Predicting Popularity: The Math Behind Hit Music
IRA FLATOW, host:
This is TALK OF THE NATION: SCIENCE FRIDAY. I'm Ira Flatow.
(Soundbite of song "Nowhere Man")
THE BEATLES: He's a real nowhere man sitting in his nowhere land making all his nowhere plans for nobody.
FLATOW: What does this song have in common with this one?
(Soundbite of song "Brown Sugar")
FLATOW: And this one?
(Soundbite of song "You're Beautiful")
Mr. JAMES BLUNT (Singer): (Singing) You're beautiful. You're beautiful, it's true. I saw your face…
FLATOW: Yes, they were all big hits, but why? My next guest says that he has figured it out: what makes one song a hit, one song a dud. In fact, he says he's picked two songs now rising on the charts to be hits way before they were. He says all hit songs from punk to rock to country fall into one of 60 patterns that he's identified. He says his company can analyze a song and tell the record company execs if they have a hit or a miss on their hands.
Joining me now to talk more about it is my guest, Mike McCready. He is the CEO of Platinum Blue. He joins me here in our studio. Thanks for being with us today.
Mr. MIKE McCREADY (CEO, Platinum Blue, New York): Hi, Ira. Thanks for having me.
FLATOW: Mike McCready, tell us about how you're able to do this. Is this a computer program, or is it an analysis on your - you sit there and listen to a thousand songs, you play, that's the one?
Mr. McCREADY: No, actually, if we look at historically how the music industry has operated over the past 50 years since its inception, the labels have two criteria they apply to deciding whether a song is going to be released as a single. The first criteria is: does the song sound and feel like a hit? They have A and R professionals - artist and repertoire professionals - people with golden ears who - at the labels who listen to music and decide that.
The second criteria they use is, are we able to promote the song effectively? That in - that covers does the artist have an appropriate appeal to the - audience or the target demographic that they're looking for? Is this - you know, the song have somewhere to fit in? Is it written within the zeitgeist of the culture?
They do also have some market research that they do, call-out research, where the radio stations sometimes call people up and play the hook of the song and try to get a feel for the audience. But using those two criteria and the best market research applications today, they're still only right about one out of every 10 times. And it can cost upwards of a million dollars to promote a single in the U.S. market, so when you're wrong nine out of 10 times, that one success is paying for all of those failures, and whatever's left over is for the…
Mr. McCREADY: …you know.
Mr. McCREADY: So we have a - we've added this third criteria. We have kind of accidentally - and this research was done both at the company I run now called Platinum Blue as well as at our former company, Polyphonic HMI, now called Music Intelligent Solutions. We've identified about 60 different mathematical patterns to which hit songs have conformed historically over time.
So we go to the music labels and we say, look, if your songs - the songs that you're about to promote - continue to conform to these first two criteria, they sound like hits and you're going to be able to promote them effectively - in addition, if they also conform to one of these optimal mathematical patterns, your chances of success increase to about 80 to 85 percent.
FLATOW: So do you, then, analyze the music to see if it conforms to the - one of those 60 patterns?
Mr. McCREADY: Right. We look - we have a computer program that can analyze a fully produced CD and isolate things like melody, harmony, beat, tempo, rhythm, octave, pitch, chord progression, fullness of sound, cadence, sonic variances - about 30 to 35 of these variables that we look at, and we look at how they fit together in the different kinds of patterns that they make up as they come together.
FLATOW: Is this an historical analysis that you've taken from hit songs and distilled what makes a hit?
Mr. McCREADY: It is. We've invented this observational tool. We can't - you know, the computer - again, it isn't telling you if the song sounds like a hit. That's still a human evaluation. Sometimes, we find these optimal mathematical patterns in songs that sound nothing like hits. They would never be successful.
FLATOW: Right. Right. And how good are you? What's your success rate?
Mr. McCREADY: Well, our public success rate, the work that we do, is very close to 100 percent. We have a lab accuracy rate of about 80 to 85 percent, but fortunately, most of the predictions and help that we've provided to our customers - which include Sony BMG, Warner, EMI et cetera - have always been on the money.
FLATOW: We're going to listen to a couple of songs now that were predicted - I think by you - to be hits…
Mr. McCREADY: Mm-hmm.
FLATOW: …before they were hits, and let's listen to that first one.
(Soundbite of song "Crazy")
Mr. CEE-LO GREEN (Vocalist, Gnarls Barkley): (Singing) Does that make me crazy? Does that make me crazy? Does that make me crazy?
FLATOW: And that is?
Mr. McCREADY: Yeah, that's "Crazy" by Gnarls Barkley.
FLATOW: Right. And you predicted that song would be a hit before it was a hit.
Mr. McCREADY: Yeah, that's right. We were actually hired by a radio group in the U.K. to look at this song before they started playing it, and we were able to give them the green light on that.
FLATOW: And how high has that risen on the charts?
Mr. McCREADY: It was last summer's biggest hit in both the U.K. and the U.S.
FLATOW: Your stock, meaning both figuratively and literally…
(Soundbite of laughter)
FLATOW: …must have been on the rise after that one. You could pick that, or somebody said, no, you just got lucky?
Mr. McCREADY: Well, you know, the music industry, in a race to adopt new technology, finishes just ahead of the Amish.
(Soundbite of laughter)
Mr. McCREADY: So you have a lot of naysayers in the music industry still, and - but then there are a lot of forward-thinking music label executives that use our technology. But when you find somebody who's just dead set against it or has the impression that we are trying to bring technology into an art form -which again isn't really the case. We have an observational tool; we didn't invent the fact that hit songs conform over and over again to one of these 60 mathematical patterns.
Mr. McCREADY: We've just been able to observe.
FLATOW: Right. Right.
Mr. McCREADY: But you know, you tell a music label, this song isn't going to work, and they say, well, we don't believe you. We're going to release it and promote it anyway. And then it doesn't work, and you would think we could go back and say, well, you know - in a nice way - we've told you so.
FLATOW: Right. Right.
Mr. McCREADY: But they say, well, you know, we're wrong nine out of 10 times anyway, so that was just maybe statistically another time we were going to be wrong. If you do the opposite and say, well, this song is going to work, and they release it and promote it and it works, and you go back and you say, we were on the money, they say, well, you know, we thought it was going to be a hit, too, that's why we released it and promoted it.
So it's hard to - you know, unless we're working on every release of every - of a particular label where we've been able to increase their percentages - and we do have cases like that - it can be a tough sell at the beginning.
FLATOW: Can you work in the opposite direction? Can you take those 60 variables that you know are going to be - make a hit and write a song that has those sounds(ph)?
Mr. McCREADY: Well, again, I mean, in a sense, but again it's just an observational tool, so what we can do if we get involved in the studio - say, you know, at the final mix stage - we can analyze the final mix that the producer gives us. We can isolate the variables - the melody, harmony, beat, et cetera - and we can say, look, we can't give you any creative direction. We don't know how you should change the song, but we can tell you that the problem appears to be in the base or the baseline of that song.
Mr. McCREADY: So before the producer brings the bass player back in, he'll usually play around with the mix. Maybe he'll muddy up the base a little bit or increase the base a decibel or decrease a - sometimes the producer is the only person in the world who knows what the heck he did. But we'll come back to him and say, look, whatever you did on mix number three - because they'll send us four or five new mixes - we'll say, whatever you did on mix number three seems to be working. Just a little bit more of that and it puts you in this pattern.
If the producer isn't comfortable with how that sounds, mix number three, obviously, they're not going to release it. It has to be a combination of the producer's ears, the A and R's ears and our technology. We actually call this technology music X-ray, because we compare it to the X-ray or an MRI machine when it was first introduced into medicine.
Mr. McCREADY: It shows the doctor a patient in a way that the doctor couldn't see it before and allows the doctor to make better decisions. Our technology shows music industry executives their music and their market in a way that they couldn't see it.
FLATOW: But is it a universal market? I mean, would a different culture have its own cluster of what works versus the American culture…
Mr. McCREADY: Yes. We work…
FLATOW: …Japanese, whatever, anything else?
Mr. McCREADY: Yeah. We work in the U.S., the U.K., Germany, Spain, and we're starting in Brazil and Japan now. What we see is that many of the clusters are present in all of those markets, and that would account for reasons that you have these international hits.
Mr. McCREADY: But you also have clusters that are very particular to individual cultures. So in Spain you have, two or three or four what we call clusters -these are mathematical patterns to which songs conform - that are present only in that market.
FLATOW: Mm-hmm. 1-800-989-8255 is our number. Let's go to Ben(ph) in Ithaca. Hi, Ben, welcome to SCIENCE FRIDAY.
BEN (Caller): Hi. How are you doing?
FLATOW: Hi there.
BEN: My - I have a question actually for you. In terms of forward thinking and the analyzation of music - you know, electronic melodic identification, et cetera, et cetera - do you feel that, you know, your information is compiled solely based on historical context? And along with that, do you think that that impedes the fuller progression of music and evolution of musical talent?
Mr. McCREADY: That's really a good question, Ben. Actually, we've gone back and we've looked at music going all the way back to the classical composers, and we haven't seen new patterns emerge, at least patterns that would - we would call hit songs. If we look at the genres of music that we developed over the 20th century - whether it's country, rock, disco, grunge, punk - we see that the hit songs from those genres continue to come from the same mathematical patterns as some of the best compositions of the classical composers and some of the best music coming out today.
In theory, we would say that the human ear probably hasn't evolved a lot since that time, and in any new genre that could be invented tomorrow, the hit songs from that genre would continue to conform over and over again to these clusters.
FLATOW: Mm-hmm. 1-800-989-8255. Mariel(ph) in Tucson. Hi. Welcome to SCIENCE FRIDAY.
MARIEL (Caller): Hi there. I was just saying, you know - and this is a good segue way, I think - that I was studying music kind of anecdotally with some friends of mine who've been music majors. And I can't remember his name to save my life, but in classical music theory, there was a man - he's a big name in the classes - who had dissected Bach's fugues and created a similar set of kind of algorithms, if you will, for these fugues.
And every fugue that you compose with it sounds technically good. But there is something missing, a life that you get in good classical music that's just not there. And then when you actually take a look at Bach's fugues, every one of his fugues, brilliant works, breaks the rules. So I have to confess a bit of skepticism on this and a fear that this could, in a small way, kind of prevent good music from being there, because it's so called breaks the rules when so did Back every single time.
FLATOW: It might homogenize the music. It would just…
Mr. McCREADY: Well, that's, kind of a two-pronged answer. On one hand, we are looking at the past, at the compositions and at the pieces of music that we have collectively said are good or collectively said there's some quality there. So we are actually looking at and taking into consideration all of the rule breaking that's going on and factoring that into part of what we like.
On the other hand, this is a tool, and just like any tool, you can use it for good or bad. There are people who use this and pay us to help them make homogeneous, formulaic, cookie-cutter sounding pop songs. And there are also producers that are using this technology to push the envelope of creativity and to take sonic risks that they may not normally take but are able to take now because we are able to show them that there is an audience for this and where that audience is going to be.
FLATOW: We're talking with Mike McCready of Platinum Blue on TALK OF THE NATION: SCIENCE FRIDAY from NPR News. 1-800-989-8255. Let's see if we - a lot of people want to talk to you. Let's see if we can get Randy(ph). Let's go to Brad(ph) from Alexandria, Virginia. Hi, Brad.
BRAD (Caller): Good afternoon.
FLATOW: Hi there.
BRAD: When I first heard this subject come on the air on SCIENCE FRIDAY, my first reaction was, how would this have worked with such things as "You Ain't Nothin' But a Hound Dog," "See You Later Alligator," those kinds of songs from the '50s? But that's not my question; I was just curious about that. My question has to do with "Take Five", which was supposed to be - most kids don't even know what this is anymore - but…
FLATOW: Dave Brubeck's piece.
BRAD: …side of the '45 and it was not supposed to be a hit. I forgot the other side now. It must have been "Blue Rondo a la Turk" or something like that. But "Take Five" was not only a big hit, it also became a mixed genre in the jazz at the time, and I'm wondering how your tool or your formula would have worked with that.
FLATOW: Because Dave Brubeck broke all the rules.
FLATOW: In that he takes key, signatures, time, everything else like the - yeah.
Mr. McCREADY: You know, we don't work a lot in the jazz genre and if - I am familiar with that piece and I'm sure that it is in our database, but I don't have the data on that song in front of me. But it would be curious to look at. When you think about, you know, people breaking the rules and, you know, where music could go, of course, there's always room for creativity, and I'm sure that we're going to be stumbling upon things that we haven't seen as well.
But one of the interesting things - you know, I say that we're able to increase accuracy rates up to 80, 85 percent. That still leaves 15 percent margin where we're wrong. And a lot of the times when we're wrong, we can attribute that to some sort of cultural phenomenon - a song like "Cleaning Out My Closet" by Eminem, which was wildly popular but didn't conform to these optimal mathematical patterns that I'm talking about.
FLATOW: "Itsy Bitsy Teenie Weenie," you know.
Mr. McCREADY: Yeah. You know, and the "Hound Dog."
BRAD: Sure you got (unintelligible).
Mr. McCREADY: A lot of these songs are actually, you know, did conform to these patterns, but sometimes you can see that cultural phenomena like the song is a hit or a - the theme song to a blockbuster movie, or it's someone releasing a song at the pinnacle of their career, where it's just the momentum of who the artist is or the social phenomena going around them drives the popular.
FLATOW: Any Beatles tune would…
Mr. McCREADY: Could be. I mean, but there are a lot of those Beatles tunes that actually did both.
Mr. McCREADY: I mean, you know, they were conforming to these optimal mathematical patterns at an opportune time in their career.
FLATOW: So they could be conforming to the patterns and you don't even realize it in your own mind that you think, gee, this sounds so different but it fits in the cluster.
Mr. McCREADY: Yeah. Yeah. I mean, even some of the songs when you played at the beginning of the segment…
FLATOW: Yeah. That's why we played them. They were so different.
Mr. McCREADY: They're different-sounding.
FLATOW: But they conform to your predictions.
Mr. McCREADY: They have - well, not only do they conform to predictions, those three songs that you played - "Brown Sugar"…
Mr. McCREADY: You played "You're Beautiful" by James Blunt. And I don't remember the other one you played.
FLATOW: "Nowhere Man."
Mr. McCREADY: "Nowhere Man." Those three songs are mathematically related to each other. They're in the same cluster. They conform to the same pattern.
FLATOW: Mm-hmm. So we got about a minute left. Where do you go from here? Can you refine this, or you - 85 percent is pretty good, I would say.
Mr. McCREADY: Well, you know, where we're going is, you know - for the last few years we've been offering these services to the music labels, and now we've decided to offer them to unsigned artists everywhere and offer kind of a basic version of the report at musicxray.com for, like, $10. But what we also see is that recommending a song to a mass market for what can be a hit is just a little bit different from recommending a song to an individual based on their own personal tastes, which may not have anything to do with hits.
Mr. McCREADY: So we're developing and have developed a music recommendation system in partnership with other technology companies where someone, you know, can go to a retailer and say, you know, I like these two songs; what else am I going to like?
FLATOW: And this will surely be up on the Internet, some place having to do with iTunes or whatever…
Mr. McCREADY: Oh, yeah, sure. And you can always find out about us at musicxray.com or platinumblueinc.com, keep an eye on what we're doing.
FLATOW: Talking with Mike McCready who is CEO of Platinum Blue. And he was here with us in our studios. Thank you for taking time…
Mr. McCREADY: Thank you very much, Ira. Thank you for letting…
(Soundbite of song "Take Five")
FLATOW: I'm Ira Flatow in New York.
NPR transcripts are created on a rush deadline by a contractor for NPR, and accuracy and availability may vary. This text may not be in its final form and may be updated or revised in the future. Please be aware that the authoritative record of NPR’s programming is the audio.