Ever wonder what goes into your favorite Spotify workout playlist? Who creates "Weekend Warrior"? How songs are selected for "Pumping Iron"? And what really happens at Spotify’s headquarters, anyway?
We spoke with Doug Ford, Spotify’s director of product and editorial head of curation, to get a behind-the-scenes look into the inspiration, technoloy, and manpower generating the music that powers your workout.
MF: So, what does the Head of Curation actually do?
DF: I run a team of 32 (and rapidly growing) curators from around the world whose primary jobs are to formulate hypotheses and put together playlists around different contexts—working out being a huge one for us. These people have different musical strengths, and all we do is work on playlisting. We try to make life better for people.
MF: What’s the inspiration behind some of these workout playlists?
DF: We put a lot of thought into the playlists instead of basing them solely off genres, album releases, or artists. When people get up in the morning, they're not—for the most part—thinking about what the latest release is; they’re thinking about their relationships, their health, their jobs, their bank accounts, getting to work on time, and the politics of what’s going on with their boss today. They want to chill, have romance, and music for their parties. So we have a need-for-life approach. When it comes to “workout,” we started seeing things to support that. We know, on the category side, that lots of people make their own playlists called, “Running,” or “Yoga,” or “Weightlifting.” We know there’s a demand for that, and we know from informal and formal questioning of people worldwide that they love listening to music when they work out.
MF: How many people come up with one playlist? Is it a team? Or do you have a software selecting songs?
DF: We’re a human curation team. Largely, it’s an individual overseeing a single playlist, but in some cases, we have a whole team of people. We have experts with heavy experience in culture and programming certain genres like EDM, rock, and R&B. We’ll lean on them when we want to have tempo-matched music across a genre. With the “running” playlist, that was interesting because that was a team of people who did a lot of research and polling. We even have an internal team of people running to music, just checking what and who the motivators were. We call them the most fit product team in Spotify. We know songs need to have a tempo people can match, and we also found people were interested in defeating the demons of running. They wanted music, not just based on performance, but they wanted music to eliminate the tedium of running, and to keep them going for longer sessions, and we found that people need peaks. They want to cool down and then ramp up to another peak.
MF: How do age, geography, and other demographic variables come into play?
DF: We have consumption habits based on age and region. [The] era of music plays into the age demo, and local artists play into the cultural aspect of region, so we have someone doing French and Asian curation, someone in Stockholm, someone in New York. We have people all over the world, because we believe in that cultural and linguistic relevance. And that goes for workouts, too. We don’t like to make a 50-year-old person feel like they’re old. You don’t give someone 70s music just because of their age. It’s such a personal thing. Some people want to listen to music they love, either to increase their performance or take their mind off it. Some people will listen to music that they normally wouldn’t listen to in any other context, just because it helps them work out. Whether it’s tempo-based, whether it’s aggression-based, or situation-based, people will always find music they will adapt to their context.
MF: Are there hypothetical listeners you try to target with the playlist titles and images?
DF: We like to think that each playlist has a very specific hypothesis. We come up with the idea, the concept, and we run it home with an image and a description, and a title. So with “Pumping Iron,” there’s no mistaking what that is, right? It’s aggressive rock. Probably, I would say, male-geared, and when you push play on the first song, it’s going to be one of these total pump-up playlists.
MF: How important is the first song of the playlist, and its sequencing?
DF: We have a cycle we like to put together and we learn through consumption what people do, what their main tempo is, their pace, and what they like about the peaks. Skips are a huge indicator for us. An initial playlist, especially for workouts, is huge about giving a cue in the first song. We need to stamp that in somebody’s mind right away because you only get 5-10 seconds to catch them. Transitions are huge in playlisting. So first song, the sequence of what goes into the second song, the pace of the transition, the spaces. These things are all critical to us.
MF: How does the creative process meld with the raw data?
DF: We’re getting constant feedback from users, and we use machines to pool together a batch of songs that have a tempo and beats per minute that match a certain pace. Then we determine what’s the clickiness that defines a set beat, and have individuals from our team oversee those recommendations. They’ll make sure there’s no cultural whiplash between songs, and there’s not a ridiculous transition, and decide which song is lyrically more motivating to run to than another. It’s always a mix of data, algorithmic suggestion, data based on genre, era, artist, and acoustic attributes. All the machine stuff is amazingly cool, but all the work that goes into the tech are inspired by athletes, musicians, and real people.
MF: How often do you update the playlist if you see a song (or songs) is getting skipped frequently?
DF: If 70 percent of people are skipping one particular song, and six months ago, it was only 20 percent it’s time to get rid of it. So maybe that song is getting stale, or it wasn’t a right fit in the first place. A huge part of Spotify is that we like to make sure people know we’re back here listening, watching, and taking care of the playlists. It’s not just, ‘Here’s a playlist. Take it. We don’t care about you.’ I think that happens a lot with playlisting. I’m not poking at anybody else, but that’s incredibly important to my team. We pay attention to the consumption, the demos, and the skip rates. We’re able to see the dead tracks, what tracks are getting stale, what tracks might never have worked.
MF: You mentioned before that some of these playlists are created by one person. Are they in charge of maintenance?
DF: Yes, and we have oversight from the team. We have a lot of fun, healthy competition amongst the curators, and a lot of shared opinion. It does come down to the one person really taking care of that playlist. But I can’t stress enough how much we rely on team. We have a whole team who build tools to help merge man and machine. We’ve done over 4,500 playlists that we maintain. We rely on this amazing, intense Spotify data that’s put together by 75 million people (Spotify users), and we analyze the 1.5 billion playlists on Spotify that have been made by these people. So you can imagine the type of information you can get from all that.
MF: What’s Spotify’s overall philosophy on playlist-making?
DF: Well, there’s a lot of stuff I can get into about how we feel about music accentuating different types of workouts. The thing is, we’re not just guessing. We’re carefully paying attention to every element of that playlist and paying attention to people and, if they gravitate to a playlist, we make suggestions, and we start to habitualize this for people. We cater to you.