In this episode, Dana Gardner, Principal Analyst at Interarbor Solutions is joined by Nick Winfrey, Vice President of Data Science and Data Strategy at The Walt Disney Company. They explore Disney's extensive use of AI, automation, and data science to optimize advertising and enhance fan experiences. The conversation also delves into the importance of diversity in data teams, the crucial role of AI and ML in media and entertainment, and how Disney leverages audience behavior and loyalty across different platforms.
In this episode, Dana Gardner, Principal Analyst at Interarbor Solutions is joined by Nick Winfrey, Vice President of Data Science and Data Strategy at The Walt Disney Company. They explore Disney's extensive use of AI, automation, and data science to optimize advertising and enhance fan experiences. The conversation also delves into the importance of diversity in data teams, the crucial role of AI and ML in media and entertainment, and how Disney leverages audience behavior and loyalty across different platforms.
0:00:06.7 Producer: Hello and welcome to The Data Cloud podcast. Today's episode features an interview with Nick Winfrey, Vice President of Data Science and Data Strategy at The Walt Disney Company, hosted by Dana Gardner, Principal Analyst at Interarbor Solutions. They explore Disney's extensive use of AI, automation, and data science to optimize advertising and enhance fan experiences. The conversation also delves into the importance of diversity in data teams, the crucial role of AI and ML in media and entertainment, and how Disney leverages audience behavior and loyalty across different platforms. Please enjoy this interview between Nick Winfrey and your host, Dana Gardner.
0:00:48.4 Dana Gardner: Welcome to The Data Cloud Podcast, Nick. We're delighted to have you with us.
0:00:52.1 Nick Winfrey: Thank you.
0:00:53.8 Dana Gardner: Few companies or brands have had the lineage, legacy, and success in multimedia storytelling than Disney. And for going on 100 years or more now, Disney has also been at the forefront of technology to not only bring their stories to life, but to also nurture their fans' lifelong loyalty by continuously enhancing their experiences. Now that AI, automation, and multi-channel data access are coming together to provide the best insights ever into creating compelling audience experiences, who better than Disney to take the pulse on what's possible and probable when it comes to media and entertainment innovation? In today's discussion, we'll do just that by exploring how Disney is bringing together all the elements to match the right messages with the right audiences at just the right time across multiple channels, brands, and mediums. Nick, tell us what Disney Advertising is, who your clients are, and how you deliver business value to them using advanced data sciences, their services.
0:01:57.0 Nick Winfrey: Yeah. Disney Advertising oversees all the advertising sales and integrated marketing for The Walt Disney Company. Entertainment and sports offers. When you think of linear, digital, social, audio, any of the ad-supported streaming businesses that would fall under our umbrella, which is a lengthy list, ABC, ABC News, Disney+, ESPN Networks, ESPN+, FX, Nat Geo, Hulu, and our eight local owned television stations, that is the breadth that we talk about with Disney Advertising. So when we talk about who we're trying to bring value for, we're talking about the thousands of brands that run across that media and across that contents. And we're really trying to make sure that when they're working with us, that we view them as part of those brands. We want to bring that quality brand experience that we think about the content we produce ourselves and make sure that that same quality brand experience is enabled for our advertisers so that overall we integrate it into that powerful customer experience.
0:02:58.3 Dana Gardner: And tell us, Nick, about your background and how you got to where you are and how that's helped shape your team and culture.
0:03:05.4 Nick Winfrey: Yeah, I have a background in economics, particularly econometrics. I studied at the University of San Diego, which is a university that put a heavy emphasis on the philosophy, the logic, attacking problems from a multidisciplinary way, really trying to remove silos and focus on the why. Out of school, I took a job at the Federal Reserve, which really put a heavy focus on that quantitative rigor, the importance also of the word choice. Markets move on the exact adjective or adverb that the chairman uses when he talks about the performance of the market. And that was really important for me to understand when you're working with a stakeholder, really know your audience, know how you're translating, know that the particular words that you speak to matter. After the Federal Reserve, I moved to the FBI. And that was really about how to make sure you are relevant, you're timely.
0:04:06.7 Nick Winfrey: It's great to do a robust analysis to really dig into something. But if you missed your avenue to be relevant, to drive a decision, then your analysis is never going to be as powerful as well as communicating on certainty and confidence in results. When you have to move fast, there are going to be times where you are giving on some of the accuracy and be able to explain where you're uncertain, what would make you more confident in your results. That all led to what I think of as a data strategy. So when I'm building a team, I'm looking for diversity and experiences, a bit of that data unicorn of someone that will dig into what the science, but also the why and the strategy and then able to communicate that to stakeholders. I think right now, as data science has really exploded, there's a heavy, almost too much emphasis on technical degrees. In my opinion, when I'm looking at experience, art history can be a technical degree. Philosophy can be a technical degree. Neuroscience can be a technical degree.
0:05:12.8 Nick Winfrey: It's about how you approach the problem, how you take that toolkit that you've been trained in, and how you apply it to problems. With AI really continuing to explode and be democratized, the understanding of why will be so key to the differentiating factor for a data team. For myself and my team, we're responsible for data science, data strategy, measurement strategy. We're responsible not only for developing those capabilities, but translating that into tactics and levers and decisions that enable our clients to have a better brand experience when they're engaging with Disney Advertising.
0:05:54.8 Dana Gardner: And when you're bringing together these different traits and experience levels and ability to analyze along with the new technology and data science, do you think that this is perhaps an inflection point or we're able to do things that we just couldn't have done before? What makes this an auspicious time for your team and the science as it's evolved?
0:06:16.6 Nick Winfrey: Yeah, I think kind of doubling down on that diversity and experience. We're in a situation right now where the AI and ML toolkits have matured tremendously. And so the personas and teams that leverage it have really expanded. The technical toolkit that you need to have has adapted. And so you can bring in people who approach the problems in a different way, and that really leads to new opportunities, new areas to innovate that get you out of that kind of groupthink, that box think. There's also such a heavy focus on first-party data right now, really going into what can be observed, moving away from some of the syndicated panel-based data, as well as everything operates in a lower-latency environment. I remember when I started out, it was like flashback to my day, I would take off a model that would take seven days to run, and you would code something wrong, and you would come back and sign in the virtual machine, and you would see red, red, red, red, red, red, red. And it was a painful experience as an analyst. That environment has changed so much where models run so much quicker. The latency, the time to get results has declined so much. You can get quicker results, and you can be more relevant. So personas focus on first-party data, and really just the time to be able to turn around those results.
0:07:50.1 Dana Gardner: And there's a great diversity, of course, across the Disney organization with different brands and media and channels and all sorts of different business lines. How are you following the fan, as I've heard you describe it, across these different aspects of your business, but bringing perhaps closer to a one-to-one relationship with each of your fans?
0:08:14.6 Nick Winfrey: Yeah, so the Follow the Fan concept is one in which you're not just a fan when you're engaging with that content. Fanship is an essence of who someone is. We'll talk about, you got the NBA playoffs going on right now, the NHL playoffs going on as well, and people will focus in on how do I reach that audience that is a Warriors fan or a Kings fan during the sporting event, but they continue to have those characteristics long after, that brand loyalty, that devotion, that local preferences. And that becomes part of the personas that we've built out. And so for us, then it is, how do we take those fanships, those insights about the audience's behavior and translate it into something that works for a brand and makes it actionable and real for them.
0:09:19.6 Dana Gardner: In order to do that, of course, you have to manage a lot of data complexity. We now have, as you mentioned, cutting-edge AI, increased automation, faster time to value from that analysis. How are you bringing these together in a way that enhances your business value and for your clients? How do we get from following the fan to the bottom line?
0:09:42.2 Nick Winfrey: Yeah, one of the areas that I start with when I hear that question is we are part of the business. Oftentimes, data teams see themselves too much as a central service. We are not a central service. We sit under advertising leadership. We sit with the sales teams. We hear the questions, the tactics, the levers that brands are trying to pull and make decisions around. And so we honestly flip a bit of that narrative on its head in that we don't look at the AI solutions, the data capabilities, and figure out how to drive business value. We're starting from how do we drive business value and what is data science role in that conversation? And so for us, solutions get built from client up. We're looking for those trends, those norms. Where can we... I have a phrase that I like to use with the team, which is scalable customization. Where can we take the building blocks that we have and put them together in a different way so that then we can align it to that business value that we already identified.
0:10:56.4 Dana Gardner: And how important is adding more automation, faster at scale, being repeatable? How is automation factoring into your ability to deliver this across all of your various business models?
0:11:10.5 Nick Winfrey: Yeah, automation, we have a goal of 75% automation by 2027 for our business. But it's always important to remember what is behind that. We want to remove the barriers to entry. We want to free up time for innovation. We want to get things into repeatable processes. If you've got people spending a ton of time on manual tasks, or what has become more of the norm with the focus on automation, which is what we refer to sometimes as swivel chair automation, where you automate something to hand it off to someone else who's automated something to hand it off, you just get into this pattern of repeating what you've always done, and you don't get that opportunity to try to innovate, try to find something outside of the norm and build a new capability.
0:12:06.9 Dana Gardner: Cool. Let's drill down a bit into how you got to where you are, your data enablement journey up until the present day. Tell us a little bit about some of your tools, maybe your proprietary tools that you use, and how you've been able to develop them.
0:12:22.0 Nick Winfrey: Yeah. Journeys always have multiple chapters, and I love to lean into analogies. The initial focus is on building the foundation, and the key pieces that you need to have in place. Just like a foundation for a house, you have to continue to invest in that foundation. For us, the foundation has always been the audience graph. I've been at Disney for over 10 years. We've been investing in the audience graph for more than a decade now, and we continue to invest in it. The audience graph is central to everything we do. It is the connection. It is the hubs. It's how we pull that Follow the Fan and all those different engagements into one central area. After you invest in a foundation and a data journey, we really then look at what are those interior spaces. Now that we have that foundation, we have that audience graph, how do we add rooms to it? How do we have the kitchen, the living room, the dining room? How do we make the different functions fleshed out at Disney Select when we talk about our audience segment offering, generation stream, our thought leadership on how individuals are engaging with streaming platforms, all the fan that we talked about before. It's making sense of the data.
0:13:43.5 Nick Winfrey: Once we have those two pieces in place, we have the house in place, we started thinking about what are new opportunities. We started thinking about externalization and the clean room. How do we expand the uses of the space? How do we make it so that Follow the Fan is not just us building the narrative from beginning to end? Follow the Fan is co-developing that narrative with our marketers so they can own part of that story. It was really creating that short-term rental for our house of letting someone come into a room for a period of time, understand what it looks like, and then build from there.
0:14:20.1 Dana Gardner: Any examples come to mind of when you've done all of that blocking and tackling properly, what it gets for you? Perhaps maybe matching the context of what's going on in the media and the content with what the next advertisement might be, that maybe there's a need for making it natural or a better experience.
0:14:39.2 Nick Winfrey: Yeah, we have a product called Disney Magic Words, which is really about leveraging AI capabilities against our content to provide what we refer to as hyper-contextualization of the content, really understanding the moods, the emotions, the interests in those scenes. That only gets you part of the way there. The rest of it then is understanding the brands and what type of audiences they're trying to tap into so that then you can get that creative that plays after the content and really seamlessly connect the pieces together. And the only way you can do that is by getting into a collaborative co-development style of model where you are speaking data systems to data systems.
0:15:34.3 Dana Gardner: Are there any other advertising experience enhancements that come to mind? What can we do now that we couldn't do before when it comes to making that experience of delivering the message about a product or a service in a way that benefits the advertiser but also perhaps is more effective to the audience?
0:15:55.7 Nick Winfrey: Yeah, my team overall, we're really focused on test and learn. And a lot of the push right now in the marketplace is this, how do you optimize to outcomes? And optimizing to outcomes really requires getting an understanding of what those outcomes are and having the models be able to react in an actionable, timely enough fashion so that you can lean into the learnings that you get from it. And you can maybe shift the narrative a little bit. Like on the measurement side for so long, it's been, you did a good job, pass. You got above a benchmark, pass. You were below a benchmark, fail. We're trying to switch that narrative a little bit and move it into a conversation of how can we optimize and move things up into the right? How can we continue to get that exponential growth in performance for brands? That would have not been possible at the scale it is today. Models, as I talked about earlier, like the latency of models was too long. The ability to run these complex models within the systems, the ability to connect systems was just limited. And so really with the AI data cloud and the AI, the cross cloud collaboration patterns, they've unlocked new capabilities.
0:17:32.1 Dana Gardner: Okay. Another aspect of building out the data, the infrastructure, finding those AI models that work in the way you want is in the ability to predict perhaps audience behaviors and preferences. Looking instead of towards what's working now, what might work later and having real insight into that rather than it being more of a guessing or perhaps an art rather than a science. What have we been able to do in order to elevate predicting audience behaviors and demands into a science?
0:18:06.8 Nick Winfrey: Yeah. Predicting anything is difficult. We all make jokes about the weather forecast, the economy, former chairman of the Federal Reserve, Alan Greenspan, used the term irrational exuberance. So difficult for predicting the market because there's a human element in it. When you start predicting audiences, it's all about the human element. So it's less about, in my opinion, necessarily having gotten to the singular process to predict better. But it's what I was talking about before is how rapidly can you test, learn, and iterate on those processes. How well can you get to a stable model that performs well against those predictions? Even going back to what I was talking about in my career path, how well can you also communicate uncertainty and confidence in your results so you don't lose stakeholder buy-in? You may say, this is the journey we're going to go down, but here's the decision tree. If something happens, it's different. We're going to change the model approach and we're going to go into a different direction. And just that pace at which we're able to test, learn, and iterate has greatly expanded. And we're able to take more shots at it in a narrower time. It's no longer running a model and waiting a week. It's running a model and waiting seven minutes and understanding how long you did it.
0:19:38.5 Dana Gardner: And what do you look for, Nick, in your suppliers, the people who are contributing, the components that you're pulling together in order to accomplish this and get to that level of being able to do things you just couldn't do before?
0:19:51.4 Nick Winfrey: Yeah, there's a few things. One, especially where there's so much easier cross-cloud collaboration, it's so much easier to connect systems together with concepts of dynamic tables or incremental ingest. I'm looking for someone that starts with the ends. Early on in data solutions, there was a lot of, what are the means to get to the end? So much focus on what is the architecture? How do we connect this architecture? I want to make sure we start with the solution and then we figure out how to get there. There's also an area where there's a balance of short-term and long-term goals. Anytime we're working with a new party, we definitely have to demonstrate the easy wins. We have to demonstrate the immediate value while continuing to focus on the North Star. Beyond that, we are going to challenge our parties to be adaptable and be transparent with us. Things become obsolete so fast. Approaches shift so quickly. We need an engaging party to call us out. And when we give them feedback that maybe their systems are on a 1.0 and they're going to move to 2.0, we need to have that transparent, agile working relationship.
0:21:31.7 Dana Gardner: And we've been talking primarily about looking at the audience and discerning inference and patterns and behaviors. But because measurement is in your title, I'm going to guess that you're also involved with measuring what's working inside the company. And so to what degree does automation and data contribute to your actually being able to know yourself and how well you're doing internally in order to then bring that, like I say, more to a science and less of an art?
0:22:04.5 Nick Winfrey: Yeah, any new capability you build, you've got to be willing to test it on your own internal areas first. It's how you're able to work through what happens when you pull one lever. What are those realm of possible paths that you can go down? When you're thinking about automation internally, Disney is a very complex ecosystem. I laid out all the brands that we're talking about, and in many ways, the internal teams are our stakeholders. They advertise on our platform, they try to reach their audiences on their platform, and we're able to see then they're trying to promote new content and they're running marketing media against it. We're able to see did that drive the tune-in that they're looking for, did that drive the engagement that they're looking for. And if not, we have more of an ability to understand the why, so that when we get in front of advertisers, we're already starting from a grounded position of a lot of tests, learn, and findings that we can bring to the table.
0:23:26.3 Dana Gardner: Great. Looking to the future a bit, can you talk about some of your projects? Maybe not in too much detail, but where do we go next with these new capabilities and the ability to see what's going on and working internally, working with more partners and stakeholders, and then delivering those end results to the fans and enhancing their experience? What's coming up on the horizon?
0:23:51.3 Nick Winfrey: Yeah, we announced in January a product called Disney Compass. It's an evolution of our data journey. We started this as an audience graph, put the rooms in the house, let people come into our house. This is almost a little bit to make a playoff of Disney content. This is almost our wayfinder moment. We are now taking that house and turning it into a houseboat, and we're following a North Star via Compass. And for us, what that's going to mean is we're going to refocus the narratives. We want, as I said, to shift away from this pass-fail, from a measurement capability. We want to move to an optimization-style approach, but we also want to move things out of silos. We want to get out of it in advertising. We talk about plan, activate, measure. We want to make all of these things kind of be self-reinforcing. We want to bring more transparency and consistency.
0:24:43.8 Nick Winfrey: So it's really going to be... The underwritten part of Compass is it's a challenge to ourselves, our vendors, and our clients to start thinking about automation as automation from beginning to end of the full flow, not just what touches the Disney part of the flow, but what touches from the very beginning when someone's thinking... An advertiser's thinking about investing in our platform to the end where they're seeing results and back again when they're trying to think about the audiences that they want to reach the next time or they want to message to again.
0:25:20.7 Nick Winfrey: We're challenging that automation to be from throughout the whole process and really move it towards an always-on framework. No longer is it... When you think about measurement, measurement study is done, four months later you're giving a readout of it, and you're not actual, you're not relevant, and so we want to move that into creating much more of a unified data platform.
0:25:46.1 Dana Gardner: And Nick, when it comes to the proper infrastructure, the blocking and tackling for your data in order to then take advantage of automation and speed and some of these higher order analytic capabilities, what do you look for and what are some of the necessary ingredients to get that right?
0:26:05.4 Nick Winfrey: Yeah, the AI data cloud has really unlocked some technologies and capabilities that allow for faster path automation. SQL native ML capabilities are one that we use, going back to those personas. It allows for more people to leverage the ML models, but then connect it to the rest of the work that they're doing. And so where before you had multiple handoffs, you've now got a singular person that can be looking at the ML model directly to insights. Dynamic tables is another one that we look at. Again, where before you had these files moving around, jobs processing the files, picking them up. Dynamic tables allow us to have the raw data in sync. And when raw data comes in incrementally, it lands there. It's immediately attached from ingest to insights. And so really in that case, the automation is done seamlessly by the Snowflake platform for us. We don't even have to think about it as much of a pipeline from raw ingest to insights.
0:27:14.7 Nick Winfrey: Similarly, when we think about streams with vendors or working on flows that go to other platforms, and the data is coming in at different points, before we have a lot of data restatements, we have a lot of incremental pipelines, and it was our heady lift to automation. With streams, those systems are connected, and we're decreasing the data processing and data restatements. A lot of these tools really have thoughtfully built automation into their capabilities instead of relying on the data teams to have to build them from scratch.
0:27:53.8 Dana Gardner: So the automation builds on itself, and you look to your suppliers to get that game well underway before you have to take over. Great.
0:28:03.3 Dana Gardner: How about digital advertising at writ large? How do you see AI improving, enhancing, and delivering perhaps new capabilities? What's AI going to bring to the digital advertising stage?
0:28:19.3 Nick Winfrey: Yeah, so going back to Disney Select, it's thousands of ML models that we run concurrently to really try to think of different ways to lump personas or lump audience engagement patterns together. That would have never been possible at scale 10 years ago. It just would not have been a scalable capability. When we think about what's going to happen within the digital advertising space, I think it's a similar thing of, there's a lot of proof of concepts around things that can be done in real time of understanding with sports, what is a really high-intensity moment and where fans are really engaged. And those types of models, again, would not have been possible a decade ago because of just the amount of time they were taking to process. And so the beautiful answer is I don't know, but I think the exciting part of that is things that were not scalable before are going to start coming as actionable techniques to be used.
0:29:35.3 Dana Gardner: I have to say that number of 75% automation as a goal is very impressive. And so my last question is going to be what words of advice do you have for other organizations that would also like to up their automation game? What are some of the basics that need to be in place? What do you recommend people do in order to put themselves in a better position to take advantage of automation?
0:30:00.4 Nick Winfrey: Yeah, the foundation is key. If you're automating off of an unstable foundation, it's going to collapse right away. Context is necessary, that building from the business solutions up so that you're not automating something that no one knows what the use case is for. Diversity of the data team is essential to get out of that box think of what the end solution looks like. But I think even more where we talk about we are part of sales, I think about broadening the definition of the data team and teams that are responsible for solution. We bring legal and privacy in early and often. So in that house analogy, instead of having an inspector come in at the last step, they're inspecting along the way so we don't have to tear down something if we've done it in a way that doesn't make sense. So you have to be automating with a purpose of the teams that are key to the end success being there early. That's also our go-to-market teams that we have who simplify our messages and make sure that again, we're automating around what would drive the business goals. And then our PR and communications team, I view as part of the data team. They're making sure that they're spotlighting our successes and getting us feedback on our shortcomings.
0:31:23.3 Nick Winfrey: We talk about automation as a data and tech capability and it's data and tech teams that sometimes get the credit for it, but you have to build a culture within the organization where everyone owns the solutions, which means everyone owns the wins and the losses. And as long as you get that cross-functional collaboration, moving towards automation is a lot easier than if you're trying to do it in the silo.
0:31:55.5 Dana Gardner: Well, great. Well, thanks so much to our latest Data Cloud podcast guest, guest rather Nick Winfrey. He's Vice President of Data Science and Data and Measurement Strategy at The Walt Disney Company. We so much appreciate you sharing your thoughts, experience, and expertise, Nick.
0:32:10.2 Nick Winfrey: Thank you.
0:32:13.7 Producer: Witness the future of data, AI, and apps at Snowflake Summit 2025. Join pioneers and industry leaders like Snowflake's Sridhar Ramaswamy and OpenAI's Sam Altman in San Francisco, June 2nd to 5th. Dive into 500-plus sessions, explore 190-plus partner solutions, and experience cutting-edge demos. Transform your career and organization. Register now and build the future with Snowflake at snowflake.com/summit.