Episode 11: Personalizing Customer Experiences Through CDP Strategies

Join Rob Anderson, Principal CDP Strategist at Tealium, as he explores the world of Customer Data Platforms (CDPs) and their role in personalizing customer experiences. Rob shares insights on leveraging CDPs for use cases such as cart abandonment, website personalization, and healthcare marketing, while maintaining data governance and regulatory compliance. He highlights the importance of real-time data activation and how CDPs enable marketers to deliver targeted, personalized experiences across multiple channels. 

He speaks to the significance of understanding customer journeys and using CDP data to identify pain points and deliver relevant messages to guide customers towards their desired outcomes. Rob also discusses the future of CDP technology, including the exciting possibilities of Tealium’s Moments API and the upcoming feature that will allow multiple users to work simultaneously within the AudienceStream tool, enhancing collaboration and scalability for CDP teams.

Aaron: We’re here at Tealium DV, and I’m with Rob Anderson. Rob you are Principal CDP Strategist with Tealium is that right?  

Rob: Yes, yes. What that entails, so I get to work with all of our Tealium customers all throughout North America, working with them on their path to value with AudienceStream and figuring out how to use that that tool, essentially. One of the things that we understand is when you purchase AudienceStream and you purchase the CDP, it’s a big investment, and we want to make sure that you’re getting the value out of that investment compared to what you’re putting into it. Some customers need more help than others to realize that value, and so we have a team here within Tealium that is here to help customers on that path. That’s where we come in.  

Understanding CDPs

Aaron: For those who might not fully understand, have heard the acronym CDP again and again, but don’t maybe fully understand what a CDP is, can we start there? Can you define a CDP, at least the CDP as Tealium has created and implemented?  

Rob: Yeah. In Tealium terms, a customer data platform (CDP), that’s going to be where you’re housing data to act on that customer in the moment, based on what you know about them from a variety of data sources. A customer that’s interacting with you, they might be touching the brand based on channel, like the website or app. There might also be offline data. I work with customers all the time, and they’ve got data in a data warehouse, whether it be a cloud data warehouse or on premises storage. There’s also data that’s going to be generated through offline interactions, through brick and mortar, and through loyalty programs. All of this creates kind of a scattered, fragmented view of the customer. CDP is really designed to bring that all together and resolve all of those little breadcrumbs into a single holistic view of that individual. Then once we’ve got that, we want to be able to act in real time on that individual as they sit. If I have abandoned a website session, and I haven’t completed a purchase, if I’m on a travel website and haven’t completed a booking, we want to be able to take actions to get that visitor to return to the website to complete that journey. That’s where the last part of the CDP comes in, is it really needs to be that real time activation engine that’s able to take that customer profile and get it to where it needs to go so that we’re messaging the right thing to that user at the right time, in a consistent way across all of your different messaging channels, whether that be paid media, like paid search or paid social, or through in-app push notifications, or through just website personalization. It’s really kind of that full stack system that’s taking data from its stores, identifying that into that single view, and then taking that customer and saying, “Here’s what’s that next best action.” 

Aaron: Someone used a great metaphor earlier today, and they describe the difference between a CDP and a data warehouse as the difference between RAM and a hard drive, which makes a lot of sense for me. The CDP holds all this data in memory so that it can be actioned very quickly. It’s available almost instantaneously. Is that correct?  

Rob: One of the things we talk about with customers all the time is actually the CDP doesn’t need to actually hold all of your data. It just needs to hold enough data to be able to know what the next action is to take on that customer. I don’t necessarily need to know 1,000 different data points. I just need to know, “Here’s Aaron, he’s a cart abandoner, here’s what was in his cart at the time.” Then what is the next thing we do with a customer like that? The other 1,000 points of data that I collected on our customer, we don’t need that to determine what the best action is, so push that off to the data warehouse because we might use it later. We might be wanting to act on that data that’s in moment, in session – kind of RAM versus hard drive. When you’re loading data from your hard drive and into RAM, you don’t load the entire terabyte content of your hard drive in and then expect everything to work right. You load what’s needed to act for that application in that moment for what’s up on your screen, what tab you have open. That’s what we recommend, load that necessary subset that’s needed to take action. You can do that from a data warehouse. You can do that from real time events that are coming in. At the end of the day, you’ve got that that visitor profile and you’re able to act on that.  

Aaron: How does the CDP gather data? What are the inputs?  

Rob: If you talk to Jay Calavas, another great person to have on the podcast, he’s very fond of this analogy, which is, “If it has a chip and a wire, it can send data to Tealium.” For us, we integrate in all sorts of ways. Basically, if it has a live internet connection, we can probably figure out some way to get it into Tealium. If you’re on the website that’s through our native collect tag, if you’re in the app, we have our collect API’s or mobile SDKs. If you just have flat offline data, that’s data that you can load in just through flat files or through multiple different buckets that we provide to load that in. Then we’ve got things like data connect, so if you’ve got data that’s sitting, say in Snowflake in Salesforce, we can load that in in real time integration. There’s all sorts of ways to bring that data in. But at the end of the day, it’s really getting down to, “Can we distill what’s happening in that data source to an event?” If it’s offline data, we treat every row as an event. If it’s the SDK, your app interactions are going to be treated as single events. If it’s a website, we’re treating that data as single events. All of those events, in their totality, make up who that person is as an individual once we’ve resolved the identity. 

CDP Use Cases

Aaron: Okay. So, we’ve collected data, it’s in the CDP. How is that data actioned? You talked about the ability to react or act based upon what a visitor to a site does or doesn’t do. Describe that process and how a marketer would go about employing that process. 

Rob: What we think about when we’re going through this process is it’s really about the process of figuring out what your use case looks like for these individual actions that you want to take as a marketer. One of the things that we say all the time is a use case, when we’re talking about Tealium or we’re talking about any CDP, it’s really interesting because there’s no place in the AudienceStream UI, in Tealium CDP, where you can click or tap into that says, “Show me my use cases,” or “I’d like to build a new use case.” If we had that great, but it’s really not something that we’re trying to solve for. The way you solve for that is by thinking about what a use case is. What a use case is, it’s a specific customer journey that has a distinctive beginning, a middle, and an end activation. If you’re a marketer, and you’re thinking about, “How am I going to use CDPs?” Think about the ideal customer journeys or customer journeys interrupted, and what those look like for the different folks that you want to target. Once you’ve got that in mind, one of the things you can start to do is map that beginning part of the use case to the data sources. Where does that data originate that starts the journey? Does it happen on the website? Does it happen in the app? Does it happen when we do the nightly upload? Once you’ve got the beginning, the middle is, “What data gets stored in the visitor profile at each step of that journey?” Then finally, the end is, “What were we getting that visitor trying to do in the first place?” Were we trying to get them to convert or trying to get them to fill a lead form, trying to get them to book a trip? It might even are we trying to get them to schedule a visit in the store? If it’s a brick-and-mortar retail, it might be a big-ticket item that someone isn’t comfortable purchasing online. They’re just trying to schedule an in-store visit. How do we connect them with a representative at the store? That’s the beginning, the middle, and the end. At each step of that journey, there are things that you would configure an AudienceStream that reflect that journey. The beginning is your data sources, the middle are the audiences, the events, their resolving identity, and the end is your integrations. Where are you sending that data out – to paid media to paid search to paid social to email – to trigger that next step of the journey to keep them moving forward? 

Aaron: Let’s continue with that example of a serial cart abandoner. Let’s say they abandoned a cart one time. What then is the action that is taken, and how does the CDP invoke that action? 

Rob: Cart abandonment is really interesting as a use case because not all cart abandonment is created equal. Someone that abandoned the cart with a high value item in it, we wouldn’t just treat them as someone who came to the website, put an item in their cart and then left. We can do some things here to actually look at who is that visitor? How likely were they to make that purchase in the first place? Are they just window shopping or browsing? You’ve got a bunch of different audiences that aren’t just abandoners that you can set up out of a use case journey like this. Then the next step is okay, once I’ve got my different audiences in mind as they’re tied to this journey, what’s the next thing that I need to show them? For everyone that abandoned in that scenario, there was some kind of either journey interrupted or a pain point that kept them from doing the thing that we were trying to get them to do on the website. So what’s the pain? Is it just they were trying to find a lower price elsewhere? Were they just not ready to make that purchase today? Do they need to do more research? Do they have maybe some kind of objection? Are they worried that the product is not going to do what it claims it’s going to do? Answering those questions is going to tell you what the next step is for that customer. If it’s an objection, maybe I need to, from an activation standpoint, send them an email that says, and you wouldn’t say this overtly, but you would just from a marketing standpoint, it’s like, “Hey, you went down to the product reviews here, you spent 10 minutes scrolling the product reviews.” These are all things that we would track as part of cart abandonment. You sorted and you bought the negative reviews at the top and you spent five minutes in there. So, clearly you didn’t just abandon your card. You had an issue with this product, and if we could send you an email that would contain messaging that would not only resolve your concerns about the product, but also maybe gave you a discount code if you purchased today, would that be enough to convince you to move to the next step of the journey? We could try that as an experiment and see how many people move to that next step of the journey and go from there.

That’s the activation piece. What’s their pain point? What kept them from finishing? Can we hit them with a message in a channel that they’re likely to use? This is another thing that we look at in CDP, where I talk about this with use cases, for a given brand, they might have 10 different ways to message to a customer – email, paid media, push notification. Of those, which is the one that the customer actually spends the most time in is most likely to be receptive in terms of messaging. For a given customer, if they are on their phone 90% of the time, I don’t need to hit them with costly email marketing. Let’s use push notifications because they’re more receptive to that. The tap through rate on that is better for them. It’s not just one touch point or like we don’t look at abandoners all the same. It’s like, “Hey, this is an individual, they started a journey, and they encountered a pain point, can we get them a message to resolve that pain point?” Then, what’s the messaging that’s most likely to resonate with them in the channel that they most likely use? 

Aaron: Can that action be automated out of the CDP then? 

Rob: That’s the beauty of the CDP. All of those rules are automated. You don’t have to go into the CDP every 24 hours and say run a new batch or a new query or something like that. On users, you configure those rules in advance, you configure those audiences in advance. Then once that data is flowing into the CDP, really it’s about evaluation. It’s like looking at those results of things that are already happening in real time and then deciding, okay, this audience seems to be growing in a way that we’re comfortable with, we’re reaching the right amount of people. Let’s keep letting the CDP keep running them. When people set up connectors and that sort of stuff, that’s typically how they use it. They’ll set it up once, monitor it, if it looks like it’s performing well – I see people set that stuff up, and they won’t touch it for years – it’s like this is a campaign, it works really well for us as a base campaign, it’s reaching the right number of people, we’re not seeing churn out of this, great. What you also might see is the actions that you’ve set up aren’t performing well. Okay, we’ve made some assumptions about how we’re measuring engagement or scoring visitors on the website or how we want to set up this audience. Looks like based on the data that we’re seeing into the CDP that that’s not doing so great. We can pivot that now. We don’t have to wait 30 days to see how our campaign was performing. We can see that real time, like no one’s entering into this audience, looks like it’s either not configured right or we made some bad assumptions, or maybe the results of this A/B test are not coming up the way we want. Let’s stop that use case and let’s move to something else. So you get that flexibility there. 

Maintaining Data Governance

Aaron: You mentioned something critical there, which is the connectors to other platforms, the pipeline through which both you maintain data governance and you also then activate the action that you want as a result of the behavior or lack of behavior. 

Rob: There’s kind of two parts to this one. One is the governance piece. Governance is really interesting, because governance is actually something that, I think, should take place well before you get to the point of activation. Governance is really about that user giving either that explicit or implicit consent to use their data in the first place and how it should be used. There can certainly be all sorts of things that come into play with that, HIPAA, GDPR, and CCPA, all of these four letter acronyms that are going to define how you can even use and process and store that data in the first place. That takes place long before the connector action. So that’s upfront, like right at the data source. Then at the end of the day, once you’ve got that data, you can use those signals, not just the signals we’ve collected, but also the consent signals as well to decide whether or not you even are allowed to trigger a connector for a given visitor. Once you’ve got that, then it’s really up to you to decide what those connector actions should be and who gets to see that data based on those signals. 

CDPs in Healthcare Marketing

Aaron: You mentioned governance, which is critical. It’s a nice transition to talk about healthcare. Let’s talk about the role of a CDP in a healthcare or MedTech setting, where you’re dealing with a HIPAA-covered entity. What value does a CDP offer there that is unique for healthcare? 

Rob: For us, the CDP market in healthcare, it solves a problem that we’re really uniquely positioned to tackle, which is how can we provide that consent and control layer on top of an organization’s data that they are collecting in a first-party context? The problem that Wheelhouse has tried to solve for years is exactly that. You’ve got the updated OCR guidance. You’ve got HIPAA that defines not only what a covered entity is, but what data you’re allowed to collect that makes up that protected health information, that makes up a healthcare record. Step one is identifying at the collection level, what that data looks like, and setting up rules to make sure that data, if it’s protected, and we don’t want to ever activate on it, if it’s too risky to store that or send it forward, don’t even collect that in the first place. Second, once you’ve collected it, the big thing that a lot of organizations need to remember is with Tealium, we will sign a BAA, and that data, while it’s within Tealium, it’s still owned by the covered entity. You have not breached HIPAA, you have not shared it with a with a third-party. You haven’t done anything that compliance needs to worry about, yet. But, what you do have is that ability to when it is appropriate and allowed to be shared with that third-party entity, you can send it out in that controlled way, and you control what leaves the CDP as well. It’s really that nice command and control structure that sits on top of data. You get the option to decide what’s allowed to come in, when is it allowed to go out, and when it goes out, who gets it and what piece of that are they allowed to get?  

Website Personalization and Experience

Aaron: What role does AudienceStream, does a CDP play, in personalization of website experience? 

Rob: I’ve had so many great conversations at Digital Velocity this week around website personalization and experience because it’s one of the really interesting use cases that folks are using a CDP to solve for. We’ve had this conversation about what a CDP is and what data is in a CDP. It’s you, right? It’s data you as a visitor coming to a website, you as a visitor engaging in the app. It’s not just the data that you’ve generated from that single session. It’s everything about you. It’s data that you’ve got, that you’ve sent to us. It’s based on your preferences, how you’ve interacted with that brand in the past. It’s everything that someone that’s doing personalization needs to drive a personalized experience. One of the things that we’re heavily involved in, in terms of use cases – remember think about use cases as that beginning, middle, and end journey state –in personalization, it’s squarely one of those things where, if you’ve landed on the website, we already have all the data. The beginning and middle is already done. The end part is actually giving you that personalized experience. It’s how can we surface that data from the CDP fast, so that the personalization that needs to run for you as a visitor can do so immediately. On that first page as you come back to a website, or you might be in the middle of a session, and we learn something about you or we already know something about you, and before you leave your session, we should give you some kind of personalized experience based on what you’ve told us in-session and based on the totality of your visitor profile. There are so many use cases that get tied to CDPs and personalization because from a personalization experience with CDP, it’s the sweet spot. It has the data that personalized marketers need to drive those experiences. 

Aaron: So what does personalization look like in a healthcare setting? 

Rob: It’s a great question.  

Aaron: What can it look like?  

Rob: It’s an interesting one because once you get to the point where you’re looking at a personalized experience in a healthcare setting, depending on who you talk to, you are looking at what borderline is a customer healthcare record. To open the door to personalization there, it’s really what data would be needed to drive a personalized experience? Is that even the ideal personalized experience for a healthcare setting? How do we do that in a way that stays compliant and on board with regulatory requirements? 

Aaron: I think the other thing is not a regulatory issue. How do you do it in a way that remains comfortable for the user? 

Rob: So it doesn’t feel creepy. That’s exactly right. The creepiness factor is something that I talk about with customers all the time. It feels totally right and fair, in a retail environment, to personalize based on all the factors – things you’ve purchased in the past, where you’ve had them shipped, the price of those products, whether or not you’re likely to buy again. In healthcare, if we take the totality of how much you’ve spent at that hospital and the kind of procedures you’ve had in the past, and we start giving you personalized experiences based on that, it’s very disconcerting. We expect our doctor to know those things, but we didn’t expect the website to know those things. When you get to that point, when you kind of crossed that line, it feels creepy. It feels weird. It feels like data has been shared with people that it should not have been shared with. In healthcare, you have to have kind of a light touch on personalization. I think a lot of the personalization use cases, and I’ve talked a lot about this, especially in pharma. We’re doing a lot of use cases in pharma around personalization experiences, and it’s really around, is the person finding what they need? Because there’s all this information on healthcare websites. You look through data sheets and look through marketing material, especially in pharma because it’s not just about care, it’s about the marketing of the product, too. If you’re considering a product to maybe talk to your doctor about or something like that, there’s all these things you can read and look at and research on the website. It’s really trying to figure out, where is a person in their journey? Are they a current patient, are they brand new? Are they trying to find specific information or not? Are they looking, are they ready to have a conversation with a provider or not? At each step we want to figure out, can we get them to the next part of that journey and do it with a light touch? We don’t want to force them into healthcare. Healthcare is an elective for most people, in a non-emergency situation. It’s an elective thing. We don’t want to force you to make that purchase or force you to have this procedure or force you to take this drug, it’s your choice. We want to make sure it’s an informed choice. Those are the things that I think lends itself really well to personalization in healthcare. It’s not about giving them a discount code or an offer or hitting them with emails. It’s can we get you the information that you were looking for faster, so you can make the choice that you need to make to better your life? 

Unique CDP Uses

Aaron: Thinking about the conversations that you’ve had at Digital Velocity, but actually this year, what are a couple of the most interesting or novel uses of CDP? 

Rob: One of the things I can talk about is some of the unique ways I’ve seen customers work with us. With our strategy team, one of the things that we do as kind of our calling card, is we like to kind of meet that customer where they are in their Tealium journey and figure out what that ideal way of working with them looks like. For different customers, that kind of entails a different way of work, depending on where they are in their Tealium journey. For some customers, we have kind of a very prescriptive, “Here’s a series of workshops with distinct pre, during, and post activities that will work people through,” and then the outcomes from that become their use cases or becomes their visitor identity strategy. For some customers, and this is where it gets into the unique part, they’re moved into a center of excellence model where they’re kind of already beyond the basics. They don’t need a workshop. What they need is a consultant will come in and talk to them kind of and work with them side by side to solve a problem that they’re having, and it’s better for that to happen either in an in-person setting or at their office. I’ve been involved in a couple of things now where I’ve gone on-site with a client and their dev team. The mission statement is we’ve already done the pre work. We walk into the room with the use case in mind, and we know how we want to activate on this, we’ve got the journey, that’s done before anyone shows up in the room. Now it’s just okay, we’re all here. We have the Tealium resources on one side of the room, we have the vendor on the other side of the room, we have our dev team on the other side of the room. We’re going to stay in this room for four hours until it’s built. Let’s just all sit around the table and program this out, set up AudienceStream, set up the website, set up the vendor. In four hours, one of the things we’re able to do is do, what for a lot of people, would take six months of Zoom calls. Think about what’s required to actually go from a complete integration, from ideation all the way to the activation. It’s a lot of coordinating those people. That coordination, if you’re doing it in a remote first environment, takes a lot of time. You do the weekly status call for 30 minutes and you get a little bit done and then you have to leave it for next time. Then the next week rolls around, then you have to onboard someone new, and then they’ve got to get up to speed, and so six months later, you finally get to an activation. If you can just get the right players, in the right room, at the right time, with the right mission statement, and say “Hey, it’s not that hard to set up some of these use cases. If we just worked together for four hours, we could have this done” Those have been some really rewarding experiences around Tealium. We’ve assembled the players, we have the mission statement, we go in, we build it, and then I get to go home saying, “Man, today we did something that would have ordinarily, in a remote first environment, taken us six months or 18 months.” 

Aaron: Well, given what I know of you, I’m guessing that’s some of the most fun work you do.  

Rob: It should be the most fun work for anyone. We don’t exist to not create things. If we’re not creating, if we’re not building, then we’re just living. That’s no way to go.  

The Future of CDPs and AudienceStream

Aaron: A final question. What are you excited about for the future of AudienceStream and the future of CDP?  

Rob: There are two things that I’m most excited about, that have been talked about in this conference. One is very boring, and I’ll talk about that in a second, but it’s also the most exciting thing that we’ve been asking for years. The other is one that’s been talked about a lot, that I don’t think enough people have their heads wrapped around just yet about how they’re going to use it. That first thing that I just referenced, the thing that is very exciting and has gotten a lot of discussion here at the conference this week is Moments API. That really fast way to now query AudienceStream and get that visitor profile data out and then be able to activate on that in real time, in that moment, without waiting for data layer enrichment, without having to run it through a third-party. We can do all these first page personalization use cases that we couldn’t do before. We can activate in an AI context, providing AI engines contextualized information in real time, appending to a query. I just saw an amazing demo at the Beast Bar, where we’ve got an implementation engineer that has this use case fully stood up now, where he’s appending to a query in real time, to an LLM, to get a result, and it’s all in context. Really, really cool. Everyone’s excited about that. For what I’m going to work on in the next year, we’re going to see a lot of that work. I’ve already seen it for customers that are in early access for that feature. We have dozens of use cases that are just queued up waiting for people to start work on them at this point, so very, very exciting. Lots of hype.  

The second thing is the really boring thing, which is a limitation of AudienceStream, for years at this point, that people been asking for, forever. Can I have a work stream and AudienceStream where I actually have two people working in the tool simultaneously, able to make changes, without overwriting each other? When I worked at Wheelhouse and worked with the team there, we would constantly have situations like that where we were like, “Okay, I need to log out so that someone else can log in and work in there for a little bit. Then when they’re done, they can save their changes, and then I can log back in and start working again.” Just for a two-person team, that’s not the ideal workflow. When you try to do that at scale, we have teams of people that are AudienceStream customers that have tried to scale up AudienceStream, and then suddenly they run into a problem where, okay, we have now 20 developers, they all need to run campaigns this week. They all need to go set up those audiences and AudienceStream, and only one person can be in there at a time, that’s not the ideal deal workflow state. We’ve been asking for that change actually for years. This year, it is going to come out. The exciting thing is I’ve seen it demoed. I’ve seen it on the roadmap. It’s going to solve the pain points that a lot of people have with scaling up AudienceStream and expanding their team to support larger, more complex, weighty deployments of AudienceStream that really lend themselves to multiple people working in the tool at once. Not a sexy feature, but one that, I think, if you’re a user of AudienceStream, and you’re trying to scale it, I think it’s one that you should be really excited about. 

Aaron: Yeah, I agree. That’s great. Well, this has been a really good conversation. Very helpful.  

Rob: It’s always a good conversation talking with you. I’m so glad you came down this week and so glad you invited me. 

Aaron: It’s great to reconnect. Thanks, Rob.  

Rob: Absolutely. 

Questions or Comments

Please let us know if you have questions or comments about this episode by emailing Grace Johnson at grace@wheelhousedmg.com. Want to be a guest on a future episode? Fill out the Be a Guest form at the top of the Digital Clinic page to submit your inquiry.

Make sure to subscribe to the Digital Clinic on Spotify, Apple Podcasts, or Amazon Music to stay up to date on our weekly episodes.