182: This Data Analyst Has Analyzed 1M+ Songs (here’s everything he knows)

Speaker: 00:00:00

Have you ever wondered how to

do data analytics with music?

2

: 00:00:03

Well, today you'll learn how.

3

: 00:00:04

My guest today is Chris Riva, who is a

music analytics savant and genius, and

4

: 00:00:10

he'll share everything that he knows.

5

: 00:00:12

He's a data analyst for an audio

streamer, and he's literally

6

: 00:00:15

writing the book on music analytics.

7

: 00:00:17

And this episode he'll tell us

how to get into music analytics.

8

: 00:00:20

How to get music data, how to

analyze it, and what he's learned

9

: 00:00:24

with years of analyzing music.

10

: 00:00:26

Let's go ahead and get into it.

11

: 00:00:27

Chris, you have analyzed over a

hundred thousand, if not a million

12

: 00:00:32

different song data over your career.

13

: 00:00:35

Uh, and specifically for your new book,

which is coming out soon, uncharted

14

: 00:00:39

territory, I will link in the show

notes for everyone to check it out.

15

: 00:00:42

Down below, you analyzed, you know,

over a thousand number one hit songs.

16

: 00:00:46

Basically every number

one hit song since: 1958

17

: 00:00:50

Which is crazy.

18

: 00:00:51

That's a lot of songs to listen to and

to look at the data for in your study,

19

: 00:00:55

like what do you feel like you learned

looking at those Number one hit songs.

20

: 00:00:59

I feel like the, this will be a

little bit of a cheesy takeaway,

21

: 00:01:02

but, and it's not even specifically

related to analytics, is that.

22

: 00:01:05

You really should keep an open mind when

you're listening to music because things

23

: 00:01:09

that seem strange to you on the surface.

24

: 00:01:12

Often when you dig a little bit deeper,

there are rich musical communities

25

: 00:01:17

that are making all different styles

of music, and if you don't like.

26

: 00:01:21

Any, so if you claim not to like one song

in a specific genre, you probably just

27

: 00:01:25

haven't listened to enough of it yet.

28

: 00:01:27

So I like to tell people, listening

to all those songs, analyzing all

29

: 00:01:31

this music has taught me to keep an

open mind about how music works and

30

: 00:01:35

what it means to be a good song.

31

: 00:01:36

But also, you know, you want to keep

an open mind when you're analyzing

32

: 00:01:39

music or analyzing data because if

you go in with preconceived notions.

33

: 00:01:44

You're liable to miss out on some

of the most interesting conclusions.

34

: 00:01:48

That's super interesting.

35

: 00:01:49

It's interesting also as well, because

for this study you are, uh, we should

36

: 00:01:53

mention that you were a data analyst.

37

: 00:01:55

You're now a senior, uh, product manager

of data and personalization at Audio Mac.

38

: 00:02:00

So you, you know, data, you know,

data analytics, and you did this

39

: 00:02:03

study, in my opinion, from like

kind of two different perspectives.

40

: 00:02:07

One, you have kind of like more

of a qualitative perspective

41

: 00:02:10

where you're actually like.

42

: 00:02:12

Listening to all 1000 number one

hit songs and you're like, thinking

43

: 00:02:15

about it and you're like, oh, that's

interesting how they did this here.

44

: 00:02:18

Or, you know, that's, I, I didn't

really expect that here, but you're also

45

: 00:02:22

doing it from a quantitative standpoint

where you actually have the data.

46

: 00:02:26

And I think I counted, it was

like, it was like over a hundred

47

: 00:02:28

columns, 105 columns, uh, worth of

data on each one of these songs.

48

: 00:02:33

Um, and you also have

that database online.

49

: 00:02:35

We'll have a show, uh, link

to the show notes down below.

50

: 00:02:37

So you're like doing a qualitative study.

51

: 00:02:40

While doing a quantitative study where

you're like looking for what your ears

52

: 00:02:44

are hearing and kind of like maybe

what your brain's thinking and what

53

: 00:02:47

your heart's feeling, but you're also

like, Hey, what numbers can I actually

54

: 00:02:51

like attribute back to those physical,

you know, sensations I'm having?

55

: 00:02:54

Can you talk about like.

56

: 00:02:55

How you did this study kind of

through those two different methods.

57

: 00:02:58

Ultimately, especially with, there's a

quote about music I, I think I put at the

58

: 00:03:02

beginning of the book, uh, writing about

music is like dancing about architecture,

59

: 00:03:06

which is sort of a funny quote is like,

music is experienced by listening to it.

60

: 00:03:09

You could write as many words about

it as you want, but ultimately, you

61

: 00:03:12

know, music is a very human subjective

thing that we all experience.

62

: 00:03:16

So data can make you dispassionate

about certain things and give you

63

: 00:03:20

a better view about what's actually

going on with certain things.

64

: 00:03:23

But ultimately we're talking about.

65

: 00:03:25

Popular songs here, so we need

to experience them with our ears.

66

: 00:03:29

And typically what would happen is

I would be going along listening

67

: 00:03:32

and I'd be like, oh, I feel like

I'm noticing some strange trend.

68

: 00:03:36

I feel like in the 1970s there's a

lot of songs that have the word dance

69

: 00:03:41

and shake and boogie in the title.

70

: 00:03:44

Now if you know about music of that area,

you'll be like, oh, that's not surprising.

71

: 00:03:47

That's when disco and

dance music became popular.

72

: 00:03:50

But the nice thing about data is I

can have this intuition and I can go

73

: 00:03:53

check, I could just scan all the song

titles, put together a chart and be

74

: 00:03:56

like, oh, there was actually a sharp

rise in a specific, in music using a

75

: 00:04:00

specific type of language in this era.

76

: 00:04:03

And that's what I sort of do throughout

the entire book is as I'm listening

77

: 00:04:06

to these songs, my gut is telling

me something, or there is some.

78

: 00:04:10

Something I've heard about before, and

I'm like, all right, let's, let's go

79

: 00:04:13

check if this is actually the case.

80

: 00:04:14

So it's the subjective and the

objective, or you know, the

81

: 00:04:18

personal and the quantitative.

82

: 00:04:20

The qualitative and the quantitative

are really wrapped together.

83

: 00:04:23

I think when you're talking

about data and art specifically.

84

: 00:04:26

But I think about data

and anything specifically.

85

: 00:04:29

You need to have a feel for something.

86

: 00:04:31

You can't just analyze your way directly

out of it without any feel for what's.

87

: 00:04:36

The data actually represents, I think

some people don't necessarily know that,

88

: 00:04:40

and they think that like everything's

data-driven, and I guess it's kind

89

: 00:04:44

of good if everything is data-driven.

90

: 00:04:46

Um, but from my experience, uh, working

in industry data is usually used as

91

: 00:04:51

a guide and like, as a suggestion

more than like absolute fact.

92

: 00:04:56

So for instance, when I,

when I worked for ExxonMobil.

93

: 00:04:58

Uh, one of the things that I did is

like predict gas demands, uh, every

94

: 00:05:02

gas station, ExxonMobil gas station

in America, or I would run a bunch

95

: 00:05:06

of simulations to try to figure

out what the best crude oil that we

96

: 00:05:10

should buy based on prices right now.

97

: 00:05:11

Is it, you know, is it Russian

or is it from Saudi Arabia?

98

: 00:05:14

So on and so forth.

99

: 00:05:14

And the decisions, the numbers that I had.

100

: 00:05:16

We're never the numbers that actually

happened in real life, even though if

101

: 00:05:20

they were optimal or that's like the best

prediction, it would always go to someone

102

: 00:05:24

that's a little bit closer to the business

than, than I am that like actually

103

: 00:05:28

understands like, I don't know, gas

demand more than me as a data scientist.

104

: 00:05:32

I, I understand I'm a chemical engineer,

so I understood gas and I understood

105

: 00:05:35

some of that, but like these people

have been in the industry for so many

106

: 00:05:38

years and those are the people who are

actually making the decisions and they're

107

: 00:05:41

using my numbers kind of as a guide, but

they're still going with a lot of like.

108

: 00:05:45

Their gained experience

and their, uh, heart.

109

: 00:05:49

So, uh, interesting to kind of hear that

from, from your perspective, uh, as well.

110

: 00:05:53

'cause you like grew up loving music, like

you've loved music for a long time, right?

111

: 00:05:57

Like played music as well.

112

: 00:05:58

Yeah.

113

: 00:05:58

That's my, my first musical love as I.

114

: 00:06:01

Of course, at first I was listening to

music as a kid and was always really

115

: 00:06:05

into bands and learning about, you know,

who played what instrument X, Y, and Z.

116

: 00:06:08

But since I was in middle

school, I've played instruments.

117

: 00:06:11

I've always enjoyed writing

and recording songs and playing

118

: 00:06:14

in a variety of bands, so.

119

: 00:06:16

My interest in music First comes

my interest in writing about music.

120

: 00:06:20

First comes from just being a fan of

making music and listening to music,

121

: 00:06:24

and it was years later that I discovered

that you could apply quantitative

122

: 00:06:29

skills to something like this.

123

: 00:06:31

But I totally agree with you.

124

: 00:06:32

You know, you need a.

125

: 00:06:33

You want to have a feel for what

your data is actually being used for.

126

: 00:06:37

Because ultimately, someone

told me once a, A model is a

127

: 00:06:40

map, it's like a map of reality.

128

: 00:06:42

You know, we're never gonna be

able to use data to perfectly

129

: 00:06:45

model everything in the world.

130

: 00:06:48

Well, maybe one day, I don't know, but

you know, the world's really complicated.

131

: 00:06:51

So it's always good to have a feel

for what's going on too, to trust your

132

: 00:06:55

gut a little bit, because sometimes.

133

: 00:06:57

The data might be telling you

something and you're just like,

134

: 00:06:59

that just cannot be right.

135

: 00:07:00

And lo and behold, often you go

and look a little bit more closely

136

: 00:07:03

and you made a mistake somewhere

or you misunderstood something.

137

: 00:07:06

Okay.

138

: 00:07:07

That's, that's really interesting.

139

: 00:07:08

I wanna, I want to chime in on

like one specific gut feeling that,

140

: 00:07:12

that you may have had, um, that

you kind of found out in the book.

141

: 00:07:15

So one of the things that, um, you did

in the book was you basically identified

142

: 00:07:20

a, a, a period where there was fraud

going on in the billboard top 100.

143

: 00:07:26

Now, um, I'm not familiar with, with the

music industry, uh, especially when this

144

: 00:07:30

occurred, which is like the 1970s, right?

145

: 00:07:32

Like I was not alive then.

146

: 00:07:34

Um, and I'm not a music buff.

147

: 00:07:35

So is it like well known that like

this fraud occurred in the: 1970

148

: 00:07:39

Um, and I guess for, for those who

don't know about this fraud, can you

149

: 00:07:42

kind of explain what happened and then

how you were able to use data to, to

150

: 00:07:45

figure it all out and back it all up.

151

: 00:07:47

Yeah, I mean, historically the, the

music industry was known as a, a place

152

: 00:07:51

of shady characters and, you know, people

involved with organized crime ran labels.

153

: 00:07:56

Uh, in the late fifties.

154

: 00:07:58

All these radio DJs were hauled in

front of Congress for what became known

155

: 00:08:02

as the payola scandal, where basically

they were getting monetary kickbacks

156

: 00:08:06

to play certain songs on the radio.

157

: 00:08:07

Congress was like, you can't do

that if you're gonna be paid.

158

: 00:08:10

You have to announce that you are being

paid to play certain songs on the radio.

159

: 00:08:15

This stuff didn't go away.

160

: 00:08:16

It just morphed into different behaviors.

161

: 00:08:18

So there was always talk about how in

the seventies and the eighties, there

162

: 00:08:23

were still some shady behaviors going

on, especially in the world of radio.

163

: 00:08:27

There's a great book called

Hitman by Frederick Danon where

164

: 00:08:30

he outlined some of this stuff.

165

: 00:08:32

But again, I had, I had, I was aware

of some of this in the back of my head.

166

: 00:08:36

But I was looking at this billboard.

167

: 00:08:38

Number one hit data, and I noticed

thing strange that during the: 1960

168

: 00:08:42

your average number one hit stays at

number one for about three weeks, and

169

: 00:08:47

then in the middle of the seventies,

it just takes a complete nose dive

170

: 00:08:49

and it gets very, very close to one

week, which it can't go lower than one.

171

: 00:08:53

You know, if some.

172

: 00:08:54

It can't be at number one for zero

weeks if it's a number one hit.

173

: 00:08:57

And then by the eighties it

sort of climbs back up to around

174

: 00:09:00

the two or three week mark.

175

: 00:09:01

And I was like, oh, that's

sort of a strange decline.

176

: 00:09:04

And again, that felt odd.

177

: 00:09:05

So I started looking into some other data

and I noticed that when songs would lose.

178

: 00:09:09

Their number one slot on the charts.

179

: 00:09:11

During that same era, they would

fall down, say five or six positions.

180

: 00:09:16

Whereas in all other eras, they would

only fall down one or two positions.

181

: 00:09:19

So say going from number one to

number three, now they're going from

182

: 00:09:22

like number one to number six when

they were losing their top slot.

183

: 00:09:25

So this is weird.

184

: 00:09:26

I mean, what we're seeing anomalous data.

185

: 00:09:28

Then I need a smoking gun

to, to write this story.

186

: 00:09:31

I can't just say something weird

is going on because sometimes there

187

: 00:09:34

is just oddities in your data.

188

: 00:09:36

A lot of things change and I find this

character named Bill Wardlow who was

189

: 00:09:40

involved with the billboard charts

at the time, he was actually the

190

: 00:09:42

director of the billboard charts, and

people over the years have said that.

191

: 00:09:46

If you were in his good graces, you

could just be like, yeah, I need this

192

: 00:09:49

record to be number one this week.

193

: 00:09:51

And for an exchange of, most people

say there was no exchange of money

194

: 00:09:56

for an exchange of something, uh,

unspecified, he would get your record

195

: 00:10:00

onto where it had to be on the charts.

196

: 00:10:02

And there are, during that

era in the seventies, there's

197

: 00:10:05

a lot of chart oddities.

198

: 00:10:06

You have songs, certain songs that

people claim should have gone to

199

: 00:10:09

number one but didn't because it.

200

: 00:10:11

It was not in this guy's best

interest to make it happen.

201

: 00:10:15

But again, find, figuring

this out was a combination of,

202

: 00:10:18

oh, I'm looking at the data.

203

: 00:10:19

I see something weird.

204

: 00:10:21

I'm aware of some sketchy behavior

in this department before.

205

: 00:10:25

And then you're like, all right, let's

see if there's actually any truth to this.

206

: 00:10:28

And it was able to find that there was

super interesting, uh, you're, you're

207

: 00:10:32

solving crimes or maybe not crimes,

but you're, you're solving shadiness.

208

: 00:10:36

Yeah.

209

: 00:10:37

Uh, with data in the music industry.

210

: 00:10:39

Uh, one of the other things that

you found, uh, in the book as well

211

: 00:10:42

was that the percentage of number

one hits, that had a key change has

212

: 00:10:46

like basically gone to, to zero in

the last like 10 years or, or so.

213

: 00:10:52

Tell us what that's all about.

214

: 00:10:53

First off, I guess maybe explain

what a key chain is for or not.

215

: 00:10:56

Key chain.

216

: 00:10:57

A key chain.

217

: 00:10:57

That's what goes where your car keys go.

218

: 00:10:59

Yes.

219

: 00:10:59

Uh, a key change in a song, uh, what that

is for us who maybe aren't as musical.

220

: 00:11:04

Uh, and then explain like why

is that maybe significant.

221

: 00:11:07

Very odd.

222

: 00:11:07

I've written about this before and it

always gets quite the reaction online

223

: 00:11:11

when you say there are fewer key

changes because I'm of the impression

224

: 00:11:15

that this is sort of a niche, uh,

you know, music theory topic, but

225

: 00:11:19

it seems to get people really going.

226

: 00:11:21

So they seem to have an

intuitive grasp on this.

227

: 00:11:23

When you think of a.

228

: 00:11:24

A key, a musical key is like a set of

notes that a song is based around, and

229

: 00:11:29

if you change the key, it means this

different other part of the song is now

230

: 00:11:33

based around a different set of notes.

231

: 00:11:34

That sounds sort of imprecise and

um, complicated, but when you hear

232

: 00:11:39

some examples of it, it's obvious.

233

: 00:11:41

In the seventies and eighties, it's

most notably at the end of a song

234

: 00:11:45

you'll hear, it seems like the song

goes up higher for the last chorus.

235

: 00:11:48

Something like Living

on a Prayer by Bon Jovi.

236

: 00:11:51

Or I wanna dance with somebody by

Whitney Houston, or if you're familiar

237

: 00:11:55

with the song Love On Top, by Beyonce,

which is from the new, new, the two

238

: 00:11:58

thousands, she ratchets the key up

like four or five different times at

239

: 00:12:02

the end of the song from the sixties

to the, I don't know, around: 2000

240

: 00:12:06

It's something like 20% of

songs have a key change.

241

: 00:12:10

Not all of them are using that

specific key change, which is sometimes

242

: 00:12:14

called the gear shift key change.

243

: 00:12:15

'cause it feels like you're

shifting a car into a higher gear.

244

: 00:12:19

And then it plummets down to

zero, and it's still pretty

245

: 00:12:23

close to zero these days.

246

: 00:12:24

Key changes are not that

common in popular songs.

247

: 00:12:28

I think the reason people find this

observation interesting, and again, this

248

: 00:12:31

was an observation at first, that I just

felt like this was happening, and then I

249

: 00:12:35

went and measured it and it was the case.

250

: 00:12:37

Is that people associate key changes with

some sort of sort of musical complexity.

251

: 00:12:42

So sometimes people try to extrapolate

this to say, oh, popular music is becoming

252

: 00:12:46

less complicated, and people get upset

about this to some degree, thinking

253

: 00:12:51

that there's no expertise or there's no

craft, um, in our popular songs, as there

254

: 00:12:55

once were, you know, when the Beatles

were topping the charts or whatever.

255

: 00:12:58

I don't think this is,

is exactly the case.

256

: 00:13:02

When you see this decline in key changes,

it's mostly when hip hop becomes much

257

: 00:13:05

more popular and, and hip hop is a genre

that's much less based around, or I should

258

: 00:13:10

say it's more based around rhythm and

lyricism in general than it is melody

259

: 00:13:15

and harmony In hip hop songs, complexity

is not really built around harmonic

260

: 00:13:20

changes or key changes in the same way.

261

: 00:13:22

That it is in earlier forms

of pop and rock music.

262

: 00:13:27

So that's, that's how I interpret

it as it's really a change in what

263

: 00:13:30

genres are popular, more in how

skilled we are at crafting songs.

264

: 00:13:34

But I know people online have

interpreted this in other ways.

265

: 00:13:38

Super interesting.

266

: 00:13:39

Um, we should mention that you, you're

on TikTok, you make TikTok videos.

267

: 00:13:43

What effect do you feel

like platforms like TikTok.

268

: 00:13:46

Have had on, on the music

industry, um, like has it

269

: 00:13:49

changed how, how artists emerge?

270

: 00:13:52

Has it changed what the

top 100 charts look like?

271

: 00:13:55

A hundred percent.

272

: 00:13:56

I mean, in a certain sense, TikTok

is a continuation of a longer history

273

: 00:14:00

connected first to social media.

274

: 00:14:02

I mean.

275

: 00:14:03

In the mid two thousands, MySpace was huge

for making and breaking musical careers.

276

: 00:14:09

But even if you go back further to like

the eighties, you know, MTV was huge

277

: 00:14:13

in making and breaking musical careers.

278

: 00:14:15

TikTok is sort of part of that

trend, but it's different in terms

279

: 00:14:19

of how things go viral on TikTok.

280

: 00:14:22

Of course, I'm sure most

people listening to this have.

281

: 00:14:25

Use TikTok to some degree.

282

: 00:14:27

Short form video.

283

: 00:14:28

Music is heavily integrated

into the platform.

284

: 00:14:30

Often the way songs go viral

on TikTok is a song becomes

285

: 00:14:34

associated with a particular trend.

286

: 00:14:36

In many senses, that means people

are dancing to a song, but in other

287

: 00:14:40

senses, trends are associated with

songs in many, many different ways.

288

: 00:14:45

And finding a song in the two thousands

that got to the top of the charts and

289

: 00:14:50

was not popular on TikTok is basically

impossible in the same way that.

290

: 00:14:54

You were almost never gonna find

a number one hit in the: 1980

291

: 00:14:58

did not have a popular music video.

292

: 00:15:01

So TikTok is where hits

really pop off these days.

293

: 00:15:06

It's not the only place and it's

fundamentally changed how artists

294

: 00:15:10

interact with their fans and how

people interact with music from.

295

: 00:15:13

One other interesting point about

this is, again, I like this comparison

296

: 00:15:16

to the 1980s because it was, the

music video is something else that

297

: 00:15:20

changed how things became popular

that wasn't completely musical.

298

: 00:15:24

When you went and watched Madonna's video

for like a Virgin, for example, you were

299

: 00:15:29

watching something that A, had Madonna

in it, and B Madonna was clearly involved

300

: 00:15:33

in making on TikTok it's different.

301

: 00:15:35

You could have, you could upload a

song and then suddenly some random

302

: 00:15:39

kid in Ohio dances to it in their

basement, and suddenly that song

303

: 00:15:44

becomes popular and it doesn't

really have anything to do with you.

304

: 00:15:47

And we see, we've seen this a bunch of

times throughout the: 2010

305

: 00:15:53

That artists almost have less control over

their work, in a sense, because anyone

306

: 00:15:57

could upload up mu stuff on the internet.

307

: 00:15:59

Anyone could make posts, anyone could.

308

: 00:16:01

Fans are, it's much more interactive

than it was decades ago, which is

309

: 00:16:05

very distinct from earlier eras.

310

: 00:16:08

Very interesting.

311

: 00:16:09

Yeah.

312

: 00:16:09

Um, TikTok is, has played a role in like

what music I, I listen to, uh, as well.

313

: 00:16:15

I want to get a little

bit into the weeds here.

314

: 00:16:17

I'm actually gonna share my screen for.

315

: 00:16:20

Those of you guys who are watching

on, uh, YouTube, because I

316

: 00:16:23

wanna talk about this data set.

317

: 00:16:25

So this is the data set, uh, that you

kind of used, uh, to write your book

318

: 00:16:29

for the analysis in your book, and

you have it online for anyone to use.

319

: 00:16:32

So I encourage anyone who's interested.

320

: 00:16:34

Uh, in music data to take a look at it.

321

: 00:16:36

Um, because one of the things that

I think is a little bit difficult

322

: 00:16:39

is when you think of music, you

don't necessarily think of numbers.

323

: 00:16:42

So, um, you basically have this, this

list of the song, the artist, the date.

324

: 00:16:47

That all makes sense.

325

: 00:16:48

Uh, but then the other categories that

you have here, or the other columns,

326

: 00:16:52

I guess I, we should say, is ratings.

327

: 00:16:54

Weeks at number one.

328

: 00:16:55

How many weeks in a row?

329

: 00:16:57

It was at number one.

330

: 00:16:58

I don't know this word.

331

: 00:16:59

What's this word?

332

: 00:17:00

Di Divisiveness.

333

: 00:17:02

Divisiveness.

334

: 00:17:02

A scale.

335

: 00:17:03

Uh, so I mean, some of this was

calculated by me, but mm-hmm.

336

: 00:17:06

How this project started was a

friend and I would listen to every

337

: 00:17:10

song and we would just rate them out

of 10, which I've anonymized our,

338

: 00:17:13

our ratings there, but there were

three people who would rate songs.

339

: 00:17:16

The overall rating, I just take

the average of the three and the

340

: 00:17:19

divisiveness was basically me trying

to figure out a way to measure.

341

: 00:17:23

Mm, which songs had the

biggest spread in their rating?

342

: 00:17:27

So if I lay, if I said a song was one

was a one out of 10 and my buddy said

343

: 00:17:32

it was an eight out of 10, I would be

like, oh, this song is divisive because

344

: 00:17:35

we couldn't agree on how good it was.

345

: 00:17:37

That's how this actually started.

346

: 00:17:39

Funny enough, and then it expanded.

347

: 00:17:40

And so that's like very subjective data.

348

: 00:17:43

But the other, most of the other

stuff is objective measures or you

349

: 00:17:46

know, objective or factual pieces

of information that I tacked on.

350

: 00:17:50

Interesting.

351

: 00:17:51

So you have like what label

they're with, their parent label,

352

: 00:17:54

the different genres and styles.

353

: 00:17:56

Um, whether it features an, features

an artist, multiple artists, which

354

: 00:18:00

I think is really interesting.

355

: 00:18:01

Place of origin, age.

356

: 00:18:04

Male whites, you have race and age in

there, the songwriters, which I know

357

: 00:18:07

in your book you do some analysis

on whether, like what affects, uh,

358

: 00:18:12

having one songwriter versus like

four or five songwriters would have,

359

: 00:18:16

um, whether it's male or female.

360

: 00:18:18

So there's lots of really

good data, uh, that you have.

361

: 00:18:20

You, like I said, 105 different.

362

: 00:18:23

Columns, some of it, like what

we talked about earlier, uh, what

363

: 00:18:26

key, it's in like a simplified key,

some of the like energy and these

364

: 00:18:31

different, um, vibes around the song.

365

: 00:18:34

Um, I don't know.

366

: 00:18:34

Yeah, those, those come from spot.

367

: 00:18:36

Spotify.

368

: 00:18:36

Okay.

369

: 00:18:36

Spotify.

370

: 00:18:37

API, right?

371

: 00:18:38

That's, yep.

372

: 00:18:38

Um, yeah, Spotify keeps track of those.

373

: 00:18:40

So they're looking at some of the

actual like data from the like.

374

: 00:18:44

Sound wave of the song, basically.

375

: 00:18:46

Um, whether there's bongos or the

banjo in there, that's really cool.

376

: 00:18:50

So you could, you could like do some

pretty cool analysis, which you obviously

377

: 00:18:54

have in the book, but anyone, anyone

listening, I think there's almost like

378

: 00:18:57

unlimited analysis that you could do.

379

: 00:18:59

Like, hey, how many artists, how many

black female artists had a number one

380

: 00:19:04

hit with the flute slash the piccolo?

381

: 00:19:06

That's an interesting question

that you could answer.

382

: 00:19:08

You could answer that.

383

: 00:19:10

Uh.

384

: 00:19:11

I, and that was sort of my goal with,

I, I've always appreciated about the

385

: 00:19:14

data community that I've at least

interacted with online, is that

386

: 00:19:17

everyone's pretty open to sharing data

sources or making things open source.

387

: 00:19:22

So I knew when I, I had this huge

data set that I wanted to make it

388

: 00:19:27

available and hopefully someone else

could use it so that I wasn't, you

389

: 00:19:30

know, I didn't just waste years and

years of building this just for myself.

390

: 00:19:34

And I, you'll see in one of

the other tabs, I have a data

391

: 00:19:36

dictionary, so I try to describe.

392

: 00:19:39

What all the columns are.

393

: 00:19:41

Um, but yeah, I'm, I'm

hoping PE people use it.

394

: 00:19:44

Uh, that's sort of the goal there.

395

: 00:19:45

Yeah, I could see some really

interesting things like talent

396

: 00:19:48

show contestant, like there was a

time where American Idol mattered.

397

: 00:19:52

Um, and now at at least the latest, I

think like Pop Star, I don't know if

398

: 00:19:57

they fit your criteria of Pop Star that

was American Idol, like dropped out

399

: 00:20:00

extremely, uh, earlier Benson Boone.

400

: 00:20:03

I think the other thing that's really

interesting that isn't even necessarily

401

: 00:20:07

tabular that people could, if they

really wanted to is you have the lyrics.

402

: 00:20:11

And the lyrics is like

a whole nother data set.

403

: 00:20:13

'cause it's super unstructured.

404

: 00:20:15

Um, so you could do some really

cool NLP stuff, like where you're

405

: 00:20:18

actually analyzing the words in

each song at a really high level.

406

: 00:20:21

So this is an awesome data set.

407

: 00:20:23

Um, and thank you for, for making

it, uh, open source for everyone.

408

: 00:20:27

I'm curious because once again, you

do work, uh, for Audio Mac, which

409

: 00:20:30

is a music streaming platform.

410

: 00:20:32

You know, you're writing this book

about the history of music and, and

411

: 00:20:35

using data to kind of analyze it.

412

: 00:20:37

Uh.

413

: 00:20:38

Do you ever get sick of like your

job and your hobby being data

414

: 00:20:42

plus music or is it something that

you could do like all the time?

415

: 00:20:45

I haven't burnt out yet.

416

: 00:20:46

I, I know I've had a couple friends that

are worried that I will, I will flame

417

: 00:20:50

out with this because it's, I've been

so invested in it, um, for so long.

418

: 00:20:55

But I mean, still, I,

I turn 30 or earlier.

419

: 00:20:57

A couple months ago, and

I, I still still love it.

420

: 00:21:00

I still love all the musical stuff so far.

421

: 00:21:02

I try to, you know, I try to do some

other things occasionally to give myself

422

: 00:21:06

a little bit of a musical reprieve.

423

: 00:21:08

Uh, but right now I'm

still having fun with it.

424

: 00:21:10

So I, but I think it's important,

you know, you, it's, I think it's

425

: 00:21:13

easier to a degree if you're deep in

the weeds of a data set, if you're at

426

: 00:21:18

least interested in it to some degree.

427

: 00:21:20

We were talking before,

uh, we, we hit record that.

428

: 00:21:23

The process of writing a book is just

extremely difficult, um, which I have

429

: 00:21:27

heard, I have not done, but I've heard.

430

: 00:21:29

And so, uh, I think it makes sense that

you wanna write a book or do anything

431

: 00:21:33

hard about something you're passionate

about because when those hard moments

432

: 00:21:36

do come, you are like, at least it's

kind of fun to, to do all of this.

433

: 00:21:40

Um, so I'm glad, I'm glad

you're still enjoying it.

434

: 00:21:43

You know, we talked about

the data set, just barely.

435

: 00:21:44

I'm curious, like we mentioned

the Spotify, API or or Spotify

436

: 00:21:48

has some data available.

437

: 00:21:49

I'm curious, like if you need to go

find like a music data set, what are

438

: 00:21:53

some of the resources or some of the

methods you try to go find that data?

439

: 00:21:56

Yeah, I, I mean there are, I know certain

people who write stuff with data is,

440

: 00:22:00

they'll basically start with a data set.

441

: 00:22:02

And be like, oh, I found

this great data resource.

442

: 00:22:05

I'm gonna write something about it.

443

: 00:22:06

And I've done that before.

444

: 00:22:07

What I typically do is I usually have

a question and then I'm like, all

445

: 00:22:10

right, I have to go find data for this.

446

: 00:22:13

Which has its, I mean, there's

been some stuff I can't write about

447

: 00:22:15

'cause I just don't, I'm not able

to locate data, even if I think the

448

: 00:22:18

question is interesting, some of them.

449

: 00:22:20

Valuable musical resources that are out

there that are generally easy to use.

450

: 00:22:26

The Spotify API is great.

451

: 00:22:28

Uh, they've locked some

of it down recently.

452

: 00:22:30

I think they're trying to prevent

other people from using it to

453

: 00:22:32

train LLMs, but, uh, you can still

access a lot of great data there.

454

: 00:22:37

Um, Wikipedia is a great data source.

455

: 00:22:40

Uh, Wikipedia has an API that it's

a little bit janky, but there's

456

: 00:22:44

a lot of great stuff on there.

457

: 00:22:46

And you can access basically

any Wikipedia page and Wikipedia

458

: 00:22:49

has lots of great lists.

459

: 00:22:50

So a have like list of every rock artist

or list of every: 1980

460

: 00:22:57

can pull all those down and access all the

pages and see certain things about those.

461

: 00:23:02

For example, something I did with that

was I try to do something about nepotism,

462

: 00:23:07

because if your parent is popular, usually

they'll be listed on your Wikipedia page.

463

: 00:23:11

Linking back to their own page.

464

: 00:23:12

So I was like, all right, let's look

at every pop star to find however you

465

: 00:23:16

want, and see how many of them have

parents who are also famous or famous

466

: 00:23:20

enough to have a Wikipedia page.

467

: 00:23:22

Uh, so there's some fun stuff

you could do with Wikipedia.

468

: 00:23:25

Music Brains is another big one.

469

: 00:23:27

This is a huge, huge open source

music project that has more data than

470

: 00:23:33

you could ever possibly dream of.

471

: 00:23:34

Billboard.

472

: 00:23:36

Is not open source, but people scrape

like the billboard charts and you

473

: 00:23:40

can, if you just search billboard

chart data, you could find the entire

474

: 00:23:44

history of the hot 100, uh, Kaggle,

they have a lot of data sets on there.

475

: 00:23:48

Occasionally there are some music

data sets that are useful, and

476

: 00:23:51

then there are some, all that

stuff I'm talking about is free.

477

: 00:23:54

There are some paid resources.

478

: 00:23:56

You could use Chart Metric as a big

one for the music industry, which is

479

: 00:23:59

pretty cheap, relatively speaking.

480

: 00:24:02

They scrape and they have data on base for

every song from basically every platform.

481

: 00:24:07

You know, you could see how many

radio stations in Kenya have

482

: 00:24:10

added a certain song or something.

483

: 00:24:12

It's pretty overwhelming.

484

: 00:24:13

Great resource.

485

: 00:24:14

You'd have to pay for it.

486

: 00:24:15

And then Lumin.

487

: 00:24:16

Which is much more expensive to the

point where if you're a single person,

488

: 00:24:18

you're probably not paying for it.

489

: 00:24:20

Luminate is the company that

powers the billboard charts.

490

: 00:24:23

So there is, there's lots

of music data out there.

491

: 00:24:27

And don't be a, if you're looking for

something and you think someone might

492

: 00:24:30

have it, don't be afraid to email 'em.

493

: 00:24:32

I've.

494

: 00:24:32

S come upon a couple data sets in

my life just because I sent a cold

495

: 00:24:37

email to somebody and they were like,

oh, yeah, I, I know how to get that.

496

: 00:24:40

I'll, I'll let you know.

497

: 00:24:41

Gimme a call.

498

: 00:24:42

So you gotta poke around.

499

: 00:24:43

But it's out there, the

power of networking.

500

: 00:24:45

That's super cool to to hear you.

501

: 00:24:48

Obviously, you know, you

work with Data at Work.

502

: 00:24:50

Uh, for this book.

503

: 00:24:51

You crunched a bunch of

numbers for your newsletter.

504

: 00:24:53

You crunched a bunch of

numbers, uh, for the book.

505

: 00:24:55

You created a, a decent amount

of charts and, and graphs.

506

: 00:24:57

I'm curious like.

507

: 00:24:58

What are your go-to data tools if you're

going to be analyzing data and what did

508

: 00:25:02

you use to make the charts in the book?

509

: 00:25:04

Uh, if I'm at work, I'm mostly in the

sql, Python, pandas, land and Ex Excel.

510

: 00:25:11

You, I don't know, you

can't really avoid Excel.

511

: 00:25:13

It does a lot of great stuff.

512

: 00:25:14

So, um, still rely on Excel, but.

513

: 00:25:16

During the day, I'm, I'm basically write

during my, my workday, I am writing a

514

: 00:25:20

lot of SQL queries, uh, and we use a

visual, an open source visualization

515

: 00:25:24

tool called Superset, which I think

was developed by people at Airbnb.

516

: 00:25:29

And they opened, they open, sourced

it for my personal projects,

517

: 00:25:33

the book and the newsletter.

518

: 00:25:34

I'm mostly using Pandas and

Excel for my newsletter.

519

: 00:25:39

I do all the visualization through

Data wrapper, which has great

520

: 00:25:43

free data visualization tools.

521

: 00:25:45

Occasionally I'll use Canva for stuff,

but not so much the book, though.

522

: 00:25:49

I worked with a graphic designer,

her name is Kaylee Nerney.

523

: 00:25:52

She's one of my good friends.

524

: 00:25:54

She, I basically gave her all the data.

525

: 00:25:57

We talked through it, and she

built all of those charts using.

526

: 00:26:01

I believe like Adobe, like Adobe

Illustrator, the ones in the

527

: 00:26:05

book are much more custom built.

528

: 00:26:08

And I think if you wanna do stuff

like that, you probably have to

529

: 00:26:11

work with someone who is not as

graphically challenged as I am.

530

: 00:26:15

But for simple stuff in the

newsletter, you know, I'm very

531

: 00:26:17

data wrappers a great tool.

532

: 00:26:19

Even the.

533

: 00:26:20

The visualization tools in

Excel are, are good enough.

534

: 00:26:23

If you're just sending out a newsletter,

people are posting something online.

535

: 00:26:27

Very cool.

536

: 00:26:28

I honestly have never heard of data

Wrapper before, but I really do like,

537

: 00:26:32

uh, the charts on your newsletter,

so I'll have to check that out.

538

: 00:26:36

And the other one you

mentioned was Superset, right?

539

: 00:26:37

Is that right?

540

: 00:26:38

Yeah, superset.

541

: 00:26:39

We use that at work.

542

: 00:26:41

So it plugs into like our data,

our data warehouse, and we could.

543

: 00:26:46

Set up visualizations and their charts

and graphs for people who aren't as

544

: 00:26:50

data savvy, um, in our organization,

though I think your development

545

: 00:26:54

skills might have to be a little more

sophisticated to get that all set up.

546

: 00:26:57

I, I didn't, I didn't do the setup

there, but it was a tool my, my

547

: 00:27:00

coworker was aware of and we've

had good success with it so far.

548

: 00:27:05

Very cool.

549

: 00:27:05

Uh, what advice would you give

someone that maybe is listening to

550

: 00:27:08

this and really enjoys music and

wants to, you know, become a data

551

: 00:27:12

analyst in the music industry?

552

: 00:27:14

Yeah.

553

: 00:27:14

I mean the, the cool thing about music and

entertainment generally in this day and

554

: 00:27:18

in the internet age is that every one of

these companies is hiring data analysts.

555

: 00:27:23

Whether you're working in live, live,

entertainment, music, streaming,

556

: 00:27:27

publishing, even the, the big

labels, I mean, they're hiring

557

: 00:27:30

people to crunch numbers because

there's so much data around there,

558

: 00:27:34

out there around music these days.

559

: 00:27:36

So it's a good industry to work in.

560

: 00:27:37

I think the most, I always say

the most employable skill that

561

: 00:27:40

I've learned in my day to days is.

562

: 00:27:42

How to write SQL queries, because

every role that I've come across

563

: 00:27:46

always involves SQL in some way.

564

: 00:27:48

I mean, there's tons of other,

uh, statistical, you know, there's

565

: 00:27:52

other tons of other data programs

that you can use, but a lot of

566

: 00:27:55

things seem to fall back onto sql.

567

: 00:27:57

But if you're interested

in music and music data.

568

: 00:28:00

Most of the things that I've been

able to do so far in my career have

569

: 00:28:03

just been because I am constantly

shouting into the void on the internet.

570

: 00:28:08

And occasionally someone is

listening and reaches out to me,

571

: 00:28:12

or I send a cold email to someone.

572

: 00:28:14

And that's how most opportunities

that I have had, especially with

573

: 00:28:19

writing this, this book would've

never happened if I didn't start

574

: 00:28:21

a newsletter or post on TikTok.

575

: 00:28:23

I don't think, you know, people

always say good enough is what is it?

576

: 00:28:26

Uh.

577

: 00:28:27

I'm gonna get it wrong.

578

: 00:28:28

Even if something's not perfect, you

should not be scared to put it out online.

579

: 00:28:32

Like you don't have to have

the perfect data visualization.

580

: 00:28:34

When I started the newsletter, it was

all just, it was just Excel graphics.

581

: 00:28:38

Um, people will, if, if what you're

saying is compelling, people will follow.

582

: 00:28:42

But it's nice to have pretty visuals too.

583

: 00:28:44

So pretty visuals help to go

make things go viral sometimes.

584

: 00:28:48

But good ideas and good

concepts can go a long way.

585

: 00:28:51

And I love what you said that like most of

the time, like 90% of the time I was just.

586

: 00:28:55

Talking into the void and no

one really cared what, what you

587

: 00:28:58

were saying until they don't.

588

: 00:28:59

And then all of a sudden

that makes the difference.

589

: 00:29:02

And it can lead to career opportunities,

it can lead to, you know, book

590

: 00:29:05

opportunities in, in your case.

591

: 00:29:07

And I agree, there's, there's so much

good that can come from posting on

592

: 00:29:11

social media and just talking about

what you're working on, talking

593

: 00:29:13

about what you are interested in.

594

: 00:29:16

That's awesome.

595

: 00:29:17

Okay, Chris, super excited

for your book to come out.

596

: 00:29:20

It's got, uh, uncharted territory.

597

: 00:29:22

Uh, when does it come out?

598

: 00:29:23

It comes out November 13th,

: 2025

599

: 00:29:29

Um, you can find it basically

anywhere it's available.

600

: 00:29:32

It'll be available online through every

major bookseller, Amazon, Barnes and

601

: 00:29:37

Noble, Walmart, all that good stuff.

602

: 00:29:38

I think it's cheaper, cheapest

through the publisher.

603

: 00:29:40

So that's what I've been

linking to for people.

604

: 00:29:42

But you should be able to get it anywhere.

605

: 00:29:43

And if you can't, if you reach out

to me, I will make sure I get you

606

: 00:29:46

a copy of that book in your hands.

607

: 00:29:48

Sweet.

608

: 00:29:49

That's awesome.

609

: 00:29:49

So we'll have a link to it, uh,

in the show notes down below.

610

: 00:29:53

Uh, if you're listening to

this beforehand, then you can,

611

: 00:29:56

uh, pre-order it or if you're

listening to it after it comes out.

612

: 00:29:58

You'll be able to, uh, check it out.

613

: 00:30:00

I haven't read every single page,

but I've read a good chunk of it.

614

: 00:30:03

Um, and I found it pretty interesting.

615

: 00:30:05

Um, so Chris, thanks so much

for coming on the podcast and

616

: 00:30:08

talking about music and data.

617

: 00:30:09

Yeah, thanks for having me.

618

: 00:30:10

I'm always down to, to chop

it up about music and data.