Spotify Charts Analysis: Where Is Everyone Listening?

Spotify shares the top 200 most-streamed songs by users around the world for that day. This daily list, or “Spotify Charts”, can be viewed for 60 countries and in many cases, is listed all the way back to January 1st, 2017.

This data allows us to investigate a wide range of questions. In this post, I’ll be seeking to answer which countries have the most streams a day in their top 200 and which countries have the highest top 200 streams per capita.

Before I continue, I should acknowledge that these analyses are only based on the streams provided in Spotify Charts. They certainly do not encapsulate the total magnitude of Spotify streams worldwide. I don’t typically listen to many of the songs listed in the Spotify Charts, so many of my daily listens are not represented here – I imagine many others are the same. They say 90% of an iceberg is underwater, and I suspect that these charts streams are similar, only representing a small portion of the total daily streams on Spotify.

I used Python and many of its packages, including: BeatifulSoup, for scraping the Charts page (I modified code shared here – and since deleted?); Pandas, to store and analyze the data; and Seaborn/Matplotlib, to generate the plots.

Does our dataset make sense?

First off, I should note that for the initial analysis I’ll only be looking at the Spotify Charts for each available country from January 1st, 2020 to April 11th, 2020, a span of 102 days. All in all, there are 60 local country charts and one global chart. If there were 200 songs for all days and all charts, we’d expect 102 days * 200 songs/day/chart * 61 charts = 1,244,000 songs. When we inspect the dataframe where all the chart info is stored, we can see that there are only 1,177,664 rows.

df.shape

(1177664, 8)

This means that we’re “missing” 66,736 songs. This due to the fact that Spotify does not seem to include any songs getting less than 1000 streams on these charts. Many countries do not have 200 songs receiving 1000+ daily streams, thus the total song count is lower than expected. This can be seen on Bulgaria’s chart for 4/11/20, which only lists 56 qualifying songs.

Another interesting “discrepancy” in the data is that the sum of all 60 represented country’s streams is larger than the sum of global streams, 44.31 billion and 27.33 billion, respectively. This isn’t really a surprise, as many countries have some songs on their local charts that are not counted on the global chart, therefore the sum of all local charts consists of more total songs. For example, the #1 song in Vietnam, “Ke Cap Gap Ba Gia (feat. BINZ)” by Hoang Thuy Linh (a total bop btw), is not included in the Global Chart’s count.

On the other hand, there are some streams that are counted in the global total that would not be captured in our local totals. The Spotify API provides access to even more data about a song, including what “markets” or countries it is available for streaming in. When we check the “available markets” key for the global #1, “Blinding Lights” by The Weeknd, we can see that this song is available in 79 countries; we only have charts for 60 countries. To assess how big this difference is, we can look at the total number of streams for a song globally and compare it with the sum of the total number of streams in each country that we have data for. For example, on 4/11/2020, “Blinding Lights” was streamed 6,480,759 times globally. When the local streams of this song for each of the 60 countries in our dataset are added up, we get 6,443,927. This is a difference of 36,832 streams. We could fret about missing out on knowing where these 36,000+ streams are coming from, but we can also take comfort in the fact that we can account for more than 99.4% of all the daily streams of this song.

What countries stream the most?

The first thing I did to answer this question was make a simple bar chart.

The global streams kind of dwarf all of the country local data, so I’ll remove it to give us a better look.

So we can see that the US dominates in the total streams totals, eclipsing 8 billion streams in the the first 102 days of 2020. See the 10 highest streaming countries tabulated below.

We can look at the above table and see that 5 different continents are represented, with half of the countries in the top 10 being a part of Europe (albeit, not necessarily the EU).

If we look at this data sorted by continent, we see some other trends emerge.

We can also compare the percentages of the total share of top 200 streams by continent with a pie chart.

So we can see that Europe and North America each make up about a third of Spotify’s total streaming. Asia and South America are each at about a sixth, and Oceania and Africa make up the remainder. It is also important to point out that only one African country, South Africa, is currently making the Spotify Charts.

What country streams the most per capita?

I feel like the above charts can be a bit misleading. They clearly show which countries are streaming the most, but the top 10 highest streaming countries bears much resemblance to the most populous countries represented in the dataset. Of the top 10 streamers, half of the countries also fall in the top 10 of population.

With more people, it makes sense that a country would be able to stream more music. If we divide a country’s total number of streams by the population of the country, our bar chart will take a different form. I used the World Bank DataBank’s population data for 2018. This population data does not count Taiwan’s population, so I added that to the dataset, pulling the data from worldometers.info.

I certainly realize that the entire population of a country is not using Spotify, and even they are, many are not listening to any of the songs listed in their country’s top 200 chart. As mentioned earlier, since we’re limited to streaming counts for just the top 200 songs in a country, there are likely billions of streams not accounted for in our data. Perhaps all that normalizing this data by population is doing is showing what country’s population have the most similar taste in music (i.e. more than 10% of our analogy iceberg is above water). But in absence of more data, I’m assuming that each country has a similar distribution of total streaming being counted in their respective top 200 chart. Normalizing total streams by population allows us to do a more apples-to-apples comparison of the prevalence of Spotify use in various countries.

Now we see a bar graph that looks completely different. The highest streaming countries’ totals don’t seem so high when normalized by population.

The 10 highest streaming countries per capita list looks much different than the 10 highest total streams list. Only one country made both lists, Australia. Sweden, the birthplace and headquarters of Spotify, vaults from 17th on the total streams list to 2nd on the streams per capita list. In fact, all Nordic countries are represented in the per capita list, along with both Oceanic countries that have available Spotify Charts data. Chile just missed the top 10 on the total streams list, coming in 11th by that metric. It jumps into 5th for per capita streams.

Conclusion

From these simple graphs and tables, we can begin to get a better idea of what countries are streaming the most music and where Spotify is most widely used. Not surprisingly, countries with larger populations make up a larger chunk of the total streaming on Spotify. The United States, Brazil, and Mexico, the 2nd, 4th, and 6th largest populations on this list make up 35% of the total streams in this dataset.

When these streams are normalized to population, the numbers tell a different story. The US and Mexico sink into the second tier of per capita streams, with 26 and 27 streams per person, and Brazil falls even lower, with 17. If we plot these streams per capita on a histogram, it helps illustrate just how much growth potential Spotify still has; if they can get the rest of the world streaming like Sweden it’d increase the total numbers of streams by billions.

That leads me to my final point: as mentioned above, the top 200 most popular songs in 60 countries were streamed over 44 billion times during the first 102 days of 2020. That’s just a staggering number. If the average song length is 3:30 and everyone was streaming at the lowest-available quality, 96 kbps (which translates to 0.72 MB/min of data), that would be 111,686 terabytes of music. That’s nearly 160 billion CD’s. That’d be a lot to carry around.