Prithwijit Mukerji: Moneyballing Music Event Presentation (Transcript: Part 1)

07 Apr 2015

Prithwijit Murkerji (Jeet): ‘Moneyballing’ is a term taken from author Michael Lewis, and was originally used in the context of baseball.  It concerned the Oakland baseball team manager’s use of evidence-based objective data analysis to choose their players, rather than using subjective wisdom to turn around players’ performance, (and on a very low budget).

Now common amongst baseball teams throughout the US, I thought that if it works there, why not apply it to music?

When it comes to talking about statistics in music, what we’re really talking about is the global online consumer activity – the conversation that’s happening around music.  That’s essentially social media activity – from Facebook ‘likes’ to Twitter followers, to Shazam tags…all the way through to pirated content.

This data really is big, it’s unwieldy, very complex and moves very quickly.  In terms of analysing it, we’re really trying to use predictive analysis to extract the value from this sea of information.

Why care?  Music consumption and the exchange of information about it has changed.  It’s weightless, automatic; it’s privatized, mobile and has become more and more invisible. You can no longer just use sales as a completely accurate indicator for a track’s performance.

Music is inherently social, it’s socially embedded and so are the platforms it inhabits.  We can use those platforms that make that consumption exchange invisible, to draw off this big data and learn about our audiences, our artists and make consumption less invisible.

However, it’s not cheap.  Searching for it, capturing it and sharing it is costly, and it’s not very accessible.  Those who can afford to do it and have the resources available are essentially major labels and services.  Spotify makes use of collaborative filtering, similar to the Amazon and Netflix platform which flags similar content to that which it knows you consume.  It’s not perfect, and subtleties within genres – e.g. between, say, east and west coast rap [almost certainly] is not necessarily recognised, but it enhances the service.

Spotify recently bought The Echo Nest, a data company, who last year collected 1 trillion points of data [2014].  Through the use of algorithms, their data on music can be used to identify [consumer habits/ trends] indicating what films music consumers like and even who you might vote for [in an election].

Whilst it’s a little bit scary, it’s also beautiful as music is no longer a ‘thing’ that somebody does in a closed room.  It’s everywhere and linked to other content.  So it becomes more relevant, interesting and important to understand how music consumption and exchange is linked to other content, e.g. the films you watch.

The Next Big Sound, another data company, which powers Spotify For Artists, collected 435 billion data, points last year.  They look at 200 artist metrics and can even use this to predict likely sales of a single for an artist who performed on a TV show.

What this tells us is that as streaming is becoming more and more important – it’s now recognised by the Official Charts Company – the services are increasingly relying on big data to optimise what they are offering by understanding consumer behavior better.

Zane Lowe’s move to Apple indicates the service’s championing of subjective, selective curation, whilst also acquiring Semetric/ Musicmetric a data service similar to The Echo Nest signals to us that key distributers of music have already realized the importance of big data.

With Warner partnering with Shazam (which has 100m monthly active users tagging 17m songs, TV shows and ads per day in 2014) the label can look at the data that goes beyond the top tier of the charts.  As A&R becomes increasingly more crowd-sourced, Shazam is becoming a viable option in helping find new artists.

I know a label that have traditional A&R meetings and separate, Shazam-based meetings, which will look at the top 200 Shazam charts to find new opportunities across territories.

Lyor Cohen, former CEO, Warner Music Group, recently founded a label called 300.  In partnering with Twitter it allows 300 to tap into the music information Twitter has, and its location tags, enabling to talent spot globally, from its New York offices.  To put the volume of that data into context, there were 1 billion music-related Tweets two years ago.

Does this mean we should stop going to gigs and just be reliant on social media data?  Absolutely not – big data shouldn’t replace what we do.  It should inform about 20% of what we do, with 80% being the subjectivity, wisdom and experience.  The idea is that we should use that 20% to strengthen the other 80%.

Using big data isn’t about finding the next Adele or Sam Smith, but it might help you to look in the right place and get you there quicker.

We’re in a subjective business, but it is a business, where we have to make money and we do bear risk. What big data can do is help minimise that risk.  Forbes Magazine recently valued A&R as a $4.5bn industry.

If The Next Big Sound can predict album sales within 20% accuracy with 85% of artists as it promises to, and thereby reduce risk in this subjective business, then why not? I don’t believe this is ‘us’ being cautious and going for safe bets only, but rather keeping ahead of the game and empowering decisions with hard facts.

Ultimately it’s how you feel. So just because you like The Vamps, it doesn’t mean you’ll like another band that looks and sounds like them, like Five Seconds of Summer.  But you just might, and this is where big data can bridge that gap.  It goes beyond demographic and genre segments to look at your actual activity at a granular level.

It isn’t trying to create generic, swooping statements about consumers for marketing but reaching that single individual.  It’s about giving people the right content, at the right time, on the right platform for the right price.

In my paper I talked about big data in relation to consumer understanding and A&R, but it can go far beyond, into ticketing, merchandising, cross-media promotion and brand partnerships.

At the end of the day what we’re trying to do with big data is understand consumers better so we can give our artists a more fruitful and lucrative career.



Leave a comment:

Your email address will not be published. Required fields are marked *