Making a Hit Song
Authors: David Spector (Project Lead), Madeline Zhang, Hyerin Lee, Morgan Chan, Peiyuan Lee
Cover Image from Here
Summary:
Aside from musicianship and artist popularity, are there other ways certain songs can be pushed to the top? What if societal conditions and trends actually correlate with which types of music are most popular? Read our article to explore the external factors involved with determining mainstream music as we all know it today.
Analysis:
Music is traditionally thought of as a subjective medium where each listener has their own listening taste and playlist they habitually turn to. However, each year, certain types of music stand out from the rest and reach the top of the Billboard Charts. Due to the varying amounts of Alternative, Electro, Hip Hop, and Pop music fluctuating at the top each year, it is possible external factors such as societal conditions or social trends affect which types of music succeed in the marketplace and which get left out. To explore this, we sampled the 31 most popular songs annually from 2010 to 2021, excluding 2020, and compared the results with real-world events. The result is a clearer glimpse into the external factors which help determine a song’s monetization and global outreach.
It’s no secret that social media influences what music people listen to, as music streaming services like Spotify, Apple Music, Soundcloud, YouTube, etc. have become such an entity themselves. Observing a timeline from the past decade of music-related events in social media and trending music genres, we look further into the relationship between these two.
Dotplot displaying the number of social media users each year from 2010–2019 by the most popular platforms
Dotplot displaying the frequency of genres in the top songs each year from 2010–2019
In 2014, a broad variety of genres including Australian dance, Australian hip hop, electropop, folk-pop, Hollywood, house, and metropolis, emerged in the top songs charts. In the same year, Twitter, with more than 223 million monthly users, partnered with Billboard to create real-time charts to track tweets related to music, and Musical.ly, an app where users create, share, and interact with short videos using music/sounds, was released a few months later. These apps heightened music-related social interaction, which ultimately affects which genres are most popular and most discussed online. In 2015, when dance pop and Canadian pop had the highest number of appearances on the top charts within the decade, both Apple Music and YouTube Music were launched. With the addition of these large-scaled music streaming platforms, users could share, stream, and discover all genres of music. The following year, Snapchat, which had over 200 million users per month, partnered with Shazam, which enabled users to scan and find songs playing in their proximity through audio recognition with the Snapchat camera. In this year, pop, dance pop, and Canadian pop had decreased from previous years, whereas genres like British soul and electropop had increased. Musical.ly later changed its name to TikTok in 2018, when it had amassed nearly 500 million monthly users, and Instagram released a music feature, also in 2018, which was accessible to 400 million of roughly 1 billion active monthly users, allowing them to add songs to the content on their Stories and expanding the overall world of music on social media. Hence, pop and dance pop saw a significant increase from 2017, and relatively less common genres like detroit hip hop, brostep, contemporary country, and moroccan pop also began to appear in the top songs in 2018.
While many mainstream artists are featured on the radio, social media has become an external source for discovering new music and artists. We can see how major events in social media relating to music within the past decade have developed a widespread presence of music in platforms such as Instagram, Snapchat, and Tiktok, growing the overall influence of social media on trending music genres.
In addition to the ever-changing platform of social media, the state of our global economy may also have some impact on mainstream music styles. Since a good economy is often paired with a higher standard of living, low unemployment rates, and greater public consumption, it would follow that the public would have greater access to music, more time to explore new genres, and even be “upbeat” about their collective financial positions. However, according to Spotify’s data, reality paints a much different picture. For almost all music attributes studied, there was no statistically significant correlation with Global GDP or Change in GDP. The sole exception, however, which ended up having a strong, positive relationship with Global GDP is Music Popularity.
Relationship between global GDP and the average Spotify Popularity Score of the top 31 hit songs of the same year
To determine what this actually means, we must first define the statistic. According to loudlab.org, “the Spotify Popularity Index is a 0-to-100 score that ranks how popular an artist is relative to other artists on Spotify”. With similar calculations for an individual song’s popularity score, the correlation between Global GDP and Music Popularity suggests that as the world becomes more economically advanced, the general public tends to listen to the same mainstream music more often, implying less overall diversity in music taste. In fact, since this correlation has an R-Squared value about 0.788, our data shows that increases in annual, global GDP account for roughly 78.8% of all increases in Music Popularity for the top 31 songs that year. Contributing factors to this relationship could be anywhere from increased commercialization to a potentially greater number of large social gatherings. In any event, while a strong economy seems to be the ideal conditions for well-established songs to thrive and rake in ridiculous profits, it may mean the opposite for lesser-known artists who haven’t fully caught the public’s attention.
Another factor we were interested in exploring were the effects of major global events on music tastes and popularity. To explore this relationship, we’ve utilized the data sets from Spotify’s Top 100 Hong Kong Tracks in 2019 and 2015, the former being the year where the protests against extraditions commenced. While examining the attributes of top tracks in 2019, we found that in comparison to 2015, the overall valence, which measures the happiness of the tracks, is lower than the top tracks in 2015. In other words, the tracks listened to in 2019, a year of chaos and turmoil for Hong Kong, are more melancholic compared to their 2015 counterparts, although the difference is marginal.
The diagram displays the valence and danceability of Top Hong Kong Tracks in 2019 and 2015.
To further explore the relationship between the valence of top tracks and music popularity, we’ve chosen to look into the data set for top tracks on Prince Archie’s (Meghan Markle’s child) birthday, an arguably joyous occasion for the British, to determine the effects of social moods towards music popularity. The data shows that the valence of the tracks were higher than that of Hong Kong’s in 2019, however, there is little to no difference between the top tracks in Hong Kong in 2015. Thus, we can conclude that there may be a possible correlation between the overall valence of tracks and major global events, albeit having a weak relationship. More data is required to accurately ascertain the complicated relationship between the two.
Boxplot displays the distribution of top tracks on 3 different occasions
Taking Spotify’s dataset of different music attributes such as energy, liveliness, artists, popularity, we tried looking at the correlation between popularity and other attributes of a song to see if any of the attributes had a strong correlation with popularity. Unfortunately, the correlation matrix showed that none of the numeric variables had significant correlation with popularity. Then, we decided to take a look at the categorical variables, such as artists, keys, and year released. Taking the mean popularity for each key showed that the key of D flat had the highest average Music Popularity rate of 43.15, and key of G flat had the lowest popularity rate of 40.64. However, all the keys had similar ranges of popularity, which led us to conclude that different keys do not have a strong correlation with popularity.
Correlation matrix of different variables within the dataset
Next, we looked at popularity by artists as artists are a significant factor in publicizing music. The top 10 artists with highest average popularity scores included Cardi B, Megan Thee Stallion, Bad Bunny, etc., all of whom were well-known artists. While we could not find a way to directly associate artists with popularity scores due to the difficulty of it being a categorical variable, there is definitely a potential that artists may have strong correlation with popularity. One good way to expand on this project would be to join this dataset with a new dataset that has an artist’s “fame” scores that rank how well-known an artist is. It also seemed that the year released showed an upward slope compared to popularity. As seen in the graph below, songs that were released more recently tended to have a higher popularity score. Songs released in 2021 had a highest average popularity rate of 57.5, while the lowest of 28.6 was in 1940.
Scatterplot of average popularity rate by each year
Finally, an interesting development in the characteristics of hit songs throughout the 2010’s is the fact that the average hit song’s accousticness has grown significantly from 2010 to 2021, becoming more prevalent than the once popular attribute, liveness. According to the Spotify database, accousticness is a value associated with how much a music uses real instruments as opposed to electronically-made sounds. While the recent rise of EDM and other electronic genres in popular culture may influence this trend, our data shows that accousticness was on the rise on top of the charts during the 2010’s. Also, the liveness statistic represents the chances a song was recorded in front of a live audience. Therefore, since accousticness has become more prevalent in popular songs than liveness in the past couple of years, the data suggests that general audiences have come to prefer well-produced songs with strong, natural instruments. Releasing tracks with a live audience in the background is evidently going out of style.
Average Accousticness and Liveness of top 31 songs annually
Outside of song characteristics, marketing for music is incredibly important for how songs will perform on music charts and the reach a song can be played. Song titles often influence how a song is perceived and indicate what songwriters and artists want to present from their song. By sorting title Spotify data, we can discover popular choices in songs through how artists choose their titles.
We have found that the highest single word is “Feat” (featuring) which demonstrates how collaboration between artists, especially if both are popular, can consistently lead to higher ratings. Perhaps in an artist’s next release, they should consider working with another artist to reach more fans and gain more popularity. Additionally, many articles like ‘the’ and ‘it’ are also popular, which is expected as they are necessary for syntactical purposes. ‘Love’ is also a common theme, possibly since songs with more emotions will connect and resonate with audiences more. The high frequency of ‘you,’ ‘me,’ ‘I,’ ‘my,’ and ‘we’ demonstrates that people like music more when songs come from the first-person and second-person perspective, like the artist is singing from their personal experience or to someone else. However, these commonalities could also be just preferred formats of music and music writing.
It is noted that these common words are typically easy to read and understand, even for children and foreign audiences who would be less familiar with English language.Trying to have more radio play time means that songwriters should consider keeping song titles and lyrics alike more family-friendly.
Bar graph displaying common words found in popular Spotify songs
Furthermore, these key characteristics of making a popular song title can be related to the length of the title itself. We have found using the same dataset that the most popular length of the title is 1 word, and more top songs use significantly fewer words in their title. This can indicate that less words in a song title is easier to remember and can be more eye catching when looking at a list of songs.
Histogram displaying common length of title found in popular Spotify songs
In conclusion, after examining popular music trends in a wide variety of lenses, we have found no perfect mixture of world conditions, music attributes, and lyrics which directly lead to Billboard success. While being a well-established artist certainly helps, there are never any certainties about which songs will make it and which new trends will catch the attention of worldwide audiences. Further exploration of genre-specific datasets and more categorical variables are needed to offer artists more concrete insight into success in their genre and in the global marketplace. However, after examining varying economic conditions and important world events, we have found that certain global conditions may shift societal moods and subtly affect which songs are popular. We’ve also found that featuring another artist may slightly increase your odds of topping the charts. Overall, while popular social media platforms, economic conditions, unpredictable world events, and other cultural movements will undoubtedly continue to influence popular culture and mainstream music, the ideal mix of societal conditions to release certain kinds of music largely remains a mystery.
Sources:
https://www.kaggle.com/datasets/leonardopena/top-spotify-songs-from-20102019-by-year
http://student.elon.edu/rwalker14/spotify/
https://ourworldindata.org/rise-of-social-media
https://data.worldbank.org/indicator/NY.GDP.MKTP.CD
https://www.statista.com/statistics/268750/global-gross-domestic-product-gdp/
https://www.kaggle.com/datasets/equinxx/spotify-top-50-songs-in-2021
https://www.loudlab.org/blog/spotify-popularity-leverage-algorithm/
https://en.wikipedia.org/wiki/Musical.ly
https://en.wikipedia.org/wiki/YouTube_Music
https://www.businesswire.com/news/home/20161213005521/en/Shazam-Announces-Integration-with-Snapchat
