Oct 26, 2017 · 24m · mad

You Are What You Stream // Christine Hung, Spotify (FirstMark's Data Driven)

Christine Hung · 19m spoken Matt Turck · 53s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this presentation at FirstMark's Data Driven NYC, Christine Hung, Head of Data Solutions at Spotify, explains how Spotify leverages its massive streaming dataset and machine learning models to decode human behavior, uncover authentic life context, and deliver personalized user experiences.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 4.1% of the talking time here. How this is scored →

Matt as informed peer 0.6 Guest teaching 1.6 Guest disagreement 0.2 Matt pushing back 0.4
05100:0010:0020:000:42–3:07 · Matt as informed peer 0/10 Spotify's Scale and Streaming Data Footprint This segment is a presentation monologue by guest Christine Hung outlining Spotify's scale and data footprint. The host is not present or speaking during this monologue segment, requiring host scores to be zero.3:07–6:11 · Matt as informed peer 0/10 Stated Preferences vs. Streaming Reality: Personal Case Study Christine Hung continues her presentation monologue, contrasting self-reported listener preferences with actual streaming behavior using her personal data. The host does not participate.6:11–8:21 · Matt as informed peer 0/10 Streaming Intelligence Framework and Life Moments The guest details how listening habits allow Spotify to map user personality traits and key life moments. Host activity remains non-existent during this monologue section.8:21–14:35 · Matt as informed peer 0/10 Machine Learning Methodology and Xbox Experiment Christine explains machine learning testing methodologies, Xbox cross-device targeting, and nostalgia experiments. As a monologue segment, host interaction scores are zero.14:35–21:24 · Matt as informed peer 3/10 Audience Q&A: Recommendations, Personality, and Creator Focus Host Matt Turck opens the Q&A and mistakenly asserts Spotify generates personalized playlists per day, which Christine gently corrects as being weekly. Matt asks informed technical questions on editorial versus algorithmic balancing and contextual data integration.0:42–3:07 · Guest teaching 1/10 Spotify's Scale and Streaming Data Footprint This segment is a presentation monologue by guest Christine Hung outlining Spotify's scale and data footprint. The host is not present or speaking during this monologue segment, requiring host scores to be zero.3:07–6:11 · Guest teaching 1/10 Stated Preferences vs. Streaming Reality: Personal Case Study Christine Hung continues her presentation monologue, contrasting self-reported listener preferences with actual streaming behavior using her personal data. The host does not participate.6:11–8:21 · Guest teaching 1/10 Streaming Intelligence Framework and Life Moments The guest details how listening habits allow Spotify to map user personality traits and key life moments. Host activity remains non-existent during this monologue section.8:21–14:35 · Guest teaching 1/10 Machine Learning Methodology and Xbox Experiment Christine explains machine learning testing methodologies, Xbox cross-device targeting, and nostalgia experiments. As a monologue segment, host interaction scores are zero.14:35–21:24 · Guest teaching 4/10 Audience Q&A: Recommendations, Personality, and Creator Focus Host Matt Turck opens the Q&A and mistakenly asserts Spotify generates personalized playlists per day, which Christine gently corrects as being weekly. Matt asks informed technical questions on editorial versus algorithmic balancing and contextual data integration.0:42–3:07 · Guest disagreement 0/10 Spotify's Scale and Streaming Data Footprint This segment is a presentation monologue by guest Christine Hung outlining Spotify's scale and data footprint. The host is not present or speaking during this monologue segment, requiring host scores to be zero.3:07–6:11 · Guest disagreement 0/10 Stated Preferences vs. Streaming Reality: Personal Case Study Christine Hung continues her presentation monologue, contrasting self-reported listener preferences with actual streaming behavior using her personal data. The host does not participate.6:11–8:21 · Guest disagreement 0/10 Streaming Intelligence Framework and Life Moments The guest details how listening habits allow Spotify to map user personality traits and key life moments. Host activity remains non-existent during this monologue section.8:21–14:35 · Guest disagreement 0/10 Machine Learning Methodology and Xbox Experiment Christine explains machine learning testing methodologies, Xbox cross-device targeting, and nostalgia experiments. As a monologue segment, host interaction scores are zero.14:35–21:24 · Guest disagreement 1/10 Audience Q&A: Recommendations, Personality, and Creator Focus Host Matt Turck opens the Q&A and mistakenly asserts Spotify generates personalized playlists per day, which Christine gently corrects as being weekly. Matt asks informed technical questions on editorial versus algorithmic balancing and contextual data integration.0:42–3:07 · Matt pushing back 0/10 Spotify's Scale and Streaming Data Footprint This segment is a presentation monologue by guest Christine Hung outlining Spotify's scale and data footprint. The host is not present or speaking during this monologue segment, requiring host scores to be zero.3:07–6:11 · Matt pushing back 0/10 Stated Preferences vs. Streaming Reality: Personal Case Study Christine Hung continues her presentation monologue, contrasting self-reported listener preferences with actual streaming behavior using her personal data. The host does not participate.6:11–8:21 · Matt pushing back 0/10 Streaming Intelligence Framework and Life Moments The guest details how listening habits allow Spotify to map user personality traits and key life moments. Host activity remains non-existent during this monologue section.8:21–14:35 · Matt pushing back 0/10 Machine Learning Methodology and Xbox Experiment Christine explains machine learning testing methodologies, Xbox cross-device targeting, and nostalgia experiments. As a monologue segment, host interaction scores are zero.14:35–21:24 · Matt pushing back 2/10 Audience Q&A: Recommendations, Personality, and Creator Focus Host Matt Turck opens the Q&A and mistakenly asserts Spotify generates personalized playlists per day, which Christine gently corrects as being weekly. Matt asks informed technical questions on editorial versus algorithmic balancing and contextual data integration.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%12:00 · Matt 10.7% · guest 89.3%12:00 · Matt 10.7% · guest 89.3%15:00 · Matt 8.3% · guest 91.7%15:00 · Matt 8.3% · guest 91.7%18:00 · Matt 0.2% · guest 99.8%18:00 · Matt 0.2% · guest 99.8%21:00 · Matt 13.9% · guest 86.1%21:00 · Matt 13.9% · guest 86.1%24:00 · Matt 2.3% · guest 97.7%24:00 · Matt 2.3% · guest 97.7%
Sharpest disagreement ▶ 14:49 Gently correcting host playlist frequency assumption

In an otherwise completely non-confrontational presentation and Q&A, the guest's most assertive moment occurs when she directly corrects the host's mistaken belief that Spotify creates daily personalized playlists.

Hardest push from Matt ▶ 15:01 Host questioning data-driven vs editorial balance

Host Matt Turck pushes past the presentation praise to probe whether Spotify's curation and top hits lists are purely algorithmic or influenced by human editors.

Biggest teaching moment ▶ 14:49 Clarifying weekly vs daily recommendations

The host praises the app based on his assumption that playlists update daily, prompting Christine to inform him that personalized recommendations are currently strictly weekly.

Matt holds his own ▶ 20:59 Host inquiring about external contextual data sets

Host Matt Turck demonstrates tech industry expertise by asking how Spotify overlays external contextual data streams, like weather, onto first-party listening behavior.

the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Spotify's Scale and Streaming Data Footprint 0100 This segment is a presentation monologue by guest Christine Hung outlining Spotify's scale and data footprint. The host is not present or speaking during this monologue segment, requiring host scores to be zero.
Stated Preferences vs. Streaming Reality: Personal Case Study 0100 Christine Hung continues her presentation monologue, contrasting self-reported listener preferences with actual streaming behavior using her personal data. The host does not participate.
Streaming Intelligence Framework and Life Moments 0100 The guest details how listening habits allow Spotify to map user personality traits and key life moments. Host activity remains non-existent during this monologue section.
Machine Learning Methodology and Xbox Experiment 0100 Christine explains machine learning testing methodologies, Xbox cross-device targeting, and nostalgia experiments. As a monologue segment, host interaction scores are zero.
Audience Q&A: Recommendations, Personality, and Creator Focus 3412 Host Matt Turck opens the Q&A and mistakenly asserts Spotify generates personalized playlists per day, which Christine gently corrects as being weekly. Matt asks informed technical questions on editorial versus algorithmic balancing and contextual data integration.

Statements from this episode (10)

Assertion Supported
Christine Hung: Spotify has 150M monthly active users across 60+ countries
“At Spotify alone, we now have a hundred and a hundred and fifty million active monthly users streaming from more than 60 countries.”
Christine Hung Oct 26, 2017 ▶ 1:02
Insight
Music streaming doesn't compete directly with social media for user attention
“The interesting thing about music is that we're actually not competing as much as you think, and that's because music is complimentary.”
Christine Hung Oct 26, 2017 ▶ 1:40
Assertion Not checkable as stated
Hung: Spotify streaming data enables mapping of critical user life moments
“As we start to capture different types of user behavior, we are able to start mapping out critical moments of your life.”
Christine Hung Oct 26, 2017 ▶ 7:22
Assertion Not checkable as stated
Hung: Predictive targeting for Spotify Xbox launch boosted engagement and lowered opt-outs
“So we saw a significant increase in cross-device usage, which is what you always want. We saw a much higher engagement rate, and we were able to dramatically reduce the email opt-out rates as well, because the messaging was really relevant.”
Christine Hung Oct 26, 2017 ▶ 9:38
Assertion Not checkable as stated
Spotify listeners consistently return to music from their teenage years
“Regardless of how old you are, at least on Spotify, people keep going back to the music that they came out during their high school, during their teenage years”
Christine Hung Oct 26, 2017 ▶ 12:48
Assertion Supported
Spotify Discover Weekly is purely algorithmic without editorial intervention
“Discover Weekly, for example, everyone's Discover Weekly is different, right? So, there's no way that we can really have, sort of, editorial, ah, intervention. So, everything is purely based on what you've been listening to”
Christine Hung Oct 26, 2017 ▶ 16:03
Assertion Supported
Spotify listening data can accurately predict user personality traits
“We actually had some sort of collaborations with academic institutions, and, you know, we can actually pretty accurately predict people's personality just based on the data, right?”
Christine Hung Oct 26, 2017 ▶ 16:58
Prediction Not checkable as stated
Spotify views AI taking over music composition as years away
“We have Internal discussions about, like, what happens when machines take over, but I would say our view is that that's probably still years away.”
Christine Hung Oct 26, 2017 ▶ 19:48
Disclosure
Spotify plans to shift focus toward music creators over the next decade
“And I would say for the next 10 years, we want to focus more on the creators, right?”
Christine Hung Oct 26, 2017 ▶ 20:15
Assertion Not checkable as stated
Spotify analysis shows weather changes directly affect user music choices
“Weather does affect how people listen. We had a partnership with AccuWeather, I think it was last year, and we did some analysis just in Chicago to understand, so when the weather changes, right, do people change what they listen to, and the answer is yes.”
Christine Hung Oct 26, 2017 ▶ 21:31
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.