365 Days of EDM on YouTube

Introduction

While some say that electronic dance music is just a fad, it is undoubtedly one of the most popular music genres today.

Though I am kind of an older armchair EDM fan and not the young party animal, I was curious to see how the genre evolves over time and started brainstorming about a potential analysis, and so more than a year ago I’ve set up a small piece of software that started collecting data.

At the beginning I have asked fairly simple questions like “Which are the most popular songs?” and “How did they become popular: instantly or gradually?”, but as I was getting the answers, more and more questions have arisen and many unexpected patterns have been discovered.

In the following I present the results of my analysis and I hope you too will find some interesting bits and pieces about the worldwide EDM community.

Dataset

The analysis is based on the content uploaded to YouTube channels dedicated to electronic dance music. The following channels were being monitored:


anjunabeats

anjunadeep

armadamusic

blancoynegro

dimmakrecords

happymusic

kontor

ministryofsoundde

mostv

musicalfreedomrecs

pandoramuslc

planetpunkmusic

spinninrec

toolroomrecords

ultrarecords

zoolandmusicgmbh

Though data collection was started a bit earlier, only the period between 1 July, 2015 and 1 July, 2016 is part of the analysis. Only newly uploaded content was tracked (videos that were uploaded before July 2015 were not being monitored).

2015
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2016
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

The period covers the second half of 2015 and the first half of 2016: during that time 4,326 videos were uploaded to the 16 YouTube channels in the dataset. For each upload the number of views were recorded every 15 minutes, resulting in more than 70 million measurements during the 12 months of data collection.

Uploads and Views

The number of uploads shows how much new music is getting released by EDM record labels (supply) and the number of views can tell us whether people are actually interested in that new content (demand).

Daily Uploads

The number of daily uploads over the last 12 months were quite balanced, there is no statistically significant general upward or downward trend in the series. A noticeable big drop happened between 28 Dec 2015 and 3 Jan 2016 (the middle of Fig. 1): labels were taking a break on the week of New Year’s Eve.

While some of the spikes are caused by a single channel uploading many videos at once (e.g. an entire artist album), the day of the week breakdown reveals a general pattern:

As one would expect, the least number of uploads are made on Saturday and Sunday. Labels in general are more likely to release new music just before the weekend: 1 in 4 of new uploads are made on Friday. The beginning of the week, Monday and Tuesday, are also a popular choice for uploading new content, these two days combined with Friday are responsible for 57% of all uploads.

Daily Views

As compared to the number of daily uploads, there is a general upward trend in the number of daily views. The spring of 2016 has brought an increased demand for EDM which is still maintained for the current months.

Important note: though Fig. 3. shows the rate at which views were captured, it cannot be used as a measure of absolute growth. In other words: the popularity of EDM on YouTube did not grow by a factor of 4 in the last 12 months!

Let’s look at the day of the week breakdown of views:

A higher than average number of views are recorded on weekends, especially on Saturdays. Not surprisingly, the number of views on Mondays are the lowest.

Uploads vs. Views

If the number of daily uploads and views were even for all days of the week, there would be 618 daily uploads and 260M daily views for each day of the week (the same for Monday, Tuesday, etc.). Fig. 5. shows the relative difference of the actual data to this ideal numbers:

Even tough that the number of uploads on an average Monday is 8% higher than the rest of the week, the number of views are 10% lower as compared to other days. Similarly, despite that on a typical Friday there are 83% more uploads as we would expect, it doesn’t attract significantly more views than the other days of the week.

Both Wednesday and Thursday has less than average number of uploads and views: it seems that both the labels and the fans are waiting for the weekend.

Popularity

Some songs become very popular, others barely get a few thousand views. While no amount of data will ever tell what exactly makes a hit, it can tell which songs can be considered as such.

Distribution of Views

Count Share Total Views Share
Less than 100,000 views: 2,680 61.95% 66.29M 3.66%
Between 100,000 and 500,000 views: 1,014 23.44% 250.41M 13.81%
Between 500,000 and 1,000,000 views: 326 7.54% 230.13M 12.69%
More than 1,000,000 views: 306 7.07% 1,266.33M 70.42%

Fig. 6. shows how unequal the popularity of individual songs is. Videos that reached at least 500,000 views (15% of all uploads) account for 83% of all views and the rest of them (85% of all uploads) got only 17% of the views.

To be in the top 10% based on views, a video must have been played at least 730,000 times and only 24 of all the 4,326 uploads (0.55%) has reached more than 10 million views.

Top 50 Most viewed EDM Songs (July 2015 – July 2016)

The following table shows the top 10 most popular videos in dataset (views are as of  1 July 2016):

The problem with determining the popularity of a song based on the absolute number of views it has received is that it favors older uploads (because in general, the more time elapsed since upload, the more views an upload has).

Growth

We need to take a look on growth to better understand which songs are the currently most popular ones.

For example, an upload with 25M views but an age of ~7000 hours (~290 days) is less popular currently than the one that reached 10M views and was uploaded ~2000 hours (~80 days) ago.

On Fig. 7. each dot represents a video: the horizontal axis shows how many hours elapsed since it was uploaded (1000 hours is about 42 days) and the vertical axis shows the number of views it got:

There are so many factors influencing the number of views a song gets that each of them has a unique growth profile. Fig. 8 – 12. provides examples of such growth profiles: on the left side you can see the number of daily views (grey area) and the number of total views on the right side (black line).

Tiësto & Don Diablo – Chemicals was uploaded in late September and was getting more and more popular throughout the rest of 2015:

After an initial short burst, Cheat Codes x Kris Kross Amsterdam – SEX was forgotten, but made a comeback in early April and got into a very strong upward trend:

Tujamo – Drop That Low has managed to maintain its popularity since its release in March 2016:

Though Benny Benassi & Chris Brown – Paradise is quickly losing its popularity, it still gets 200k views a day:

Mahmut Orhan feat. Sena Sener – Feel is an example of song that was uploaded recently and is getting more and more popular day by day:

Top 50 EDM Songs of Summer 2016

Based on the analysis of the growth profiles I have compiled a list of the songs that are likely to dominate the sound of summer 2016 (age and views are of 1 July 2016, age is the days elapsed since upload):

Keep in mind that this is just one of the many possible ways to rank the songs based on their potential to become the major hits of summer 2016 (and some of them are already have became), so don’t be offended if your favorite new song is not on the list!

Record Labels

Not only that each label puts a different emphasis on YouTube, but their artists usually have their own channels which are not part of the dataset.

While it would be risky to draw conclusions about the exact market share of each label based on the data available, it is safe to say that the top 5 EDM labels are as the following: Spinnin’ Records, Ultra Music, Kontor Records, Armada Music, and Proximity.

This fact comes by no surprise considering that these are among the EDM labels that were started the earliest (Kontor and Ultra in 1995, Spinnin’ in 1999, Armada in 2003). However, it is worth to note that Proximity was started in 2009 and it is already the 5th most popular YouTube channel for EDM.

Conclusions

Though this analysis is based solely on one platform, it provides widescale insights into the worldwide EDM community. The key points are:

  • The popularity of EDM on YouTube is still increasing: labels supply a steady flow of new content and fans watch more and more of it.
  • Only 1 in 14 releases (7%) become hits (more than 1M views) and 1 in 180 (0.55%) megahits (more than 10M views).
  • Although five record labels dominate the scene, smaller labels can find success as well because the environment is constantly changing.

The music industry is so complex that predicting future hits is impossible, but seeing what resonates with the audience is essential in supporting an artist’s career. While this report shows what happened, record labels could gain actionable business insights by getting to know why it happened so.

When will the bass drop?

Has EDM already reached its peak or it will become even more popular in the coming years? Only time can tell that for sure but currently there’s no sign of any slowdown!

Contact

If you have any comments or questions, feel free to contact me at balazsdavid987 at gmail dot com. Alternatively, you can join the discussion of this analysis at reddit in r/EDM.

Appendix

List #1: Top 50 Most viewed EDM Songs (July 2015 – July 2016)

List #2: Top 50 EDM Songs of Summer 2016

45Bob Marley feat. LVNDSCAPE & Bolier – Is This Lovespinninrec20837.68k

Advertisements