We're listening to more new music, but that creates its own challenges
Smart speakers are having a huge moment – the sales, projections and uptake of these devices is staggering. But while they’re changing many aspects of how we live, it’s the way we enjoy music on our devices that’s really getting flip-turned upside down.
A quick reminder of the numbers. According to Canalsys, 33.2 million smart speakers were sold in 2017 and that number is expected to rise to 56.3 million for 2018. What’s more, 60% of smart speaker owners use their device to listen to music – by far the biggest use case – and it’s changing the way we enjoy our favourite artists.
FutureSource found that owners of smart speakers are four times more likely to pay for a music subscription, and an NPR and Edison Research study revealed that 28% of smart speaker owners claimed that their purchase had caused them to fork out for a music sub.

We all know about the success of Spotify and Apple Music (which only plays via the recently released Apple HomePod) but the success of Amazon Echo is causing its own ripples in the music pool. Amazon Music is now the third biggest service out there – but the really surprising fact is that it sees more consumption via Alexa devices than via smartphones.
Of course, Alexa can access Amazon Music’s limited library, bypassing the need for an Amazon Music Unlimited pass, thus any Prime member can take advantage. But the effect of millions of Amazon Echos entering people’s homes is having a seismic effect on the music industry.
Algorithmic taste makers
But it’s the way in which music is consumed by smart speakers which is shaking things up. While streaming services put music at our fingertips, when you add voice search into the equation, our demands change. We’re more likely to bark vague, yet niche commands at our speakers: “Play me upbeat party music”, “Play Sunday morning vibes.” Those types of commands are very open, yet personal to you. What I like to listen to on a Sunday morning is different to you.
That’s a headache for music services, which are having to turn to machine learning in order to create the metadata required to service our needs. As Stuart Dredge wrote on a Medium post:
“Amazon has been pretty open about its willingness to create metadata that Alexa needs, if it doesn’t already exist. In fact, this is a task well suited to technology companies: they can use machine-learning to comb a catalogue of tracks and figure out what’s upbeat, what’s sad etc.
“What comes out in response may be hyper-personalised to our tastes, using everything the music service knows about us,” he wrote.

People are using smart speakers to discover new music, perhaps more than on phones or PCs. FutureSource found that 30% of people use smart speakers to discover music, and 78% of them do that daily.
But that in turn creates fascinating problems when it comes to discovering music via voice – in other words, finding music that users don’t even know the name of, as Ryan Redington from Amazon Music explained to Billboard:
“We had never thought about radio impact date as a critical metadata field in the past, but in voice-enabled environments, customers no longer have visual access to ‘Chart,’ ‘Browse’ or ‘New Release’ tabs and campaigns.”
“When customers ask for the ‘latest song by Justin Timberlake,’ we want to pull up the song that [the label] is working on the radio circuit, without the end users needing to know the title of the song.”
It highlights something that we’ve maintained on The Ambient since the beginning. That switching to voice isn’t a magic bullet; and when it comes to music, vocal search issn’t always the best vehicle for getting what we want.
But the evidence is that the power of smart speakers is such that, rather than opt for something less convenient, people are changing their habits to suit the device. And one could argue that represents an even bigger shift than streaming, which broke down the barriers of album-based consumption to the embarrassment of music riches we enjoy today.
What this represents is a lurch towards algorithmic listening, which will place our tastes into the hands of technology giants. And it’s impossible to know the effect that will have.