We all have that one musically inclined friend in our circle. Who recommends new music. Belts out every word of every song in the pub, as you struggle to keep up. Who wipes and offers you their headphones, asking you to catch up to that insane new artist, right?
Spotify's working hard to replace that friend, promising to punctuate the monotony of your existence with a heady mixture of sound and melody. Unlike your real friend (or Diya Koshy, Nikhil Pradhan & Siddharth Parwatay, in my case), this one's music recommendation game is always on point, is pocket-friendly, and it won't pass out or get into a brawl with the pub's DJ.?
A lot has already been written about Spotify -- how it's one of the leading music streaming services in the world, has a cult following among its users, curates over 3 billion playlists, and is probably the best music discovery service you'll ever come across.
Reuters
I had a chat with Owen Smith, Spotify's Director of Product Growth, about the service's much-awaited India launch and also try to understand how it all works. He oversaw the planning and launch of Spotify in India, and had some interesting insight on the challenges of building Spotify for India.
"The overarching challenge was: India is like many, many different people and many, many languages, with such a rich music culture," explains Owen Smith. "Our challenge wasn't to make the same Spotify for everyone, we wanted to make Spotify feel like it was made uniquely for every single user. So how do you do that for a country of 1.3 billion people was our fundamental challenge that we've we set out with."
Owen feels launching in India has been Spotify's biggest expansion yet, in terms of research, preparation and local development.
Spotify
"I think we really gave it our all," Owen said. It involved multiple visits to meet users, fans and artists across different cities from Mumbai, Delhi to Chandigarh, and more. "It basically involved asking them how they listen to music, when they listen to music, what music they listen to, what they liked and didn't like about the way they listen to music, and then collecting all of that feedback," recollects Owen.
Ensuring they understood the relationship between music of different languages and the different users was key to getting the Spotify India experience right, according to Owen Smith. Indian users are very particular about their music's language buckets, he recollects. "Users told us that they really don't want their Hindi music and Punjabi music in the same Mix. So that's a simple thing for us to do, we just give you different language clusters of music up front."
Another thing to figure out for India was how Spotify should be different for every user. "How your Spotify experience, should be different than my Spotify, and your mother's Spotify, and so on," Owen mentions. "But then we have to figure out what that means for India. What relationship does one actor's music have to another actor's music? What relationship does this language have to another??For this you need lots of pieces of data about Indian music and its relationship with the listener."
Building up all those relational data pieces and making sure "we can tweak our algorithms so that on day one we could actually deliver an experience that's personal and made for each individual user" was no easy task, as per Owen Smith.
Spotify's integration into the Indian music-o-sphere was never going to be complete if Owen and his team didn't find a way to fine-tune music discovery and recommendation within the app. This is where Spotify's playlists had to shine under the spotlight. And "collaborative filtering," too.?
Spotify has over 3 billion curated playlists of music. "It's both users, it's brands, but it's also our in house editorial experts," Owen describes. "So three different types of curation making up the biggest library of playlists in the world. There's a playlist for literally everything you could ever want and every occasion."
Now, Owen and his team's job with the algorithm is to find if you like a playlist. "If you like the Bollywood romance playlist, for example, and we find that you and I have similar tastes, but I've never listened to the Bollywood romance playlist. Then the algorithm will recommend that playlist to me." This is what Spotify calls collaborative filtering, learning from each user's musical tastes and exposing them to newer music based on collaborative filtering. This is something that Spotify has seen huge success with, says Owen.
But collaborative filtering for India is also sensitive to language preferences, and makes sure just because someone in Punjab listening to Punjabi music and someone in Tamil Nadu listening to Tamil music don't get music recommendations from opposite languages just because they might have Alternative Rock as a common musical preference.
"When it comes to daily mixes," Owen says, "we don't just literally give you a playlist with the artists that you've chosen. Because that would be boring. We also take a look at similar artists and give you suggestions."
So the algorithm looks at what other music fans of your selected artists also play, and then it recommends a couple of tracks from those other artists that Spotify thinks you may like, and then the algorithm watches how you react to the new music.
"Everything from whether you like a playlist or song, whether you skip a song in that playlist, or whether you don't finish listening to a song. All these are signals that you as a user give to Spotify's recommendation algorithm," Owen reveals. "And over time, your daily mix will evolve, you will start seeing less of the tracks and music you skipped, and start seeing more of the tracks and more similar tracks that you hearted and listen to all the way to the end. So that's kind of the general principle of the whole product as well."
So from the type of music, to artist, and band, even languages and locations, Spotify continues to learn what you like over time, and recommends you music you'll most probably love all the time.?
"Sounds a lot like Netflix for music, doesn't it?" I ask Owen. He thinks before answering, "If it's about taking a lot of signals from users and trying to understand what they want, in terms of content, then yeah you can make that comparison."
Just when I think I've scored a point, Owen chimes in, "But I think we have a much bigger catalogue than Netflix." I think about Spotify's incomplete music catalogue compared to the likes of Gaana and Apple Music, and the tall order the service faces in India's music streaming war, and I let Owen enjoy his moment.
It's still early days for Spotify in India, and I'm definitely not cancelling my Apple Music or Gaana subscription, choosing one over another. For someone who never had a shoebox full of cassettes or CDs, whose music discovery arc ended with Shenaz Treasuryvala's MTV's Most Wanted, who greatly relied on friends for musical cues -- I? think Spotify will help me miss Diya Koshy, Nikhil Pradhan and Siddharth Parwatay a little less. And who knows, even help me widen their musical horizons at our next get-together.