Using data to help artists maximize their potential | Adam Husein @ Firebird | Data Science Hangout

Transcript#

This transcript was generated automatically and may contain errors.

Hi everybody, welcome to the data science hangout. I guess welcome back because we weren't here last year or last week for the 4th of July. But I'm Rachel Dempsey. I lead customer marketing at Posit and Posit's the open source data science company building tools for the individual team and enterprise. So thank you so much for hanging out with us today.

This hangout is our open space to hear what's going on in the world of data across different industries and to get to connect with others who are facing similar things as you. And we get together here every Thursday at the same time, same place as long as it's not a holiday. But if you're watching this as a recording in the future on YouTube, and you want to join us live, there'll be details to add it to your own calendar below. And I just always want to add, make sure it adds for 12 to 1 Eastern time so that you can join us live.

I know people really love connecting with other attendees here. So if you are interested in connecting with others, I'll encourage you to say hi in the chat, maybe share your LinkedIn or introduce yourself or your base something you do for fun. But we're all dedicated to keeping this a friendly and welcoming space, the welcoming space that you all have made it and love hearing from you no matter your years of experience, titles, industry or the languages that you work into.

I'm so excited to be joined by my co host today, Adam Hussein, SVP of data and analytics at Firebird music. And Adam, to get us started here, I'd love to have you introduce yourself and share a little bit about your role, but also something you do for fun outside of work too.

Sure. So my role, I had all data and analytics and cloud technology for Firebird. And something I do for fun is I am a classically trained pianist. I play piano sometimes. And, you know, when I have three young kids, so when they're not all trying to call on top of me and play the piano with me at the same time.

Overview of Firebird and the music industry

So let me give a little bit of an overview of Firebird just for people who may not be familiar. We're a pretty new company and, um, you know, we are about two and a half years old now coming up on three years in October. And the, um, you know, the company is small and growing. So we're, you know, just over around 40 people right now. And, uh, what we are is we're kind of partnering with a lot of different companies, um, either as an investor or as an affiliate or, um, music management is one of our primary areas. So that's like red light management, which is the largest management company out there.

And so, um, so that's primarily our, uh, our spaces. We've also recently gotten to the live event space too, and invested in a particular festival that we're working with. You know, the main thing is that, you know, across the music ecosystem, there's labels, there's management, there's also publishing companies. And we have a couple of those too, uh, like One Too Many and, um, and Tape Room, which is the, uh, Ashley Gorley's publishing company where, um, you know, they have like over 40 number one country hits. Uh, that new one, um, Post Malone's, uh, country single that's at the top of the charts, like Ashley Gorley was the writer on that one.

And our goal really is to create a cohesive ecosystem for the artists where, you know, they can kind of be in one place and we can have incentives to work with them. The canonical problem I always talk about with people is, you know, you'll go to your record label, like your Universal or your Warner to create your album and you might have a multi-album deal there. You'll put out the album and then you'll go on tour to promote the album and you'll have a promoter like the Live Nation or maybe a smaller promoter or a series of promoters and your manager and your agent will work together with them to really coordinate your tour.

And the economics for the music industry really shifted over the last 30 years. And so now the artist is making most of their money from the tour and not that much from their recorded music. So you can think about it really clearly in the nineties when we used to buy albums, they would cost 15 to $20 for an album. Now you pay $11 a month for Spotify and you get unlimited music and you get to listen to as many songs as you want. So the economics are very different now for an artist than they used to be.

Now you pay $11 a month for Spotify and you get unlimited music and you get to listen to as many songs as you want. So the economics are very different now for an artist than they used to be.

And this is why music management is our largest investment is because it's core to our thesis of we want to partner with the artist and then make money when the artist makes money so that our incentives are aligned with the artist. And that goes back into what we're doing in data science analytics here is really our goal is to help the artists connect with their fans, grow more fans, deepen their relationships with their fans, and then build a longer lasting, more profitable and more impactful career.

I find that there's a little bit of a curse of data sometimes, where the bigger your internal data, the more likely you are to only look at your data for answers. And there's really a wealth of data out there that you can use for augmenting your internal data and helping you build models and make decisions.

And I would say that, you know, I spent a lot of time early on on Kaggle and, you know, that's really where I thought a lot about, you know, there were these competitions with external data. And, you know, it really taught me to think creatively about what other data could you bring to solve these problems. Whereas, you know, like when I would work in companies, I was like, oh no, we have the Call of Duty data. We know everything everybody did on our platform. Why do we need additional data? And I'm like, well, we had a 20% drop in users and nothing changed in our game. And another competitive game came out.

Communicating difficult findings

That's a great question. I would say that that's not just in music, that's everywhere, where, you know, when you think about it, sometimes as data professionals, we are the bearer of bad news. And sometimes we're the bearer of something unexpected, which could be really exciting. And so it's a double-edged sword, right?

I think the, you know, where something didn't go their way, right, or something wasn't what they're expecting, I would say in music, people are pretty positive. Like, you know, it's not that difficult to produce new music. You're not spending hundreds of millions of dollars on it, like a film. I would say it was a lot tougher in film and TV. Like, if you're working on a marketing campaign for a film where we've already spent $150 million, we're spending $50 million to market this film, and like, no one is showing any interest in seeing this film. But it's, those are tough conversations to have, where you have to go, like, I'm sorry, I know that we're really optimistic about this movie, but the numbers clearly indicate that, you know, we are way behind on presales, from where we should be.

It's one of those things where I have learned that different people respond to that information in different ways, and you have to be attuned to that type of person, and understand what type of person you're communicating with. Some people, you know, they're very receptive, and they're open. And with those people, it's good to share additional context, and, you know, make sure they understand why other people, they shut down, become very defensive, and feel like you're attacking them. And so if it's that kind of person, you have to be very careful when sharing negative information. Don't only share positive information, that's a mistake. But you may need to, like, couch it in softer terms, and make sure that you, you know, give something that seems more like a balanced perspective.

Working with artists and educating teams

Yeah, for sure. So, so for the first one, I would say that typically we have a, we work with artist managers directly, not the artists as much. The artist managers, we have an affiliate program where sometimes, you know, if they don't want to go join a bigger company or something, they can affiliate Firebird. And that's typically how we work with them.

Yeah. So I would say that it's a combination of when we roll out the tools to people, we kind of walk them through how they work and educate them a little bit. And we always answer questions or do follow-ups and that's helpful. You know, I have a neuroscience background from my time at UC San Diego. And so one of the most important things we learn in neuroscience is, you know, that the repetition really breeds learning. And so, you know, you want to strengthen those neural connections between the different areas. And so it's important to have follow-ups and touch bases and make sure that people are finding it useful.

And then reinforcing those concepts and tying things together is another way that, you know, really helps with learning. You know, if you have one concept, you understand you tie things to that concept. You know, there's a concept of like a neural attractor, but basically like you can think of it as if you tie things to other concepts, it makes it easier to remember because it's part of something that you already know. And then in the tools themselves, we had little tool tips and things like that. Sometimes with question marks, Rebecca, my team does a great job at this. And so I'm putting these in places you can mouse over and it explains to you, this is the definition of this metric or, you know, in context.

Modeling and data quality

Yeah, that's a great question. I would say that that's not just in music, that's everywhere. You know, as I imagine that occasionally you get big events like, you know, Taylor Swift announces her Eras tour or Coldplay drop a new album, which seem, you know, has potential to disrupt large parts of supply chain merch, you know, production, vinyl pressing and so on. So do your models allow for those kinds of big events happening?

Yeah. So our label service team does all the we have Scott Bergman, who is the VP of physical over there, who does all the like vinyls and everything like for Slash's album that I mentioned and, you know, a few others. And so, you know, we've discussed this a little bit. I think the issue there is that the release schedules for these things aren't announced that far in advance. And so it's difficult to do that. We've talked about building a predictive model to predict kind of, you know, there's some seasonal trends, but there's also like just a lot of variation, right? I mean, you can think of it like a bite to eat eye model, right? For lifetime value for those who've done lifetime value modeling, you know, it's kind of like there's a like a distribution that you can figure out of how often artists are releasing albums. Some people release really often and it's like clockwork and some people it's like, hey, it's been six years, maybe there's an album coming out, you know?

So definitely the streaming data is messy because it comes in a lot of different formats, and, you know, there's not much of an incentive for the DSPs to really clean a lot of that up and make it easy to drive insights from it. And so, in talking to other people, I find that, you know, some of the larger labels, they tend to just, like, archive it and use it for reporting because of that and not even do any modeling off of it.

I would say the other thing is when you get data from different data providers, like third parties that you license data from, it can come in various forms, and so sometimes you can get really clean, easy-to-use data that, you know, you just run a query on and it just works. I would love to meet those providers because most of mine, there's still work we have to do with modeling the data, cleaning it up, enhancing it, augmenting it. Like you get an album and it's like, who's the artist that is this album actually being released or a song? You know, it's just a list of artists. It's like, who is the one that actually was the primary artist on this track?

With other data sources on concert ticket sales, it's like, great, here's the name of the event and on the secondary market, and, you know, apparently on the secondary market, whoever's plugging the events on Subhub or VividSeats, one of those platforms, decided to put Death Cab Ampersand Postal Service on one, Death Cab A&E Postal Service on another one, Capitalize on some, lowercase on others, Slash on some in between Death Cab and Postal Service. So, and then, you know, there's festivals and all sorts of stuff, and so there's lots of data cleaning and enhancement that's needed on a lot of different data sources.

I would say that the data cleaning and enhancement seems to be one of the barriers that a lot of people have had in the music industry, from what I've been learning. We are investing that time and effort, and, you know, we are using AI to help with some of that too. But, you know, it's definitely something where I feel like the scale of the data, because it's so big, like just on our scale alone, you know, we're at the hundreds of millions of streamers just from our catalog, and from these extracts and billions of rows of streams, and we are not big by any means, shape, or form.

Developing data science teams

How do you approach developing and maturing new data science or analytics teams? Our team wants to expand capabilities to more complex data projects, but we don't really have an experienced data science member to mentor us.

Oh, that's really interesting. Well, I would say, first of all, that, you know, we've done a few different things on different teams I've been in before. Sometimes you don't have a more experienced member to mentor you. Sometimes, that if you have a more experienced person, but they might have projects assigned to them and be busy, and you want to grow other people on the team into this discipline, then that can be another reason why. It's not just because you don't have so much experience, but we've done a few different things. So, we've enrolled people in courses before, and then we've done a few programs before.

There's, you know, AI is really good too. Like, ChatGPT has a good understanding of these things, the algorithms. I think it's important always to make sure that you understand how the algorithms work, though, when you're using them, and that you take the time and invest in understanding how it works, and not just making it a black box where, hey, I ran the code and I got an output. Because, you know, it's very easy to make mistakes that way, and to use the right algorithm on the wrong problem, or use the wrong algorithm on the right problem. You know, you need to frame the problem correctly, and you need to understand what is the right algorithm to use to correctly model that problem.

And I would say that, you know, if you don't have the ability to send people to longer term things, like, you know, some sort of, you know, advanced degree program, or some sort of certification, then Coursera is great. Also, you know, I would say leverage ChatGPT and the latest AI, you know, the paid solutions are quite good to give you some insight. And then the most important thing is giving people a concrete project they can work on. Because if you have team members who are naturally curious and interested, you know, the human capacity for learning is great.

And so I think the most important thing is, you know, have an open learning mindset. And then if you give people a problem, and they have this mindset, you give them some space to actually do the research, you know, read papers or read materials online. And make sure that you're just you're checking the assumptions. That's a really important thing. There's sometimes I'll hear things about like, oh, well, there's an article that said to do it this way. So that's the way I did it. And it's great as a starting point, but always make sure that when you have people learning things, especially that you're checking the assumptions.

Interesting music industry data facts

So I think the most interesting things are really, you know, there's a lot of stuff about like, like one hit wonders or like people who, you know, they have their first album is a hit. And, you know, they have multiple hits on it. And it actually tends to be a minority of artists. What's really interesting is that, you know, it's about like one third of artists roughly have a hit in their first year of their career. And, you know, and what's interesting is we now see self releases too, because people can self publish. So that number goes down when you include self releases more than like one eighth.

And then the other thing that we've noticed is that, you know, there's more and more artists looking at like the hot 100 data, you know, talked about, that's one of the things we look at more and more artists having hits on the hot 100 further on into their careers than ever before. And so, you know, you can look at like, Billy Joel had a top 100 hit last year when he released his new album, right. And, you know, he's like 50 years into his career at this point, still having hit songs.

Because, you know, artists are having longer lasting careers, which is part of our goal at Firebird. But it also means that, you know, there's only so much time and so much space to listen to music. And there's platforms have made a lot easier to release music. And then on top of that, you know, it's also easier for artists to have communication and relationships with their fans. Now in this new era, right, you have social media, you have email.

One of the things we never got to see before, which now we can see, which is really interesting. So we get to see through the streaming platforms, how much you listen to music that's not new, it's not frontline. It's called catalog, you know, music that's 18 months or older, and catalog is actually the majority of streaming, which is a really fascinating statistic. It's not something that people outside the industry really know. But you know, if you look at the top 1000 artists, actually, there's a ton of artists who have been deceased for sometimes decades, who are in the top 1000 artists on streaming, because people love the music and keep listening to it.

Catalog is actually the majority of streaming, which is a really fascinating statistic. It's not something that people outside the industry really know.