7AM

Time to wake up, grab a shower and get my train. I live in Newton, Massachusetts, which means a half-hour ride into work. And I spend that time bouncing around between various streaming and podcast platforms on my phone—half to entertain myself, half as market research. It’s a great opportunity to put myself in the shoes of someone who doesn’t work at Spotify, listen with fresh ears, and figure out how to improve things for everyone in the outside world.

9AM

When I get to the office, I have a quick breakfast then get down to business. I work in the part of Spotify that builds human + AI experiences to help listeners discover music they’ll love, and our team figures out how to do this via Voice assistants in our app or on voice-controlled speakers. As a Group Product Manager, I keep track of the projects we have in flight, the explorations we’re doing and the strategy we’re trying to pursue, whether that’s broadly in the Personalization department or specifically in Voice. And on any given morning, I could be getting together with our data scientists and user researchers to discuss a feature we’re building and how people are reacting to it. I could be sitting down with our engineers to figure out a thorny technical problem. Or I could be meeting with other product people to think about our overall strategy and plan out the weeks, months and quarters ahead.

One of the fun challenges in Voice is that natural language is inherently ambiguous, meaning that the same statement can often have multiple meanings. For instance, imagine that someone asks a Voice speaker to ‘play work from home’. Are they referring to the Fifth Harmony song ‘Work From Home’? Or are they asking to play the Rihanna track ‘Work’ from the speaker they’ve named ‘Home’? Figuring out all the possibilities is a puzzle that lies at the intersection of linguistics, technology, and personalization, and we’re always trying to get better at it. Recently, our team came up with a new strategy and deployed it successfully to a small percentage of our audience—soon, we’ll be rolling it out to everyone. This kind of rapid experimentation at scale makes it possible to make lots of small bets, then roll out the winners and cheaply discard the losers. It’s way better than spending months building a monolithic solution and just crossing your fingers that it works.

12PM

I usually break for lunch around noon and head to the 8th-floor cafeteria to eat with colleagues. Although if I’m mulling over some really tricky strategy problem, I’ll sometimes go to the gym in the building instead—it’s a good way to clear my head and reset before getting stuck back into work again.

If I’m in need of an extra boost, I often check out what’s playing on the office music system too— we have an internal-only piece of software, which was hacked together by a bunch of engineers years ago. This system lets any employee add tunes to a collaborative music queue that plays throughout the office. It’s actually one of my go-to sources of music discovery—people get excited about sharing their favorite tunes and there’s such a diverse set of tastes represented.

1PM

My afternoons involve a little more thinking and writing time, as well as some more exploratory, experimental stuff if I’m lucky. Something we do regularly are ‘Soundchecks’, in which squads can submit prototypes or features in development for user research testing and get immediate feedback from six real-life Spotify listeners. It’s a super-interesting way to understand how people incorporate our product into their lives and see how they react to ideas in a really fast feedback cycle.

These sessions are also great because they encourage teams to be more experimental and scrappy, and create a low-investment, low-stakes environment that allows them to fail fast. I think it’s great that this kind of thing is so well supported at Spotify—the people who succeed here are the ones who are passionate about an idea and find some way to do it. It’s really inspiring, and it’s one of the things that’s kept me working here so long.

It’s also really nice to be involved with a product you’d use even if you weren’t working at the company. I think that’s pretty rare actually. The things we do impact the lives of listeners and creators everywhere. And I really respect the way our leadership encourages us to think about this and ensures our work has a good net impact on the world.

6PM

I wrap things up about 5:30 or 6:00pm and head home on the train—I try not to do too much work on weeknights these days because I’d much rather spend time with my wife and two-year-old son. Having kids has really affected the music I listen to—nowadays I have a whole new appreciation for bands like The Beatles, who are innocuous enough for small children but entertaining enough for the whole family to enjoy! If you compare my top songs from a few years ago with my top songs of today, it’s like night and day, I can tell you.

Tags: backend

Published by Spotify Engineering