Thanksgiving is over, a chill is in the air, and it’s now that very special time of year – the time when Spotify tells us things about ourselves that we’re not ready to hear:
It sometimes feels like Spotify Wrapped knows us better than we do. But its seemingly magical insight is made possible by a good, structured data model.
A simple data model like this goes a long way:
|
ID
|
Start Time
|
Title
|
Artist
|
Play Duration
|
Genre
|
|---|---|---|---|---|---|
|
1
|
12/01/23 1:00 PM
|
Jingle Bells
|
Frank Sinatra
|
3 mins, 15 secs
|
Christmas
|
|
2
|
12/02/23 10:00 AM
|
Deck the Halls
|
Nat King Cole
|
2 mins, 30 secs
|
Christmas
|
|
3
|
12/03/23 5:30 PM
|
Rudolph the Red Nosed Reindeer
|
Gene Autry
|
4 mins
|
Christmas
|
This basic structure can produce most of a Spotify Wrapped with just simple counts and sums. For example, here are some of the insights included in a Wrapped:*
*If you see anything embarrassing in the listening history, don’t judge Eric – he has children.
You can get this information by just adding columns and counting rows.
And sorting and taking the top of the list will get you these:
And create a fancy-looking heatmap by mapping monthly distribution of listening, and putting a little dot on the highest one:
In spite of the fact that 7 in 10 people are too embarrassed to share their Spotify Wrapped, it’s estimated that Wrapped graphics and information were shared over 60 million times in 2022. (We still don’t have an estimate on how often people shared their Wrapped data this year, but it’s safe to bet that the number isn’t too different.)
Spotify Wrapped drives a ton of user engagement, and deepens relationships with subscribers. Turns out, people love seeing their data and quantifying their behavior!
But beyond just repackaging data, Spotify also introduces Wrapped elements that encourage subscribers to engage with the core product. For example:
By assembling a list of songs that a subscriber listened to frequently, Spotify has created an enticing playlist that this listener will want to return to – plus, they’ve included a CTA to remind them to add it to their library, so they’ll be even more likely to look at it again. And by providing listeners with personalized, exclusive content from their favorite artists, Spotify is creating yet another point of active interaction for their subscribers.
Creating opportunities for subscribers to engage with the platform through Wrapped keeps them invested, and it helps make Spotify “sticky”.
But back to the data. We’ve made it through most of my Wrapped without doing anything more complex than counting, summing and sorting, but there are a few more advanced tricks up Spotify’s sleeve.
Look-Alike Modeling
Spotify came up with geographic profiles of how people listen in various parts of the world, and matched Eric to the one closest to his behavior. It then pointed out how although he may act like a Quebecois, there are some differences:
Behavioral Clustering
The last piece here is as much art as science. Spotify is categorizing the music users listen to by “vibe”, grouping users with similar taste, and assigning personas based on how the groups differ from the others:
Are these personas a little arbitrary? Sure. But they’re also a lot of fun. And like a personality test, these personas scratch our itch to know more about ourselves.
It’s arguable that Spotify Wrapped is one of the most effective ways that user data is leveraged to engage subscribers, because it gives people deeper insight into their behavior. So it’s worth asking – how can your business engage customers by giving them insight into their own information?
