Data isn’t life. They say music is.
Well, is it? Wouldn’t it be awesome to simply combine both worlds?
Let’s find out if this could turn out to be a perfect marriage!

Today, with approximately 286 million users to date, including 130 million subscribers, Spotify can be considered as one the world’s most popular music streaming services. In a remarkable coincidence, I happen to consume their service every once in a while myself. Actually, this is an understatement as I stream music basically on a daily basis: During travel, development, while exercising or even when writing this blog at home. Below some general statistics.

%

Of days streaming

Minutes of streaming music

Songs listened to

Above statistics cover a period of one year of streaming behaviour, starting on May 1, 2019.

What would we like to find out?

In regards to my streaming behavior, I came up with the following questions:

What artists did I favor the most
What songs did I listen to the most ?
Can certain events, such as festivals, cause a change in my streaming behavior?
When did I stream music?

What KPI’s can we define to provide us the answers with?

Of course we need to quantify these questions based on a KPI. I believe there are two ways of measuring whether I like or dislike a specific song. The first one is the time I actually listened to the song, which I will be referring to as the listening time (in minutes). The second KPI is the song count, being the number of times a song started, regardless of me manually pressing the play button or Spotify’s autoplay starting the song.

Where can we gather the data we need?

Spotify offers their users to retrieve data through their Spotify API. Recently, they launched a refined version of this API which, at the moment of writing, is still in BETA. I can only praise Spotify for how well they document their API. They offer a clear process description on how to connect to their API, written down call definitions and even providing developers a test environment console. More information can be found here.

Evaluation time!

As we now have all data in place, we can assemble our very own Spotify dashboards with Power BI. Let’s analyze my streaming behavior of the past year by trying to formulate an answer to our predefined questions. 😊

What artists did I favor the most?

Post Malone seems to be the clear winner based on my streaming behaviour, there’s no doubt there. I tend to have a preference for R&B with Tyga and Blackbear claiming a spot in my top 5 artists.

What is interesting here is that my top 5 artists slightly differs when switching between our two predefined KPI’s. Spot 1-3 are identical, regardless of the KPI we base ourselves upon. However, in terms of song count, Nickelback and Loud Luxury are my respective number 4 and 5 whereas, in terms of play time, Loud Luxury and Blackbear take in those positions. This can have two possible explanations: Songs of Nickelback are simply shorter than Blackbear songs or I’m excited to hear Nickelback songs’ refrains and get bored afterwards, forcing myself to switch songs once the refrain passed by.

What songs did I listen to the most?

It wouldn’t surprise you I listened more than 150 times to Post Malone’s top songs such as A Thousand Bad Times, Sugar Wraith and Goodbyes.

Can certain events, such as festivals, cause a change in my streaming behavior?

Yes! Last August I attended the Sziget festival in Hungary. According to my streaming history, I enjoyed Post Malone’s performance so much that the following months, Post Malone skyrocketed and became my number one streaming artist.

When did I stream music?

It turns out I’m actually getting value out of my monthly subscription. The past year, I streamed 357 out of 365 days. My listening time stays more or less steady throughout the week. During the work week, I mainly stream music traveling between home and work. And if you’re interested; yes, I’m not an early bird. 😊

Wrap up

This article described one way of vizualising your Spotify streaming behavior with Power BI. Of course there are many more insights these reports can be enriched with. I tried to provide a high level summary of my journey. If you enjoyed reading this article, feel free to connect via LinkedIn or Twitter. All aboard the data train!

Lou Segers

Analytics Manager @ Plainsight
Public speaker
Blogger