And we also have integrations with partners like OneTrust, whatever makes sense for you and wherever you are in your cookieless journey in terms of privacy and consent, there are options here.

So we've gone through four kind of big things, and depending on where you are with your cookieless journey, you might be feeling a little bit overwhelmed of where do I start, how do I get this all done? And this is where we want to talk to you about utilizing GenAI capabilities and AI capabilities that are available in Adobe, in Experience Platform in Real-Time CDP, Journey Optimizer, so that you can get more done more quickly and you can really take advantage of the data that you have from your customers. So the first thing I'm going to talk to you about today is something that you saw demoed on mainstage yesterday. And we actually have a really great demo of this at the booth. So if you want to dig into it a little bit more, feel free to stop in. I think they have T-shirts and other swag, too. So with AI Assistant in Experience Platform, which is available for Real-Time CDP and Journey Optimizer customers, you can really get more done faster by using a natural language chat interface that's built right into the product. And in the alpha phase that we were in, we really saw customers lean into solving three types of problems. The one is increasing and supercharging their productivity. So this could be things like, you're new to platform, you don't know all of our language and all of our concepts. Maybe you ask a question like, how do I set up a schema for Experience events, for event data that I'm collecting in Adobe Analytics? AI Assistant can give you a clear, concise answer, which means you don't have to read through all of the documentation we have, and our documentation is thorough on purpose. We try to anticipate every question, but maybe you don't need all of that detail and you don't want to dig through all of that to get to the one thing that you need to do. The second thing that we saw our customers do in the alpha is democratize access of the platform and their customer data internally. So again, if you have somebody in your organization who's maybe a new hire, or they were using other tech before, and now they have to learn Experience Platform or Real-Time CDP, they can ask questions and learn quickly and get their jobs done a little bit faster. The third piece here is unleashing new ideas, which we'll be actively testing in the beta. And I have a QR code at the end of this. If you want to participate in the beta, we'd love to have as many of you as involved as possible to influence where we take this, but this is helping you answer questions that maybe let you imagine other possibilities for how you engage your customers. So maybe you want to know, I created a bunch of audiences a while ago, but which ones are my least used? Which ones did I sort of forget about? And maybe there's an opportunity for me to do something with them. So it's just finding those answers quickly so that you can get moving and get all of this stuff done that you have to. - And, David, you were in the alpha? - Yep. Can you tell us about your experience in the alpha? Yes. So US Bank participates in a lot of alphas and betas with Adobe. I kind of love telling them what I think of them, and so. We appreciate it. Yeah. Yeah. Well, more now than in the past. But anyway, so we were invited to be on the Assistant alpha, and to be honest, it was one I wasn't excited about. I was thinking, okay, I know AEP pretty well, and I know how to get around, and why would I need this? So we were a month or so in and confession, I cannot use Experience League. It doesn't matter what question I'm asking. When I type in a question in the browser for Experience League, I get 20 listings, and none of them are relevant. And I will constantly go to Adobe. Jamie's sitting over here laughing, our CSM. And I'll go to Adobe and say, look, you got to help me find this, because I cannot ask the right question. I am incapable of asking the right question. And so what I learned in Assistant early on was I could ask the exact same question in Assistant, and at least 70% of the time, even in the alpha, I mean, we're talking, it's not perfect yet. I would get the right answer. I would get the perfect answer. It was kind of scary. I mean, I don't know why they don't put that same capability into Experience League, but they haven't yet. But Assistant was amazing. It just, they've really added some strong natural language processing. The second thing it helped me with was we often submit tickets on things like, hey, why is our total storage consumption so high? And Adobe engineering, being engineers, would send us back a list of data set IDs, and they would have these long, random-looking strings. Look a lot like a cookie, actually. But anyway, these random-looking strings, and then they would have a total amount. Now, I don't know what the data set IDs are, and it wasn't a quick way for me to find them. I can go into Assistant and say, hey, tell me the data set name for this data set ID in quotes, and it pops it right up. So it has become, for me, an easy way to get to the data. Now, one other cool thing about both of those in the Experience League, not only did it give me the answer, it would give me links into where it got the answer. So the source material, but the one that really got me when I asked questions like the data set ID, or I would test questions like, how many segments has my team built in the last month? It would give me a query, and that query uses metadata. It's the same query tool that we all use to do whatever we're doing, but it would be query based on the metadata. And I didn't, I wouldn't, I'm not an expert in querying metadata within AAP. But now I can go in and modify these queries. It gives me the query. I can click into it, I can modify it and ask a slightly different question straight through the query. So it actually taught me a lot about how the metadata works by using the tool. I kind of fell in love with this. This is sort of now my favorite alpha because of that. So I went from skeptical to bit of an evangelist around Assistant, and Adobe keeps asking me to talk about it for some reason. - So anyway. But I highly encourage you. - I wonder why. Yeah, I can't imagine, I highly encourage you to be in the beta if you are interested. They're pushing it really hard. They're using it for reporting. There's now graphics.

I'm not using it for those purposes. So if anybody wants to really push those capabilities, I highly recommend you get involved in do your pushing. Yeah, no, that's awesome. I think in the examples that you gave, what's really cool about them is you describe something a marketer might do, a non-technical person. Like you need to know the names of the audiences because all you have the IDs. But then you also described knowing the query, which is maybe something a data engineer or data architect wants to do. I'm sorry to interrupt you. I've got one more example. - Yeah - This was one. I don't know if any of you have ever tried to do permissions, if you are an admin, but I'll sometimes hit a section of the screen or the UI and it'll say, you can't do this because you don't have this permission. And it uses a string I've never seen before. I have no idea what that string means. It's like manage dash thing dash, you don't have, you know, it's just really ridiculous. Well, I then put in Assistant, hey, what's the actual permission name for, in quotes, whatever it just showed me on the screen and it gives me the real permission so I can actually go out and add it to whoever needs it, myself or someone else. So it's those little things where the tool, the UI is not perfect. Assistant lets you get around them. That was the other use case. I forgot about this. You added a third Persona, right? You got a marketer, you got data engineer architecture, and now you have an admin. So we're trying to make this easier for everybody who interacts with platform and really make it more accessible, especially to those non-technical roles, so that you can really democratize the use of the customer data and profiles that you've spent the time building across your organization, across your channels and teams. The other thing here is AI is not new to Adobe. It's not new to Real-Time CDP. We have three capabilities that I want to talk to you about really quickly that help you get more out of your data right now. They're available out of the box right now. They're generally available. The first is look-alike expansions. We have out-of-the-box capability for you to expand your audiences and find more of your best customers using your first-party data. And it could also be first-party data that's enriched with information from your data partners. And the other cool thing about this model is in terms of transparency, it lets you know what the influencing factors were when we recommended a model to you so that you can kind of learn, like, what is making the difference here, and maybe use that information for the next time you build an audience or a campaign. The second is propensity scores. So you might have teams today of data scientists, engineers, more technical folks who are building your propensity scores, and you're trying to ingest that into AEP if you are or you're keeping it somewhere else and doing marketing with those propensity scores separately. We wanted to make that a lot easier. And with Customer AI, which is built into the product for Real-Time CDP, a marketer can push a few buttons and generate propensity scores based on the profile data and the event data that you have ingested into the system. The other cool thing is you can attach those scores to your profiles that you have in Real-Time CDP, and use that for segmentation. So you can create a segment of people who are likely to churn, or you can create maybe a segment of people who are close to converting because you want to make sure you give them two different messages. And for B2B customers, we have a similar capability to calculate propensity to convert, propensity for an account or a lead to convert with B2B lead and account scoring. So, again, the goal of all of this is to help you do everything you need to do a little faster, be less reliant on the rest of the organization and other tech, and really just keep your focus on your customer and the experiences you're trying to deliver to them.

So I think we are a little short on time, but-- I can blow through the takeaways -No, I think-- -if you want to skip through. We have a little bit of time after this between this and the keynotes, so let's go through the takeaways. - Okay. - Yeah. And for anybody who the clock's gonna ding and you turn into a pumpkin, there's a survey at the end, so please-- Do we need to do the survey real quick before people leave? I think they'll get a notification in the app. - So we would love-- - Yeah. So we would love feedback, love to hear from you, but we'll go through the takeaways. So the things that we recommend that we're asking you to do, be really clear what you want to do. What's the audience, again, unauthenticated, known to someone else, not you. What channels, it sometimes matters by channel. Are you doing.com? Are you doing an authenticated channel, et cetera? Be really clear about what you're trying to accomplish. Secondly, make sure your org is internally on the same page, basically that everyone speaks the same language, that you're aligned and there's not someone opposing you on a regular basis. Third, be real clear about who your tech partners are. And we've mentioned Axiom, Merkle, Epsilon, Neustar. There are others, LiveRamp. Adobe integrates with partners in sequence and the tool's evolving, so they have a priority. So they've done some things with Axiom that they haven't done with, let's say Neustar or Epsilon or Merkle, - if you have-- - ...for any real reason. Yeah, they were the ones that came first. They had a client that went to them and said, hey, we need this thing with Axiom. So tag extensions, a perfect example. We need this thing with Axiom. So they did it first because they had a customer that asked. It's real important that you identify what tech partners you want to integrate with AP. And then finally, that you communicate that to your Adobe team, that you communicate your use cases clearly so they can help you make sure there are many options here and they can be really clear. And secondly, make sure that Adobe is working with your tech partners, your identity partners in particular, and that's, yeah. Awesome. So like I mentioned, we have compiled resources for folks who are interested in documentation, videos, blog posts on how to get these strategies into action. This deck will be posted in the Summit portal so you can get all this content on all of these links I mentioned we have a QR code if you want to participate in the AI Assistant beta. So if you didn't get a chance to scan that yesterday at the keynote, please do. And actually, if you're an experienced platform customer now, you should be seeing a message in the product itself that is inviting you to sign up for the beta. We'd really love your feedback on that. Thanks, folks.