This project demonstrates how dynamics within the music industry have shifted since the advent of streaming services. It used to be that radio DJs were the gatekeepers of the music industry, with their ability to offer up-and-coming acts a level of regional exposure that could change careers. But over time, radio has lost its relevance as a tastemaker to services like Spotify, which can break new artists quickly and globally through its massively popular, curated playlists.
We wanted to quantify the degree to which streaming is able to surface new music faster than radio in 2019. Using data from Spotify and Billboard, we calculated how quickly songs reached the top 50 on the Spotify and radio charts and how long they tended to stay there. Our results speak to a shift that most industry insiders know was happening anecdotally, but illustrate the point through data and interactive graphics.
What makes this project innovative?
This project featured a completely original dataset that we gathered from Spotify, Billboard, and through our own manual efforts. The dataset was amassed in an attempt to quantify how much faster new songs gain traction on streaming services than radio, which has never been done before in a rigorous way. The story also hinges on an innovative narrative structure, with an introduction that focuses on one example in the dataset to illustrate the thesis of the piece more tangibly.
What was the impact of your project? How did you measure it?
We typically measure the success of any project based on four criteria: 1. How well it engages / retains people on the site 2. How well it resonates with the target audience on social media 3. The quality of the media mentions it garners 4. The rate at which it drives signups for our weekly newsletter, Data & Eggs Some highlights for this project: -The project was shared on social media by numerous music industry insiders, including radio personality Dom Alessio, a product manager at Spotify, and the CTO of Kobalt Music. -Readers viewed the page for over four minutes on average, more than 20% above a normal page on The DataFace website. -Drove newsletter signups at a rate 40% higher than the average page on our site.
Source and methodology
We used Billboard’s weekly chart of “Top Radio Songs” in the US between December 25, 2016 and December 25, 2018. We used Spotify’s weekly list of Top 200 U.S. songs over the same timeframe, but culled each week’s list to the top 50 to match our data from Billboard. There were two missing weeks from Spotify data because the links appear to be broken, but it did not significantly affect the averages. We determined the date of each song’s release by retrieving data from the MusixMatch API and manually supplementing it with information from Wikipedia. We classified each artist by genre based on results from the Spotify API and Wikipedia. Songs released before December 25, 2016 were removed to avoid muddling our calculations of average trajectories on radio and Spotify (because we don’t know whether these songs charted in prior weeks). Similarly, we removed songs on the December 25, 2018 charts because the number of weeks they’ll remain in the top 50 was unknown. However, all songs were included in our analysis of artists that appeared on one chart but not the other, as seen in the final two visualizations.
Data was scraped and analyzed in Python. Data visualization was developed in D3.js and scrollytelling was implemented using the Scrollama library.
Jack Beckwith, Michael Hester, Wyatt Shapiro