The battle for streams (and what it’s doing to music)

Julian Freyberg

Hi there! I’m Julian. Usually, I develop new features for the Datawrapper app. Today I’m doing something different but equally fun: Analyzing Spotify top 100 song data to understand how music is changing in the streaming age.

It’s widely recognized that songs are getting shorter:

It looks like we can blame the internet, especially music streaming platforms like Spotify and social media platforms like TikTok, for this phenomenon. Why?

1. Platforms like Spotify pay artists on a per-stream basis, and only 30 seconds of listening are needed for a song to count as streamed. So shorter songs simply earn more money per audience listening hour.
2. Videos on TikTok are often accompanied by viral background music, but they’re very short. Why bother producing a six-minute song when only the “best 20 seconds” are used for the dance videos anyway? An artist’s time is better invested in working on the next track, which increases their chances of a viral hit.

But the duration of songs is not the only thing that’s been changing recently. In this Weekly Chart article, we’ll look at the most successful songs on Spotify over the past 10 years and analyze what’s been going on.

Where this data comes from

I first gathered this data and published on kaggle about a year ago. It contains all of Spotify’s weekly country-by-country charts from 2014 to 2023. For today, we’ll consider just the top 100 songs and only count each song once per country, whenever it first hit its best chart position there. For example, if a song hit #10 in Italy in 2020, #5 in 2021 and #5 again in 2022, we’ll count it as a #5 hit in 2021. The same goes for all countries, so if a song is internationally successful it might occur several times at different chart positions in different countries and years.

After this cleanup, the dataset contains 66,906 unique songs and 217,439 data points.

What’s happening to popular music?

First of all, it’s not just the songs themselves that are getting shorter.

This trend is harder to explain. It might be that shorter titles are streamed more often because they’re easier to remember or search for, or that they get people to click faster because they’re quicker to read. There’s probably also an aversion to very long titles that don't fit well on the user interface of mobile streaming apps. However, it could also be a side effect of other trends, such as the possibility that different genres tend to have titles of varying lengths.

Unfortunately, the goal of getting people to click onto song titles also leads to more drastic actions.

While I'm fine with 3% lowercase song titles, I'm offended by the heavy-handed attempt to get clicks with all caps, and it's tragic that this seems to work so well. It will be interesting to see how long this trend lasts. Maybe at some point “normal” titles will be so rare that they become eye-catching again, and we'll reach a Nash equilibrium. At the moment, as much as I hate to say it, writing the song title in capital letters seems to be a dominant strategy.

The data doesn't show this yet, but I predict the next title trend will be the inclusion of emojis. There is one song in the current German top 50 and it seems like the next logical step to me.

But the influence of the streaming age doesn’t stop at the titles — even lyrics appear to be changing. While I don’t have lyrics available to analyze directly, we can take a look at the “explicit” attribute that Spotify gives to songs with sexual, violent, or vulgar content.

While only around 7.5% of the top 100 songs were marked as explicit in 2014, the share rose fivefold to over 40% by 2023. Daniel Parris from Stat Significant shares his research on this topic in The Rise of Explicit Music: A Statistical Analysis — a nice read if you want to learn more about explicit music.

Genre likely plays a big role here (rap has become more popular and is often explicit), but another factor is the decline in dependence on censorship-ridden media such as radio or TV.

When looking at the country-by-country charts, cultural differences seem to play a major role. In some cultures, explicit language in music has become a lot more acceptable, while in others this trend has stopped or even flipped.

Finally, let’s take a look not just at the songs but at the artists who create them.

In 2014, around every second chart song had a feature. While there are regional differences, everywhere has seen a similar trend: an increase until 2019, followed by a downward slump that comes to a halt around 2022.

The overall increase in collaboration is once again explained by a mix of shifting genre popularity (hip hop being more collaborative than other genres) and the commercial interest of reaching a broader audience. It has also been going on since the 90’s, so this is not exclusively a result of streaming.

This makes the sudden drop in 2019 even more interesting — but maybe not so surprising. While collaboration can certainly happen online, meeting other artists on tour and hanging out in the studio together just wasn’t possible for most of the pandemic. This also explains why the downwards trend seems to reverse around 2022.

Like all art forms, music is subject to constant change. Next time you’re streaming, brace yourself for short, all caps song titles and a long list of artist names — but you won’t get to read them all anyways, because the song will finish before the vertical scrolling effect even reveals them all. That’s it from me for today, I hope you had an interesting read!

---

Julian Freyberg

(he/him, @jfreyberg) is a full-stack developer on Datawrapper’s app team and eager to make charting as smooth and fun as possible. Before joining, he studied data and computer science and worked on scientific apps. He enjoys going to the gym, cooking, and learning about economics and cinematography. Julian lives in Frankfurt am Main.

Liked this article? Maybe your friends will too:

Twitter Facebook