Design. Illustration. Data.

︎ ︎ ︎


What Makes A Song Popular is a data visualization project that seeks to analyze how the top 100 songs have changed over time. This project stemmed from a desire to see how trends in music have changed over the past decade by looking at the top streamed songs every year and their associated attributes through the Spotify database.


Timeline: 1.5 month
Collaborators: Hana Kim & Jason Cho


The primary audience included all music lovers across the globe and also extends to those who are interested in the analysis of music trends over time. The data story hopes to answer the questions of what trends affected the popularity of certain genres and attributes of music over the past decade. Some overarching goals through this project are to give a better insight and understanding into the music world to the broader audience, and perhaps also predict what future trends may be based on current trends.

Through our visualizations, we hoped to answer:

  • What songs were most popular in each of the years from 2010-2019?
  • What genres of songs were the most loved by the public over the last decade?
  • What genres of songs were most popular in each year for the past decade?
  • Who were some all-time favorite artists over the past decade?
  • What types of music does the public like more?
  • What’s the overall characteristics of songs that the audience likes (regarding bpm, energy, danceability, loudness, live recording, positivity, duration, acoustic, and amount of spoken words)?


Datasets were provided by Spotify, which accounted for the top 50 songs every year from 2010-2019, as well as their associated traits of genre, bpm, energy, danceability, loudness, liveness, valence, and length, acousticness, speechiness, and popularity.

After acquiring the dataset, each team member provided initial drawings for visualizations.

Our storyboard focused the main message on identifying the attributes of top songs from 2010-2019 that made them the most popular that year. To do so, a hook was set that relates to the audience and made them curious about the topic. Then, the songs that were popular in each year were analyzed. With these supporting insights, the main message identifies the top attributes, and in the end, it sees the overall trend and predicts what attributes are likely to make songs more popular.


I prototyped a preliminary design of our website using Figma, which was inspired by Spotify's desktop look.

I was in charge of implementing the stacked area chart for showing artists with the most popular hits and three donut charts for showing specific attributes of the most popular song for each year.


Two think-out-loud testings were conducted to test the usability and learn more about user needs.

According to the feedback from testings, we gained helpful feedback and then iterated designs based on it. Here are some important improvements:

  • Bigger font-size and style for messages and captions to be more aesthetic and visible
  • Scrollable and ‘sticky’ main menu
  • Coordinated color palette for each visualizations chart description and axes labels
  • Have tooltip popup for only the hovered-in section
  • Coordinated color palette for each visualizations
  • Revise wordings of some messages/captions (including Introduction and Conclusion) to target all our audience
  • Make Conclusion more comprehensive with a strong final concluding statement and call to action



Michelle Liu

Currently in: Cambridge, MA ︎


︎ I am a senior at Harvard College majoring in Computer Science and minoring in Studio Arts. I have interests in design, visual arts, and data.

︎ Outside of school, I love to paintdraw, animate, and photograph in my free time. You can find me bouldering, trying out new cuisines, and watching sunsets.


Let’s connect!

︎ Email
︎ LinkedIn
︎ Instagram

︎︎︎ Resume