Spotify’s Adam Bly breaks down why understanding people through music matters to marketers.
Spotify’s research initiative to understand people through music has been spearheaded by Adam Bly, VP of Data. Before he joined Spotify, Bly founded the data innovation firm Seed Scientific, which used big data to solve complex problems of all kinds—improving road safety, optimizing vaccine deployment, and more. We sat down with Bly to find out what drew him to studying music, the value of streaming data, and what’s coming next for Spotify’s data mission. At Cannes Lions, he’ll dive even deeper into these insights with his presentation: Melody, Harmony and Metadata.
Interview has been edited for length and clarity.
Can you explain your initial vision for Seed Scientific, and what drew you to bring that approach to Spotify back in 2015?
Adam Bly: The thesis behind Seed was: let’s work with some of the biggest brands and organizations in the world to take on problems that, historically, you might have solved through the lens of design thinking or management consulting but wouldn’t have viewed as territory for data and science. We got working in the music domain as a similar type of problem space. It became really clear to me that Spotify’s data set—100 billion events a day over 10 years, and over 140 million monthly active users—is pretty unique in its potential to understand people. Then the question is: If the data seems pretty unique, what kind of value can we create for listeners, for artists, for Spotify, for marketers, based on that data?
We wanted to apply scientific rigor to how we investigate this problem, and also bring a level of empathy and creativity. Music is a very personal domain, it’s such a mirror of ourselves. So studying it begs a more human-centered approach.
Spotify's Adam Bly, VP of Data
Once you explored Spotify’s data set, how did you specifically develop the vision for Understanding People Through Music?
Like with any good science, it starts with observation. You start with the qualitative research—talking to people about how music figures into their day, how it modulates mood, how it connects people to one another. And then, quantitatively, we have extraordinary data, really unrivaled in the industry, because we’ve been at this for 10 years as a company.
The fact that Spotify is really a mobile-first product also means it’s with you in so many different contexts. When you combine the diversity of contexts, the richness of user engagement, the size of the audience, all the music in our catalog, and 10 years of data…if you buy into the argument that music is ripe territory for interesting science, then Spotify has the best data set to explore.
At Cannes, you’ll be demonstrating what this research can reveal about individual users. Can you give a hint of what we might expect?
So the first generation of user understanding for Spotify had been about knowing what people like at a very high level, in terms of their affinity for certain genres and artists. If you start from that premise—that we have best-in-class (and always-improving) understanding of music taste and now want to go deeper—then the next thing to look at is: How does that taste change in time and space?
One thing that’s really interesting to understand is how much a user forms routine. Are there times in the day, for a certain individual, where a certain kind of music, or a certain kind of listening, is really important to them? Are they the the kind of person who has a very specific weekday behavior, so Sunday is noticeably different than the rest of the week? For the first time, we’re really starting to see these musical journeys at an individual level.
How does the research benefit Spotify as a product? And how can it benefit brands and marketers?
Within the product, we are working to set the standard for state-of-the-art personalization. That doesn’t mean just getting a general sense of your taste spot on—it means being able to truly soundtrack your life in a profoundly meaningful way. This takes us to another level in making the product experience more relevant to our users.
Of course, as the product is more relevant to our users, it becomes ever more valuable to brands. They can use Spotify to interact with those users in moments that could be mutually beneficial to both the brand and the user. That’s sort of the Holy Grail of advertising.
What all of this data affords Spotify is the opportunity to create a level of ad personalization that is quite unprecedented. It stems from what we can learn about our users, and what we can express to brands anonymously, so they can craft the right strategy and creative, and present the right message to the right user at the right time. There’s a level of emotional significance we can offer to brands that goes beyond just, “I’m stepping in front of the grocery store, so I need food.” It’s much more complex. We can communicate with the user on a more fundamental emotional level.
You mentioned that deeper insights are coming in the future. Where do you think this research is headed next?
From a research standpoint, this is just the beginning. The more we demonstrate our ability to predict traits and attributes about people, and the more we demonstrate both internally and externally that our first-party data is uniquely rich and our methodologies are quite advanced, the more it excites us to go deeper. It pushes us to try to model that next unique attribute about people to make their connection to music and artists—and Spotify—ever more meaningful and relevant.