The world’s leading publication for data science, AI, and ML professionals.

Hot or Not: Analyzing 60 Years of Billboard Hot 100 Data

Drake and Taylor Swift are the undisputed King and Queen of the Billboard Hot 100. With 157 and 109 different songs on the charts…

Photo by israel palacio on Unsplash
Photo by israel palacio on Unsplash

Drake and Taylor Swift are the undisputed King and Queen of the Billboard Hot 100. With 157 and 109 different songs on the charts respectively, they have shattered the numbers posted by the Rolling Stones (57), Elton John (64) and the Beatles (69). After realizing this, I wanted to discover some of the other ways that the Music Industry has changed over time, namely by identifying patterns in how much music is released, who is releasing it, and how it performs on the Hot 100. To do this, I will be downloading the Hot 100 charts as structured tables and creating custom metrics to quantify these trends and more.

Exploratory Visuals

The Billboard Hot 100 was first released in August 1958, so for this analysis my data includes every week of every year between 1959 and 2019 (leaving out the two incomplete years, 1958 and 2020) – special thanks to Chris Guo’s API.

The number of different artists that crack the top 100 is decreasing over time, meaning the arena of pop stars vying for Hot 100 entries is about three fewer each year. This market is ripe for superstars to release more songs each with fewer peers to crowd them out.

Pop Music experienced peak diversity in the late 60’s, and made a steady decline to peak repetitive-ness in 2001 (coincidentally the year that Napster was at peak popularity and got shut down). Since 2001, song variety on the Hot 100 has increased, but is still not as high as the 60’s.

Another considerable metric is a song’s turnaround when it debuts to the Hot 100. in the 90’s and early 00’s, songs used to debut with lengthy premiere streaks. Fast forward to 2010, and most songs land on the chart for a week or two, and then bounce off.

While the median does describe most cases, there is space for outliers to disrupt that norm. Here are some of the major outlying cases from 2010:

Take a trip down memory lane... (Image by author)
Take a trip down memory lane… (Image by author)

How can the median debut length be two weeks with such glaring outliers?? I think the data supports a reality where there are a few artists releasing a ton of music – most of it doesn’t last, but the songs that do stick, stay for long. Who can forget CeeLo Green and Bruno Mars’ almost year-long reigns on the charts!

In conjunction with fewer and fewer artists on the charts, each of those artists is charting 1.5x to 2x as many songs per year, with an explosion in 2018. Compare the 5 most prolific artists in 1965 vs 2018:

The Beatles and Drake in the year that each released their fifth studio album (Image by author)
The Beatles and Drake in the year that each released their fifth studio album (Image by author)

"Help!" (1965) by the Beatles contained 14 songs on the album while "Scorpion" (2018) by Drake contained 25 songs, and all 25 songs made it to the Hot 100 that year. The topic of albums getting longer has been extensively studied, and Andrew Mays has written a piece on the topic:

Are Rap Albums Really Getting Longer?

At the dawn of the streaming era, Billboard began treating songs on the album as singles for Hot 100 sales reporting, because they received on-demand plays on platforms like Spotify, Apple Music and Tidal. That contrasts with the 20th century, where singles reporting was mainly derived from radio airplay. These days, lesser-known or throwaway songs have a better chance of making the charts. And with artists like Drake who have the sheer volume to drown out the noise, the streaming era is bound to have pop stars shattering the singles records of artists from the LP and CD eras.

Trends in New Music Seasonality

In addition to solving the question of all-time singles records, I wanted to end my analysis with a breakdown of the best and worst times to release music through the calendar year. Conventional wisdom and music marketing blogs dictate some rules about releasing music: don’t release in December except for holiday music, don’t release during SXSW, and so on. But I wanted to see which months actually saw the most new music released, and how that has changed over time. (click to zoom in)

(Image by Author)
(Image by Author)

January, February and September are consistently poor months to release music, while October and March (when SXSW happens) are generally very good months for new music. June, November and July are ‘okay’ months it appears, and the rest have too much noise to make a solid conclusion.

Final Thoughts

Releasing music as a new artist in 2020 is a wildly different experience than in the 20th century. Some of the gate-keeping forces of those days have been eschewed by social media, YouTube and the internet as we known it. However, new-age gatekeepers like algorithms, playlists and digital advertising continue to impact the industry in unique ways. All in all, the effect of these forces seems to be shrinking the opportunities for new musicians, but I believe that as music revenues return to former glory, opportunities will increase for new artists to make it big.

Thanks for checking out my analysis! To see how I went from raw Billboard Charts to the metrics you see above, checkout my GitHub repo for this project.


Related Articles