How big data is transforming the music streaming business

Nielsen Music’s 2016 half year sales numbers reported the worst year for album sales (vinyl, CD and downloads) since records began, way back in 1991. That was the year “Ice Ice Baby” topped the charts. Sales of albums were down 14% from 2015 to 100m; CD sales continued to decline 11% from 2015, selling just 40m units.

On the face of it, streaming is coming to the rescue. The format has already overtaken the album/download market in terms of song volume – 209bn songs this year alone. In 2015 Spotify’s revenues jumped 80% to £1.5bn, and in June reached 100m active users, 30% of which pay for the premium service with no ads and offline functionality. Apple now boasts almost 50% of Spotify’s paying subscribers (13m v 30m) and is catching up fast, mainly due to Spotify switchers and Apple’s huge installed base of 800m iTunes users, complete with credit card details.

The benefit of personally “owning” music (even in the nebulous, digital sense) has been trumped by choice, flexibility and on-demand. Smart phones, apps and hi-fi headphones are the enabler here, making streaming accessible, cost effective and cool.

There is, however, a dark side to streaming’s impressive growth. Alongside Spotify’s jump in revenues was a jump in its losses too – $215m in 2015, up from $184m the year before. The company was forced to raise a $1bn round of convertible debt financing earlier this year to fund operations as investors were not prepared to put their hands in their pockets at the current (and apparently unrealistic) valuation.

Deezer postponed a run of funding due to bleak forecasts, and after Rdio filed for bankruptcy was quickly snapped up by Pandora – who itself lost $85m in one quarter last year and are being pushed by investors to find a buyer. Live 365 and Songza have closed up shop already.

If streaming is the future, why is it so fragile?

Streaming service’s underlying margins are wafer-thin due to what looks like an unsustainable business model (except for tech giants like Apple and Google, of course). Music industry veteran Mark Mulligan estimates that a streaming music subscription business can at best only achieve a 3-5% operating margin – this puts the cutting edge of the music business right on the edge.

The fundamental problem is that the majority of streaming revenues are paid out to rights holders – record labels and recording artists. As competition for exclusive deals with top artists grows, so do royalty costs (Spotify’s jumped 85% in 2015).

Compounding this, license deals are commonly front-ended, guaranteeing record label’s revenues no matter how well the service, and their particular content, performs. This safety-blanket ends up saddling the service with huge amounts of debt and increased risk. (It is worth noting that the safety blanket is mainly warming those who need it the least due to sell-out stadium tours – major artists like Adele and Coldplay).

ARPU is not rising at the same rate as costs, and consumer pricing seems stuck at around $10 per month. On top of that, music streamers are young, mobile-oriented digital natives making up a massive 50% of the UK’s 12-24 year olds – this is generation free-content, making margin relief from price increases challenging indeed.

Amazingly, when Mark Mulligan predicted 3-5% operating margins, he caveated this by saying it was only sustainable if the streaming service under-invested on product development and marketing.

Sky-high royalties and inflexible front-loaded license deals means an enormous amount of working capital is needed to build and operate services – plus invest in the next generation. This is giving advantage to tech giants like Apple and Google, and making it harder for startups who might bring something different to the market, cater for niche segments and enhance customer choice.

How big data is on track to save the music streaming business

The music streaming business has gone through successive waves of digital transformation from albums to CDs, then from downloads to streaming, and was battered by its battles with Napster and its copycats over IP. The investment into digital platforms has been huge; worldwide standards have been established, billions of dollars spent on branding, cross-platform apps with slick user experience launched and huge catalogues amassed.

Big music labels are starting to fight back against the "stream-rippers" – sites that enable music streams to be converted into downloadable files. Capitol Records, Warner Bros records and Sony Music, to name a few, filed a law suit against YouTube-mp3.org in September 2016 and their alleged 60m unique users. But more is needed.

To survive, streaming music platforms have to move into a second stage of digital transformations, and the contours of a new industry have already begun to emerge. Although digital transformations are underway in operations and business modelling, I believe the biggest impact is being felt in two big areas; how music is made and experienced.

Data is transforming how music is made

In 1999, the pioneering data scientists (also presumably music nerds) at the The Music Genome Project sought a systematic, digital method of analysing and organising music. They broke down music into 450 distinct “genes”, like guitar solos, vocal style, sound distortion and harmonic arrangements; for example rock has 150 genes, rap has 350 and jazz (predictably) has 400. After the project floundered, the complex algorithms became part of a new venture called Pandora that started out by assembling digital radio channels, the forerunner of playlists. Another pioneer, Hit Song Music, went further by predicting chart success by comparing new songs against past hits.

Later platforms like The Echo Nest (snapped up by Spotify) processes over 30m song scraped from the web using complex big data algorithms to identify songs via audio fingerprinting. The University of Antwerp in Belgium used this data to predict 65% of Billboard’s top dance tunes. A much better strike rate than Simon Cowell and X Factor.

The impact of these digital technologies on how commercial music is made and selected is profound. Labels can use predictive tools to select which artists to sign, which tracks to release and, when combined with listener data, how to target individual audiences. Music X Ray offers a service directly to bands that analyses songs and predicts the likelihood of them being offered label opportunities.

All of this data and insight helps match music to listeners – not just mainstream, but across a myriad of niche audience segments. It makes services sticky and acts as a doorway to discover new products, increasing ARPU, reducing churn and importantly helping transition freemium users to paying customers.

Music by the numbers must seem highly artificial when set against the musician with nothing but a guitar, notebook and existential angst – however it is proving to be highly effective. And furthermore, when data predicts successfully what TV shows we will like (Netflix and House of Cards), what books we would like to read (Amazon) and who we should connect with (LinkedIn, Facebook) – why should music be any different?

Data is transforming how music is experienced

Note the word listening, long associated with music, is missing in this heading. In many industries, the biggest and most visible impact of digital transformations has been in customer experience – and the streaming music business is no exception.

The most basic part of this transformation is the inclusion of video. Since the advent of music television (MTV) in 1981, music and video have been inseparable – now they are becoming indistinguishable. Spotify added videos to its apps in January 2016.

YouTube Red (originally named YouTube Music Key) is the service’s long awaited subscription service will feature music playlists and the ability to listen to music with the screen off, or when using another app. And how much will this set you back? You guessed it, $10 a month. 

Ironically, the biggest transformation in listener’s music experience is also likely to be the biggest transformation for the fortunes of recording artists as well. Lets look at both sides.

From the listener’s perspective, next generation streaming interfaces will soon go beyond curated and predictive playlists to feature bios written by the band (not critics), videos, video blogs, backstage picture streams, their music recommendations, social streams and ratings.

Services like Bandpage already serve as a platform to aggregate a basic version of this experience, and now license it to big players like Spotify, Deezer and Soundcloud. Soon our music streaming services might resemble Facebook more than music playlists, however intelligent.

 This development is arguably far more valuable for the recording artists. In the vinyl and CD years, they had little information on who bought their music (neither did their labels, for that matter). This started to change with the advent of iTunes downloads and now two-way portals on music streaming services look set to change the game completely.

Artists will soon have rich information on their listeners preferences, what they listened to when, and on what device. It will know where they are located, enabling them to “customise” concerts to play the right songs, in the right order. Using Bandsintown, they can understand what fans are interested in attending concerts on what dates (see below). The opportunity is emerging to offer tracks for free to test out new ideas, sell T-Shirts, VIP concert passes, and with companies like Songkick, even sell tickets direct to customers and get customer data withheld by traditional ticketing channels. In short, one massive backstage party, bringing artists and fans together.

The changes underway in streaming music have a number of parallels with the ongoing battle in TV between broadcasters and subscription based streaming services like Netflix and Amazon. Direct relationships with customers, analytics, fast feedback loops, customer service expertise, personalisation and the ability to acquire new customers online are now key success factors. How broadcasters respond to these challenges, especially using their live programming and strength across the value chain, will be interesting to watch.

In conclusion...

The wafer-thin margins of most music streaming businesses are unsustainable, as are their business models that are tipping the playing field in favour of tech giants like Apple and stadium-filling super-bands. It's stifling innovation and reducing choice.

The good news is that in recent years, a better understanding of how to build digital capabilities, lead transformational journeys and implement appropriate digital governance has accelerated the rate of positive change significantly. Andrew McAfee described this era as the “Second Machine Age” where technology and business combine to deliver new levels of productivity, performance and profit. We have arrived at this tipping point in many industries, but in music streaming faster than most.

Music streaming is fast becoming a convergence of music, broadcasting, social, e-commerce and mobility. Industry boundaries are blurred. This digital transformation is going further than mere salvation – it’s enabling an inherently creative industry to innovate and grow, be more inclusive and diverse, and share value with all stakeholders more equably. This should be music for everyone's ears.