During this week’s episode, Amy Chyan, Product to Product podcast producer and host, chatted with David Cutler, Product Director at Spotify.
David is a data-driven PM. He turns murky data into knowledge and insights for his teams. During this Recess break, David shared some of the features he’s launched at Spotify and how he uses data to make informed product decisions. He also shared how important he thinks data teams are to product orgs.
David shared a lot of great insights into using data to make product decisions. We highly recommend watching the full talk but if you’re tight on time, we’ve pulled out some highlights below.
(Highlights have been condensed and edited for clarity)
David’s path to product management (1:29)
Amy: Can you tell us a little bit more about yourself and how you got involved with product?
David: I’ve been at Spotify for almost three years at this point. So at the end of the summer, it’ll be my three year anniversary, but yeah, I’ve kind of bounced around in my career before I actually came to Spotify. I went to school in Washington, DC, and I’ve always been kind of a tech nerd at heart, I call myself, but I had a major in information technology, which is sort of a mix of business and computer science in school. And when I came out of school, I wasn’t entirely sure what I wanted to do. But Accenture was a consulting company that was popular in terms of software consulting and tech consulting. And they approached me and I sort of liked what I heard about the opportunity, and I joined Accenture for three years.
And the fun story I have about consulting is at least when I was working there, it was before the term of product manager was really popular. It wasn’t really known yet at the time, and when I think back to what my role was as part of Accenture for three years there, it really was what we think about being a product manager is today, right? So we went into different companies. Our role was essentially to understand what the employees and what the company was trying to achieve through technology and software. So they had a particular business challenge. They knew they needed to innovate and enhance their current processes and operations. And they knew tech was the way to do that, but they weren’t sure how to bridge those two things. And they relied on us as consultants to come in and understand what they were trying to achieve, what their metrics were, what are their KPIs? How do you understand how to be successful? And then we can offer different options of software that can really help them reach their goals, which essentially, if you’re going to boil down what a product manager does, it’s kind of it.
And from there, I moved into a data analytics company. I was there for around six years where I did a bit of consulting work, similar to what I did at Accenture, but also that’s where I got my feet wet and to kind of more pure product management, where we’re using a set of tools and products and capabilities to build more of a platform product for our customers, that then we took to market and actually sold as a separate product. And that is where it sort of clicked for me as I really enjoy the concept of product management. I love innovating. I love creating, but I also love collaborating with not only the user, but also the engineering team and others that are involved in actually developing that.
So from there, I went to Bloomberg where I worked on a product that built out as a data analytics platform that was servicing compliance organizations within financial services firms, where we were, I’m not going to say fraud detection, but we were trying to get ahead of some of the illegitimate, I would say, behavior that traders would sometimes engage in. So insider trading and front running and all that fun stuff that people think are cool until the SEC comes knocking at your door. But that again was kind of your typical product management type of role. We had a significantly sized product team and then a big engineering group. And we had to work together with them plus our clients to build the right products.
And then from there, I had a couple of opportunities I was considering, and I realized one of the things I haven’t done yet in my career is work on a product that had millions of end users, right? And I think there’s a fundamental difference in building products for millions versus hundreds or thousands or even teens, right? Depending on what type of product. So I wanted to move into that B to C world, a business to consumer world, rather than the B to B that I was in before.
David’s role at Spotify (7:44)
David: In terms of my particular role, so Spotify, a large company, a big part of our employee base is the engineering organization. And the engineering organization has multiple different competencies, right? You have product management, obviously, we have 250 plus PMs at this point, we have data science, we have machine learning engineers, we have data engineers, infrastructure, app designers, and essentially the engineering organization as a whole is a partnership of all of these different competencies that are coming together to really build different features, products, capabilities. And my particular role as a Product Director is I am responsible for leading the product strategy or, not necessarily the roadmap, but more the direction and the strategy of a particular product area, which is a collection of teams that are coming together for a common goal.
But I’m also responsible for helping to support product managers on my team, their particular career, make sure they’re aligned, try to make sure that it’s a healthy environment for PM’s to work and grow and innovate. And a lot of my role is making sure I understand what their interests are and what their passions are, but also how that fits with what other products within the company are being built out, so that we can provide the greatest experience, full stack, right? It starts out with infrastructure, and then data, and then the app itself, and then that experience that users have with Spotify.
Importance of data teams to product (9:22)
Amy: I mean, you’re touching a lot on analytics, data science, machine learning, and Spotify obviously uses that to understand user behavior. We get the year end list, “This is your most listened to, these are top 10,” or whatever. How important, then, is data teams to product works?
David: Yeah. So this is something I learned throughout my career and through different paths that I took and different opportunities I had working with different companies, which, again, was actually an amazing part of consulting, is you get to learn what are the needs and considerations and interests of different companies in different parts of the world, which gives you a better holistic perspective of what people really care about. But as I kind of went through my career, each year, I noticed more and more that data itself was becoming more of a primary asset of a company that could help them make multiple different types of decisions, whether these are big business decisions of, “Should I acquire this company? Should I make this content exclusive?” Particular feature decisions of even really granular stuff, like, “Should I put a sleep timer on my app itself? Or should I offer podcasts in the primary app? Or should I build a separate app? I have different user cohorts. Should I create a new app for that, or just allow that to go within the flow of the primary application?”
And what I really noticed and what I’ve seen, especially at Spotify, is data has become the most important asset that a company can actually maintain and develop to help through all of these processes within a decision making framework, right? So you mentioned user behavior. That’s a very big part of one of the things that my team does on a day to day basis, which is manage a handful of data sets that really look at user behavior, and not for anything much more than understanding what is that experience for our users? What do they like to listen to? What are some trends and kinds of behavioral signals that allow us to understand how to provide them a better experience?
And that is actually a good example, where if you’re going to consider data as a primary asset and a primary product, then you need to respect that and hire teams that focus just on data, right? And this didn’t exist, I would say five plus years ago, in a lot of the companies that I worked with. It was always a commoditized component of what you’re building and business intelligence teams started being built out, and reporting was a big part of it. And then analytics came into play and then more predictive analytics as you work your way up that chain, which became important, but essentially, I’ve never seen data used the way it is now. And even if you look at other tech companies and you look at some of the roles that they’re offering in some of their career sites, there’s a lot of data science, data product managers, data infrastructure roles, because companies are really focusing on building out teams for the sole reason of maintaining data as an asset.
Can data lead PMs astray? (12:41)
Amy: Do you think data can actually lead PMs astray if they are slicing the data the way they want to see it?
David: Yeah. So we talked about that a lot, is yes, we want to be a data driven company, right? We want to make sure that we’re using data in any decision that we’re making and help basically create evidence or backup for arguments. Something that I always think about and talk to my team about is I would rather see us become data informed, where we use data as one of the input mechanisms that help us make a decision. So it’s almost unfair to product managers, if we say we’re just going to use the data to help us make a decision, and it will tell us we don’t have to worry about a thing. It almost commoditizes us as product managers. I think the value that we bring is we can offer contextual evidence or a story to what the data is telling us. And yes, there is a risk that we could spin that in our own way to try to drive whatever decision we think should be made. But there’s a bit of social responsibility there and trust that you have.
So one of the things we talked about as a team is, yes, there’s quantitative evidence and analysis that we need to consider. There are certain KPIs that we want to define and track progress against and see if we’re successful. But there’s also a qualitative component to what are we trying to achieve here, and do those line up? And that’s what helps us reach a decision or reach a way to figure out what direction we want to go in. So I always say it’s kind of a mix of quantitative and qualitative, but again, I like the term, “data informed” rather than, “data driven.”
Features David has launched at Spotify and how the data informed them (14:36)
Amy: Can you talk about some of the features you helped launch at Spotify in a product lead role? I mean, in our pre-interview you mentioned Spotify for kids, how did you come to the conclusion that you needed to launch that, and what was sort of the data or the dashboard that your team’s like, “This is it, we need Spotify for kids.” And now it’s like the Spotify duo as well?
David: Yeah. So one of the advantages of being in the data world within a company like Spotify, is that you have many users and there’s kind of a broad cohort that you can look at, we’re relatively horizontal. So a data team within an organization, they’re not necessarily making the decision of, “Yes, we are going to launch Spotify for kids.” I think there’s a lot of different teams that are involved in that. I would do a disservice to other people if I said that our team was one who did it, it’s more of, when you’re looking at some of the core data sets within your company of user behavior, for example, one of the inputs you would actually look at to say, “Do we even need Spotify for kids?” Right? “Do we need Spotify for artists,” which is an app that artists have access to, to understand the behavior of their fans. And I mentioned before, do you need to create your own podcast app or just build it in?
A lot of times you do start with the data that is created based on the behavior of the user. So one of the things that happens with Spotify for kids is one of the use cases where parents would use Spotify to entertain their children and they would use their own profile, and they would play children’s music whether it’s at the dinner table or just to keep them occupied, which is what my sister does. But what ultimately happened is, yes, that was great for the parents, but we also noticed that it started to, I don’t want to say balloon, but intermix the children’s music with their own music that they like to listen to on their own time. And if a big part of our value prop is our suggestions and our playlists, like Discover Weekly, then we started to devalue those products that we are offering.
And so you can actually look at the data to see, “Well, is the user behavior that’s occurring, actually having an impact on other data products that we are providing that gets surfaced in the app itself?” So not only do we look at what is our user base, what do they look like in terms of location, geography, what are their tendencies and their behaviors? And then we look at what’s the impact that it has of not creating a separate app. And so then you can actually experiment and test. You could say, “Well, if we were able to keep this separate, how much more improved would their Discover Weekly be, and how much more of a better experience would it be for the kids themselves?” so it’s hard to say that any one team is the one responsible for launching, but again, being part of the data teams, you get exposure to all of these different initiatives that are happening across the company, whether it’s something for the artists, whether it’s something for the kids. Whether we launched Spotify stations, which was more of an experience for somebody who wanted to interact with Spotify more like a radio. I’m not going to say that’s my parents, but my parents love it. I said, “Hey, dad, here’s stations. And press a couple buttons.”
And you’re always looking at how the data can inform you, of if you should actually go in a particular direction, such as launching a separate app, or maybe launching a new feature within the app itself. Not to mention when you do launch a new product or feature, you could determine if it’s being utilized efficiently, and if people are in favor of it, just by looking at the interaction data itself on a day to day basis. And so there’s constant back and forth and input, and there’s a hypothesis that you make, and then you see the results of that experiment. And that all comes from the core data infrastructure and data feed that a lot of product managers at Spotify are involved with, right?
A lot of our role is understanding who are the different stakeholders in the company that care about this data? And usually you learn pretty quickly that there’s hundreds of use cases, hundreds of people in teams that really become super important stakeholders, too. And it’s a weird way to think about it because when you join Spotify, you think, “My end user is the end consumer, right? The person who downloads Spotify and they’re listening to music and I’m building a product for them,” but you quickly learn a lot of your primary stakeholders are actually other product teams within the company. And that’s also where your value as a PM comes in, you need to understand what are different initiatives? What are dependencies that you have on them? What do they have on you? Who are the different roles of individuals that we care about serving as a product team that has data as our primary product?
(20:05) Is the data team set up like a shared service at Spotify?
(22:08) Can you give an example of a feature that you were surprised was priority based on user data, or conversely, a feature you guys launched and you’re just like, “Let’s sunset this. They don’t like it at all.”
(27:50) What competencies would you recommend current PMs develop in the next 10 years?
David is a Product Director at Spotify and resides in Brooklyn (in non-quarantine times). He works with various teams across the organization that use data and insights to understand how user behavior and App feature/functionality can be attributed to the performance of Spotify as a product and company. He previously worked at Bloomberg, Teradata and Accenture in various product management roles, many of which focused on how data can be used as a competitive and operational advantage. David graduated George Washington University with a dual major of Information Technology and Finance, and has picked up various hobbies that are solely focused on keeping busy over the previous 4 months.
Amy’s a Content Marketing Specialist at Roadmunk on the Marketing team. She produces Recess, the Product to Product podcast and video content. Prior to Roadmunk, Amy worked as a journalist in various Canadian newsrooms and wrote for publications like NBC, CBC, Vice and more.