Billy Eilish is a pop megastar, with billions of streams across platforms and a huge social media following. And she’s just dropped a new single! MY FUTURE was released on July 30. Your task is to recommend what platform(s) she should focus on to promote this new release. In order to help you think through this problem, you will have access to Chartmetric’s data on Spotify, Instagram, TikTok, her YouTube channel, and her YouTube video views. You can also go to (www.chartmetric.com), sign up for a free account, and search for her to get some recent stats. But you also have loads of content to to draw upon on these platforms, as well as your own experience of how you and your friends discover music. Use all of this information as you think through this problem. There is truly no right or wrong answer here – only the story you choose to tell, and how you combine the facts available to you with your own creativity in order to tell a compelling one. Bear in mind that careers in data aren’t just about crunching numbers – often, the best insights are those that combine simple quantitative analysis with great storytelling. We’re excited to hear your recommendations.
The first step in any analysis is exploring the data, usually through visualization. We can see the data much more clearly when it’s on a graph rather than a spreadsheet. You may graph the data yourselves using a tool of your choice, but we’ve also provided a number of visualizations for you, below. Study the graphs carefully, both individually and together. What do they tell you about Billie Eilish’s career to date? Consider what they tell you about the quality of the information available as well. If something looks “funny” or surprising, have a look at the spreadsheet data to help you figure out what might be going on. Include your observations on data quality in your analysis of each graph. You will come across multiple Challenge Questions throughout the brief. Pause to consider them and write down your observations before moving on.
Spend some time on Spotify exploring Billie’s music. What kinds of playlists is she on? Do you listen to any of these? You may be interested to know that BAD GUY has been streamed 1.5 billion times, while LOVELY (with Khalid) has racked up 1.1 billion streams, WHEN THE PARTY’S OVER is at 939 million, and EVERYTHING I WANTED another 628 million. On Aug 5, NO FUTURE was at 18 million. How does what you learned about her Spotify presence help to explain the graph below?
ARTIST_NAME | FOLLOWERS | TIMESTP | |
---|---|---|---|
82828 | Billie Eilish | 15449 | 2016-11-26 |
82829 | Billie Eilish | 15526 | 2016-11-27 |
82830 | Billie Eilish | 15643 | 2016-11-28 |
82831 | Billie Eilish | 15740 | 2016-11-29 |
82832 | Billie Eilish | 15843 | 2016-11-30 |
82833 | Billie Eilish | 15945 | 2016-12-01 |
82834 | Billie Eilish | 16028 | 2016-12-02 |
82835 | Billie Eilish | 16111 | 2016-12-03 |
82836 | Billie Eilish | 16187 | 2016-12-04 |
82837 | Billie Eilish | 16299 | 2016-12-05 |
Spend some time on TikTok trying to understand Billie’s presence on the platform, both as a content creator and as someone whose music is used in content. (Unfortunately, we only have data for how Billie’s music is used on the platform, not her official TikTok account.) Explore some of the hashtags for these songs and others like #billieeilishchallenge. How do your observations help to explain the graph below?
ARTIST_NAME | TRACK | TIMESTP | NB_POSTS |
---|---|---|---|
Billie Eilish | idontwannabeyouanymore | 2020-06-08 | 416100 |
Billie Eilish | idontwannabeyouanymore | 2020-06-09 | 416000 |
Billie Eilish | idontwannabeyouanymore | 2020-06-10 | 415800 |
Billie Eilish | idontwannabeyouanymore | 2020-06-11 | 415500 |
Billie Eilish | idontwannabeyouanymore | 2020-06-12 | 415200 |
Billie Eilish | idontwannabeyouanymore | 2020-06-13 | 415000 |
Billie Eilish | idontwannabeyouanymore | 2020-06-14 | 414800 |
Billie Eilish | idontwannabeyouanymore | 2020-06-15 | 414500 |
Billie Eilish | idontwannabeyouanymore | 2020-06-16 | 414100 |
Billie Eilish | idontwannabeyouanymore | 2020-06-17 | 413800 |
Spend some time on Billie’s IG. What kind of content does she post? How often does she post? What kind of engagement does she get? Does any of that help explain this graph?
ARTIST_NAME | FOLLOWERS | TIMESTP | |
---|---|---|---|
8339 | Billie Eilish | 22156 | 2017-03-07 |
8340 | Billie Eilish | 22322 | 2017-03-08 |
8341 | Billie Eilish | 22466 | 2017-03-09 |
8342 | Billie Eilish | 22583 | 2017-03-10 |
8343 | Billie Eilish | 22892 | 2017-03-11 |
8344 | Billie Eilish | 23191 | 2017-03-12 |
8345 | Billie Eilish | 23372 | 2017-03-13 |
8346 | Billie Eilish | 23553 | 2017-03-14 |
8347 | Billie Eilish | 23725 | 2017-03-15 |
8348 | Billie Eilish | 24495 | 2017-03-16 |
Explore Billie’s videos on YouTube, both her official music videos as well other content. For instance, you may want to watch her video on body shaming (https://www.youtube.com/watch?v=ZlvfYmfefSI), or this one of her surprising her fans (https://www.youtube.com/watch?v=uyyQlWNesGM). Notice that you can filter her videos for certain time periods and rank them in different ways. How does what you learned help to explain the graph below?
ARTIST_NAME | DAILY_VIEWS | TIMESTP | |
---|---|---|---|
17689 | Billie Eilish | 17674048 | 2019-07-14 |
17690 | Billie Eilish | 18459967 | 2019-07-15 |
17691 | Billie Eilish | 18844114 | 2019-07-16 |
17692 | Billie Eilish | 18901320 | 2019-07-17 |
17693 | Billie Eilish | 18944697 | 2019-07-18 |
17694 | Billie Eilish | 20234367 | 2019-07-19 |
17695 | Billie Eilish | 20163242 | 2019-07-20 |
17696 | Billie Eilish | 18808721 | 2019-07-21 |
17697 | Billie Eilish | 19672413 | 2019-07-22 |
17698 | Billie Eilish | 19906008 | 2019-07-23 |
Search for Billie’s YouTube channel and explore the content there. How does what you observed help explain the graph below?
ARTIST_NAME | SUBSCRIBERS | TIMESTP | |
---|---|---|---|
50666 | Billie Eilish | 11860720 | 2019-04-25 |
50667 | Billie Eilish | 11971853 | 2019-04-27 |
50668 | Billie Eilish | 12122774 | 2019-04-29 |
50669 | Billie Eilish | 12267148 | 2019-05-01 |
50670 | Billie Eilish | 12413372 | 2019-05-03 |
50671 | Billie Eilish | 12574591 | 2019-05-05 |
50672 | Billie Eilish | 12696028 | 2019-05-07 |
50673 | Billie Eilish | 12750001 | 2019-05-08 |
50674 | Billie Eilish | 12803789 | 2019-05-09 |
50675 | Billie Eilish | 12931710 | 2019-05-11 |
Can you draw any conclusions about what platform she should focus on from what you’ve learned? How else could you look at the data to help you develop your thinking further?
Since the question we want to answer is focused on what Billy should do in the future, we may want to take a closer look at what has happened in the last year. We can filter the data to include only records for 2020, and graph that.
At this point, you are probably starting to compare these graphs to one another to help you understand how Billie is performing on each platform. Are there any further transformations that we could apply to the data to make this information more easily comparable?
One way to approach this problem is to make the YouTube video views a running total (also known as cumulative sum), instead of a daily count, so that it matches the format of the other platforms.
ARTIST_NAME | DAILY_VIEWS | TIMESTP | DAILY_VIEWS_CUMSUM | |
---|---|---|---|---|
17689 | Billie Eilish | 17674048 | 2019-07-14 | 17674048 |
17690 | Billie Eilish | 18459967 | 2019-07-15 | 36134015 |
17691 | Billie Eilish | 18844114 | 2019-07-16 | 54978129 |
17692 | Billie Eilish | 18901320 | 2019-07-17 | 73879449 |
17693 | Billie Eilish | 18944697 | 2019-07-18 | 92824146 |
17694 | Billie Eilish | 20234367 | 2019-07-19 | 113058513 |
17695 | Billie Eilish | 20163242 | 2019-07-20 | 133221755 |
17696 | Billie Eilish | 18808721 | 2019-07-21 | 152030476 |
17697 | Billie Eilish | 19672413 | 2019-07-22 | 171702889 |
17698 | Billie Eilish | 19906008 | 2019-07-23 | 191608897 |
Does having all the data in a similar format help you draw any further conclusions or ask new questions? What are some ways that we could continue to transform the data/visualizations to help us compare these platforms further?
(Note: We will work with the running total of YT video views from now on, so that we keep a standardized format across the platforms. This way, we make sure to compare apples to apples.)
One possible transformation is to combine all the data in one graph so we can look at all these lines overlayed on one another. To do that, we need to make the column names of each dataset the same, and also add a new label so we can distinguish each data source. For TikTok, we will filter for two tracks, BAD GUY and LOVELY, so we don’t have too many TikTok lines clogging up the graph.
Here is each dataset reformatted to contain the same column names and a new column identifying the source.
TIMESTP | RUNNING_TOTAL | SOURCE | |
---|---|---|---|
82828 | 2016-11-26 | 15449 | Spotify Followers |
82829 | 2016-11-27 | 15526 | Spotify Followers |
82830 | 2016-11-28 | 15643 | Spotify Followers |
82831 | 2016-11-29 | 15740 | Spotify Followers |
82832 | 2016-11-30 | 15843 | Spotify Followers |
TIMESTP | RUNNING_TOTAL | SOURCE | |
---|---|---|---|
8339 | 2017-03-07 | 22156 | Instagram Followers |
8340 | 2017-03-08 | 22322 | Instagram Followers |
8341 | 2017-03-09 | 22466 | Instagram Followers |
8342 | 2017-03-10 | 22583 | Instagram Followers |
8343 | 2017-03-11 | 22892 | Instagram Followers |
TIMESTP | RUNNING_TOTAL | SOURCE | |
---|---|---|---|
202 | 2020-01-01 | 796300 | Tik Tok Posts - bad guy |
203 | 2020-01-02 | 796900 | Tik Tok Posts - bad guy |
204 | 2020-01-03 | 798800 | Tik Tok Posts - bad guy |
205 | 2020-01-04 | 800400 | Tik Tok Posts - bad guy |
206 | 2020-01-05 | 801800 | Tik Tok Posts - bad guy |
TIMESTP | RUNNING_TOTAL | SOURCE | |
---|---|---|---|
17689 | 2019-07-14 | 17674048 | YouTube Video Views |
17690 | 2019-07-15 | 36134015 | YouTube Video Views |
17691 | 2019-07-16 | 54978129 | YouTube Video Views |
17692 | 2019-07-17 | 73879449 | YouTube Video Views |
17693 | 2019-07-18 | 92824146 | YouTube Video Views |
TIMESTP | RUNNING_TOTAL | SOURCE | |
---|---|---|---|
50666 | 2019-04-25 | 11860720 | YouTube Channel Subscribers |
50667 | 2019-04-27 | 11971853 | YouTube Channel Subscribers |
50668 | 2019-04-29 | 12122774 | YouTube Channel Subscribers |
50669 | 2019-05-01 | 12267148 | YouTube Channel Subscribers |
50670 | 2019-05-03 | 12413372 | YouTube Channel Subscribers |
Now we can combine all the datasets into a single one and take a look at it on a single graph.
TIMESTP | RUNNING_TOTAL | SOURCE |
---|---|---|
2016-11-26 | 15449 | Spotify Followers |
2016-11-27 | 15526 | Spotify Followers |
2016-11-28 | 15643 | Spotify Followers |
2017-03-07 | 22156 | Instagram Followers |
2017-03-08 | 22322 | Instagram Followers |
2017-03-09 | 22466 | Instagram Followers |
2020-01-01 | 796300 | Tik Tok Posts - bad guy |
2020-01-02 | 796900 | Tik Tok Posts - bad guy |
2020-01-03 | 798800 | Tik Tok Posts - bad guy |
2020-01-01 | 189500 | Tik Tok Posts - lovely |
2020-01-02 | 190500 | Tik Tok Posts - lovely |
2020-01-03 | 192300 | Tik Tok Posts - lovely |
2019-04-25 | 11860720 | YouTube Channel Subscribers |
2019-04-27 | 11971853 | YouTube Channel Subscribers |
2019-04-29 | 12122774 | YouTube Channel Subscribers |
2019-07-14 | 17674048 | YouTube Video Views |
2019-07-15 | 36134015 | YouTube Video Views |
2019-07-16 | 54978129 | YouTube Video Views |
As you can see, YouTube videos have so many more views than the other platforms have followers, subscribers or posts that we can’t really see what’s going on. What are ways that we could transform the data further to make the picture clearer?
We could try filtering out YouTube video views and TikTok posts. This makes sense because Spotify, IG, and YT Channel are all follower/subscriber counts, so they are more comparable to one another than video views or number of posts. Because they are the same kind of measure, they also have a more comparable range.
Can you draw any further conclusions from this graph? Do you have any ideas on how you could transform the data even further to help you look at all the different sources on a single graph that actually makes sense?
One method for achieving this is called normalization, which means transforming the measure variable – in this case, the RUNNING TOTAL – to be expressed as a percent change instead of an actual number. By showing the daily increase or decrease in followers/subscribers/views/posts as a percentage instead of an actual number, we should be able to compare all these sources better.
Let’s add a daily percent change calculation to the data and then graph it.
TIMESTP | RUNNING_TOTAL | PERCENT_CHANGE | SOURCE | |
---|---|---|---|---|
83857 | 2020-01-02 | 18170970 | 0.3071709 | Spotify Followers |
83858 | 2020-01-03 | 18220143 | 0.2706130 | Spotify Followers |
83859 | 2020-01-04 | 18275747 | 0.3051787 | Spotify Followers |
83860 | 2020-01-05 | 18332312 | 0.3095086 | Spotify Followers |
83861 | 2020-01-06 | 18388070 | 0.3041515 | Spotify Followers |
What do you think? Has this helped you compare the sources any better? Any ideas on how to get an even clearer picture of how these sources compare to one another?
How about looking at MONTHLY percent changes instead of DAILY? We can aggregate the data by month, and then graph it.
TIMESTP | MONTH_YEAR | RUNNING_TOTAL | PERCENT_CHANGE | MONTHLY_PERCENT_CHANGE | SOURCE |
---|---|---|---|---|---|
2016-11-26 | 2016-11 | 15449 | NA | NA | Spotify Followers |
2016-12-01 | 2016-12 | 15945 | 0.6438175 | 3.210564 | Spotify Followers |
2017-01-01 | 2017-01 | 18372 | 0.3495740 | 15.221072 | Spotify Followers |
2017-07-01 | 2017-07 | 66228 | 1.4832976 | 22.794527 | Instagram Followers |
2017-08-01 | 2017-08 | 92328 | 0.9071237 | 39.409313 | Instagram Followers |
2017-09-01 | 2017-09 | 147662 | 0.8041834 | 59.931982 | Instagram Followers |
2017-10-01 | 2017-10 | 201138 | 0.8084200 | 36.215140 | Instagram Followers |
2020-08-01 | 2020-08 | 361900 | 0.0000000 | 1.117631 | Tik Tok Posts - lovely |
2019-04-25 | 2019-04 | 11860720 | NA | NA | YouTube Channel Subscribers |
2019-05-01 | 2019-05 | 12267148 | 1.1909320 | 3.426672 | YouTube Channel Subscribers |
2020-01-01 | 2020-01 | 24400000 | 0.0000000 | 6.086957 | YouTube Channel Subscribers |
2020-02-01 | 2020-02 | 25800000 | 0.3891051 | 5.737705 | YouTube Channel Subscribers |
2020-03-01 | 2020-03 | 27200000 | 0.7407407 | 5.426357 | YouTube Channel Subscribers |
That’s a much clearer picture, isn’t it?
That’s it for now! We’ll end our graphical exploration here, but of course, there’s a lot more you could do. Feel free to jot down ideas and share them.
As your formulate your ideas on what platform(s) Billie should focus on for MY FUTURE, you may want to consider some business factors. For instance, Spotify streams generate a lot more money for the artist than YouTube or TikTok, and IG doesn’t generate any money at all. However, all of these platforms can offer a big audience boost to Billie, which in turn can lead to more paid streams.
We suggest not working on all this in one go – let your ideas “bubble up” and come back to the graphs and the content over time to explore them further.
Have fun! Can’t wait to hear what you think about Billie’s social and streaming strategy for this new song.