The Magic Formula: How Does Spotify’s Algorithm Work?


Written by:

Have you always wondered how Spotify manages to recommend you that perfect song or playlist?

The magic of data has allowed major streaming service, Spotify, to accumulate over 248 million active users since late 2019. With their innovative algorithm system, it makes a far more desirable listening experience for their users compared to their rival counterparts, Amazon and Apple.

As if we were under one big microscope, Spotify’s algorithm examines and studies our listening habits, songs we skip, our social media activity and even where we live; and recommends personal playlists that is acquired to only your taste (and the other users with the same music taste as you.)

The Magic of Discover Weekly

Spotify’s ‘Discover Weekly‘ encourages users to find a fresh and insightful playlist each week that seems familiar and brand new at the same time.

Like a huge melting pot of music, it uses the magic of collaborating millions of playlists from other users and your own music profile of preferred playlists, to give you 30 undiscovered songs for you to listen and favourite simultaneously. Better save your favourites, because they’ll be replaced same time next Monday …

Everything we do – from talking to Siri – to publishing a fancy blog post about a particular artist –

is all examined and fed back to the system to engulf so it can recommend more and more music to expand our musical knowledge as Spotify users.

Discover Weekly is clearly a favoured project for Spotify users, as it racks up over a billion streamed songs across the week since release on the site.

It’s possibly why we are so enthralled with Spotify when it delivers our ‘Decade Wrapped Playlists‘ at the end of year – we simply love finding out about what we love to listen to. We can’t get enough of the numbers.

How Powerful is AI in Music?

Spotify’s algorithm is governed by the AI system, named as BaRTBandits for Recommendations as Treatments. This is the key system in how the algorithm functions, in which it recommends songs based on what you’ve listened to prior, but also recommends fresh music in which you haven’t heard of before, just so it gives you a chance to escape the listening loop you’ve found yourself in.

BaRT also works on a 30 second rule to decipher what is accepted and what is not. If the listener skips the second after 30 seconds, Spotify then takes it as a check on their recommendations. The longer you spend on a recommended song, the more chances of this type of music will appear in your Daily Mixes and Discover Weekly’s. So, if you love something, let it play out!

How Does it Recommend New Artists to us?

The way Spotify recommends the new artists is to actually analyse the audio itself, in an effort to recognise different desirable characters to music. These ‘desirable characters’ of music – from specific genres, tones, tempo and even that of a guitar fuzz – are then clumped together for a desirable ‘Discover Weekly‘ for new artists to be discovered from Spotify users.

We now have more technology than ever before to ensure that if you’re the smallest, strangest musician in the world, doing something that only 20 people in the world will dig, we can now find those 20 people and connect the dots between the artist and listeners…

Matthew Ogle, 2015, ex-product director of Discover Weekly

What does your ‘Discover Weekly’ look like?

With us delving a bit deeper into how it all works; it may be worthwhile to dive into your own ‘Discover Weekly.’ It is almost uncanny – and slightly disconcerting – that the unheard titles of Egyptian Blue’s Nylon Wire, Delaire The Liar’s Shovel and Weird Milk’s Anything You Want are actually surprisingly enjoyable to listen to. I mean, I suppose that is the point.

Otherwise, I feel that the algorithm may be displeased with any other news. It is designed to suit the user’s most favoured after all.

At the end of it all, you are often sat there wondering how on earth they have managed to conjure up a playlist that is so, well, me.

Once you delve further, it is quite scary how well Discover Weekly matches to your music preferences. Some are hit or miss, truth be told. Like that from, Hotel Lux’s Tabloid Newspaper and Big Spring’s Too Late from my own. But a few will miss the cross checks, no doubt.

As well as this, it has been found that sometimes, users do share similar – if not identical – Discover Weekly playlists each week. And here, you’d have to ask; did two billion playlists really churn out the same two playlists for different users, or is there something else at play here?

You can argue that Spotify may be purposely feeding those particular songs via certain prompting packages with certain labels.

You could argue that. But, Spotify have repeatedly said that they have never intentionally supplied songs into playlists, despite many requests.

“Some people have said, ‘Oh, all three of us had this track on our Discover Weekly, did someone put it there?’” “And the answer is yes, someone put it there: other Spotify users who were {play-listing}, which means that something is happening in music culture, in the world.”

Matthew Ogle, ex-Product Director of Discover Weekly

Of course, if some users listen to the same music, it is highly likely that they will end up with the same songs in the playlist. There’s only so many songs, after all.

Nonetheless, Spotify’s algorithm is their trophy and sets the standard for music fans to find new music to their tastes without having to trudge and search through countless playlists. It may show just how addicted the music industry is to data, but regardless of the fact, it’s a marvellous phenomenon and I, for one, have discovered hundreds of new artists that have become songs on repeat for me.

Discover yours below on the link:

Articles for further information:

How Spotify’s Algorithm Manages To Find Your Inner Groove

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Latest Stories

%d bloggers like this: