Why Zane Lowe Could Do More For Discovery At Apple Than Echonest’s $25.6 Million Does For Spotify

BBC Radio One DJ Zane Lowe just announced a shock move to Apple. For the non-Brits and non-Anglophiles Zane Lowe is arguably the most influential radio DJ in the UK and is renowned for being a tastemaker with an eclectic pallet. His left of centre focus and his commitment to supporting and breaking new acts has allowed Radio One the freedom to be unashamedly mainstream in much of its other output. So why does this all matter for Apple? While it is not yet clear what sort of role Lowe will assume at Cupertino it is a move bristling with significance and a clear statement of intent from Apple.

Fixing the Tryanny Of Choice

The Tyranny of Choice remains one of the biggest challenges for streaming services, namely how to make sense of 35 million songs. It has been challenge enough for the Aficionados at the vanguard of the first wave of subscription service adoption. It is a problem of far greater proportions for the next wave of subscribers, the later adopters who do not have the expertise nor intent to invest great effort into discovering new music. It is not as simple as ‘lean forward’ versus ‘lean back’. But instead gradations between the two. Beyond Apple’s inevitable Spotify-subscriber win back efforts, these early followers will be at the core of Apple’s streaming strategy.

The 6th Of March: Man Versus Machine

Spotify showed its own music discovery statement of intent when it acquired the Echo Nest on the 6th of March 2014. Zane Lowe’s final Radio One show will broadcast on the 5th of March 2015, leaving him free to join Apple on the 6th of March 2015, yes, 1 year to the day after the Echo Nest. Coincidence? Perhaps. Either way, the symmetry of Spotify making its bet on algorithmic curation and Apple making its bet on human curation is unavoidable. It is man versus machine, with Apple for once coming down on the side of flesh and blood over technology.

However expensive Lowe’s salary might be, it will be far short of the millions Spotify paid for the Echo Nest, which had burned through $25.6 million of investment to get to that point. Yet there is every chance that Lowe, used properly, could deliver more value to Apple’s music discovery than the Echo Nest can to Spotify. Don’t get me wrong, the Echo Nest is a fantastic outfit with some of the smartest music analytics people going. Along with Pandora’s Music Genome Project the Echo Nest is as good as it gets for music discovery algorithms. In fact when it comes to implementation and cool data driven projects, the Echo Nest leads the way. But there is a limit to how far algorithms can fix the problems posed by the Tyranny of Choice.

Filter Bubbles

As Eli Pariser identified in his excellent Ted Talk ‘Beware Of Filter Bubbles’ there is a risk that recommendation algorithms actually narrow our choice and limit discovery. That by continually refining recommendations based on previous taste and choice they make our world views increasingly narrow and ultimately boring. Music discovery is not simply about finding music that sounds like other music we already like. It is also about serendipitous moments of wonder when something comes at us from the left field and leaves us breathless. That is the antithesis of ‘here are three other bands like this you might like’.

Of course it would be unfair to suggest that the Echo Nest is not sophisticated enough to engineer serendipity and surprise into its discovery system. (And Spotify is beginning to double down on human curation too). But the ability of a stack of code to perform this task versus an expert tastemaker is significantly less. And, another ‘of course’, it is impossible to definitively prove this one way or the other because ultimately the results are subjective and not properly measureable. Because one person’s awesome discovery is another’s sonic tripe. But that is entirely the point of the whole debate.

People Don’t Want Discovery, Well They Don’t Think They Do

There is a fundamental problem with algorithmic discovery: people don’t want it. In numerous consumer surveys I have fielded for numerous clients, respondents show little or no interest in discovery or recommendation features. Yet in the same surveys the vast majority of them state that they regularly listen to music radio, which is of course recommendation and discovery. The big difference is that it doesn’t feel like it. Instead it is an inherent part of the DNA radio. It is not an awkward artificial appendage that most people just don’t get.

Earned Trust

During his Monday – Thursday 2 hour show Lowe will play 20 to 30 or so tracks. Listeners know and understand that these are the tiny tip of the iceberg he has sifted through that week, that these are the songs he has decided are the ones that need to be heard. And when he announces his ‘hottest record in the world’ they know it is probably going to be something pretty special, even if they might not actually like it. His audience appreciates him that way because he earned their trust over weeks, months and years. That is the asset Apple are buying. Even if he has to earn that trust all over again with a new audience, that is the model.

If Lowe was simply to push 20 to 30 songs a day to Apple users (whether that be on a radio show on iTunes Radio, as an iTunes podcast or as an iTunes playlist, or all of the above) the odds are in favour of some or most of those resonating with a large swathe of the target audience. Even if just one track blows away just a quarter of the audience each day, the impact of one fantastic discovery will have more impact than a torrent of ‘sounds a bit like’ recommendations.

30% Not 80%

An Amazon Prime executive recently said that when commissioning shows he didn’t want hits that 80% of his audience quite liked, he wanted shows that 30% of his audience loved. That is what discovery is all about. Not being content most of the time, but being blown away some of the time.   Zane Lowe is not going to solve Apple’s discovery problem all by himself, but the hire shows that Apple is putting its money on moments of human magic being the nitrous oxide in its music discovery engine.


Google’s Acquisition Of Songza And ‘Fixing Discovery’

Google yesterday confirmed the much rumoured purchase of curated music service Songza for somewhere between $15 and $39 million. While it is not a vast investment for a company with the recent $3.2 billion acquisition of Nest as a benchmark, it is nonetheless a significant one for a company that already has a couple of streaming music services of its own. It is not a Beats sized deal but then if Google had wanted one of those it would have bought Spotify. So just why did Google splash the cash on Songza?

Access to all the music in thee world can be overwhelming, with so much choice that there is effectively no choice at all. This is the Tyranny of Choice. For all the efforts and intent of music services to ‘fix’ discovery no one has yet nailed it. Listen Services like Nokia Mix Radio, O2 Tracks and Pandora present one solution: effectively removing the burden of excessive choice by delivering a curated stream of music that requires little or no effort from the user. But this approach does not translate well to All You Can Eat (AYCE) services like Spotify and Googles’ Play Music All Access. These services are built on the foundations of giving access to everything, the exact opposite of what Listen Services are about. Which is why AYCE services are doubling down on enhancing their internal curation and recommendation capabilities. Spotify moved first with its acquisition of the EchoNest, Rdio followed by acquiring TastemakerX and now this move from Google. Beats Music took a different route entirely, building its service on the foundations of programming rather than superimposing it.

Google should be able to extract great value from Songza but as with all of these technologies it is just part of the solution. Human programming, as resource intensive as it might be, remains a pivotally important part of the equation, and though all the AYCE services have teams of curators, only Beats so far has done it at large scale.

First, Show People How To Find What They Have Already Found

And still the discovery problem is not fixed. Progress has been made in the last few years, but in many respects it is a case running before learning to walk. Recommendations, discovery and programming are just one part of the music consumption journey i.e. discovering new music. Arguably the most important aspect of the journey is the one that is most neglected: navigating the music people have already discovered. As counter intuitive as it may sound, people first of all need to be shown how to find what they have already found…their pre-existing music collections but also the music they have listened to in a service. Creating playlists and tags of songs is an often burdensome task that requires no small amount of discipline. Which means that newly discovered gems can all too quickly disappear back into bottomless pit of 30 million songs, rendering a discovery journey wasted.

Smart of use of data can provide the foundations for the solution, ensuring that people’s streaming ‘collections’ are dynamically created and programmed. But data alone is not enough. What is needed is an entire new paradigm in music navigation. For all the faults of CDs they were visual reference points. A consumer might not remember the name of an artist or an album but would know roughly where the CD was on a shelf or what colour the cover was. (I remember as a DJ often identifying a record I was about to play only by the colour of the label on the centre of the vinyl).

Digital music lacks such visual reference points. iTunes transformed our music collections into featureless spreadsheets, with playlists emerging as simply another means of sorting the data. New visually rich interfaces in music services help enhance the user experience but most often simply try to shoe horn in the old album art approach into a digital context. This new navigation paradigm must start with a blank sheet and think in terms of multimedia, interactive, dynamic experiences. It will need to leverage rich visuals, touch, dynamic context aware programming, sound, voice control and Shazam, to create an immersive whole that gives the consumer clear, immediate results in a way that engages multiple senses. Only once we have fixed this first step of the music consumption journey can we really start thinking about ‘fixing discovery’.