“It’s really helped – both externally and even internally to be honest – to dispel the myth that discovery is a niche thing. A tastemaker thing.”
The speaker is Matt Ogle and the ‘it’ is Discover Weekly, the personalised playlist that Spotify launched in July 2015, customised to each listener’s tastes – and originally developed by Ogle’s team at the company for an internal hack day.
Ogle joined Spotify early that year following three years at This Is My Jam, the music startup that he co-founded to help people share their favourite songs. Before that, he worked at Last.fm and The Echo Nest.
Discover Weekly has been a hit. In its first ten weeks, Spotify users streamed more than 1bn tracks from the playlists, with the total reaching 1.7bn by December. Now, in 2016, Spotify is preparing its first marketing push for the feature.
Ogle and his colleagues, meanwhile, are working on some aspects that he thinks would ideally have been included at launch.
“When we tested Discover Weekly early on and realised that it was potentially the hit that it has now become, we actually took a lot of shortcuts,” he admits.
That included no ‘onboarding’ process whereby the new feature would be explained to Spotify users – for example a pop-up message explaining what it was and how it worked.
“We literally just added it to users’ playlists, which was how we tested it. All the way along I’d been saying internally ‘Oh, we’d never launch it like this. We’d never be so rude as to just jam it in everyone’s playlists!’” says Ogle.
“But this will now become a thing we talk about as core to the experience, so we have been figuring out the best way of telling new users ‘Hey, guess what, it’s your third week on Spotify and we’ve made something for you’, and getting their expectations up and running around that.”
WEEKLY HABITS
Discover Weekly has clearly been making ripples within Spotify, and not just around discovery of new music. Its weekly update has quickly become a habit for many users: on the one week so far when the refresh was late, many protested.
That idea of an ‘appointment to listen’ may seem strange on a service where people can listen to whatever they want, whenever they want. But just as scheduling has become important to popular YouTube creators, so it has found a place on Spotify.
“Something that Discover Weekly really has ignited within the company, which we’re going to be exploring in a bunch of different ways, is the power of that weekly ritual or cadence,” says Ogle.
“Right now we’re thinking about what the things are you might want to do on Spotify every day, what the things are that you might only want to do once a month. This periodic delivery isn’t the right fit for everything, but we think there are perhaps more opportunities than we perhaps realised initially to creating rituals and habits like that.”
Specific details on Spotify’s plans are under wraps for now, but Ogle does offer some hints. For example: “’Hey, here’s the hip-hop on Spotify at the moment’ – but you’re actually seeing a personalised ordering of all that content.”
He also suggests that Spotify is thinking about whether a standard playlist is the best format for this kind of personalisation. Spotify users have made more than 2bn playlists of their own, so they understand how they work. But an ordered list of songs could be improved.
“Playlists are the atomic unit of Spotify. We’ve got over two billion of them, so obviously we’ve got a pretty robust system for storing and serving them. However, as robust as that system is, it was designed for user playlists,” says Ogle.
“It was not designed for ‘Let’s set 75m of them to all get updated in a ten-hour window on Sunday night!’. So behind the scenes, we’re going ‘Alright, if we want to have 20 personalised playlists in a given week, what sort of a system would we need under the hood?’ What would we build so that you could just seamlessly serve that?”
Ogle also explains that while there are advantages to Discover Weekly being a playlist – Spotify’s users understand how to use and share them – there are also disadvantages.
“If we get a recommendation wrong you can’t tell us, and if you share Discover Weekly it’s just called ‘Discover Weekly by Spotify’. Hopefully your face is on it, but beyond that you can’t really tell whose is whose,” he says.
“We want to play with the playlist format a bit. It’s not the right tool for every job, and you’ve seen the start of this with our Genius stuff that came out recently.” By which he means the “Behind The Lyrics” playlists launched with Genius in January, which show lyrical excerpts, annotations and stories as the playlists play.
“If you imagine that every day of the week we might have a new type of content experience for you, being able to play with the playlist format itself is also something that we are thinking pretty deeply about,” says Ogle.
AUDIENCE BUILDING AND DISCOVERY
Spotify is also getting smarter about understanding how artists that feature in Discover Weekly are growing their audiences as a result.
That includes a new feature in Spotify’s Artist Insights analytics added a few weeks ago, where streams from all Discover Weekly playlists are aggregated into one source, so that artists and managers can see how many fans are discovering them through that feature.
“What we found is pretty shocking: there are 2,000 artists for whom Discover Weekly is currently 80% of their streams, and something like five or six thousand for whom Discover Weekly is half of their streams,” says Ogle.
He adds that many of those artists are seeing a knock-on effect where popularity in Discover Weekly – as measured by the number of people who save those tracks into their own playlists and collections – leads to them featuring in Spotify’s programmed playlists.
That journey starts before Discover Weekly, though. Spotify has launched a set of “Fresh Finds” playlists that aim to pick up brand new music even earlier in its journey: when it’s just being picked up on by blogs.
“Fresh Finds is doing an even better job of taking the stuff that’s really low-reach right now – almost no listeners by definition – and elevating it,” says Ogle.
“Quite often, getting featured in Fresh Finds helps an artist go from 500 listeners to 50,000 listeners. And if they’re making something good, people will save that track to their own playlists.”
“Meanwhile, Discover Weekly is powered by the curatorial actions of all our users adding things to their own playlists. So Discover Weekly is then the thing that takes those artists from that 50,000 level and tries to reach more fans. Fresh Finds accidentally but really serendipitously plugs in to Discover Weekly, which plugs in to this broader audience.”
For Spotify, which is trying to convince emerging artists that it can mean more for their careers than just (small) royalty cheques, explaining that journey with specific examples of artists who’ve built audiences (and just as crucially, made money) from this path will be important.
FINDING EARS WITHOUT OVER-PROMISING
Ogle remembers his time at Last.fm, when the company regularly argued that the service’s potential was “finding the right ears for any music – if there are even just 12 people in the world that like what you’re doing, we should have a platform that allows you to find those 12 people, or 12,000 people, or 12 million people”.
Now, he hopes that Fresh Finds plus Discover Weekly plus Spotify’s programmed playlists is finally putting that in place.
“I think we may be at a tipping point. I remember that when Facebook was growing, they were like ‘Oh, we’re going to make money because our ad platform means even if you’re doing something really niche, you can reach those people’. And it just wasn’t working for them,” he says.
“And then all of a sudden they hit this scale where kinda everyone was on Facebook, and so that idea that in a way, niche personalisation only works when you have a certain scale – I feel we’re arriving at that scale [at Spotify] and it’s maybe the first time in music tech that it’s really happened.”
Ogle stresses that he is keen to ensure Spotify does not over-promise on its personalisation, highlighting Discover Weekly’s relatively low-key launch: like “a friend giving you a mixtape” rather than any grand claims.
“When you have a bum week, people are like ‘oh, not a great week, I’ll see you next week!’ – there’s that forgiveness there, which I think might not have been if we’d been going ‘This is the ULTIMATE PERSONALISED EXPERIENCE! EVERY SONG’S A WINNER!’,” he says.
“Which it’s very easy to fall into when you’re building these things, because of course that’s what you strive for. Also, we have to remember that music culture is larger than any one service or platform, and it’s happening all around us. At our best, we just amplify the potential of these things. You can’t take credit for everything!”
DO MAINSTREAM LISTENERS WANT DISCOVERY?
Do most people really want music discovery, though? A recent post on Medium by industry consultant Cortney Harding included an existential question for anyone involved in music tech. Pondering the death of numerous music-discovery startups, she suggested this:
“Startup founders also overestimated just how much music matters to the average person. When you love music, you surround yourself with similar people, and that creates a confirmation bias — everyone wants to share playlists and discover new bands just as much as you and your friends! But really, they don’t. The average consumer is happy to listen to the radio or Pandora, see a few concerts or a festival once a year, and leave it at that.”
An alternative, optimistic view might be that the average consumer isn’t that interested in ‘music discovery’ because the bars have been set too high in the past to do it.
Ogle tends towards the latter, although he quickly admits that “it is possible to get high on our own supply with discovery” and assume that every listener wants it as much as people working in the field do.
“But i also think you really need to make it easier for people. Most discovery happens pretty serendipitously. Waking up on a Sunday going ‘Ah, I have half an hour, time to go discover some music!’ isn’t really a common mode of thinking,” he says.
“But of course, at Spotify we care about music discovery because we do want artists to be heard, and to find their fans and make real fan, not just drive-by listeners. So it’s absolutely something we wake up every day thinking about.”
“I do think it’s how do you package it: how do you present it, how do you take the intimidating-ness out of it, the fear factor out of it?”
That’s bringing changes to the way Spotify works internally. Ogle says that the company is realising that with things like Fresh Finds and Discover Weekly, it is “almost creating brands or shows” – he compares them to what Netflix has with shows like House of Cards and Arrested Development.
“You might come to Spotify for Fresh Finds, for Discover Weekly. They’re almost like franchises. And that’s a really different way of thinking. Until now we’d thought about pages or sections or features or screens,” he says.
“Spotify used to be organised around ‘Oh, I own the artist screen’ or ‘I own the Discover page’, and that was your kingdom. In this new way of thinking, it’s very different, and we’re still learning how to program to users.”
“It’s going from a world where we were a utility and you do all your own curation, through to a world where you can still do all of that if you want, but if you don’t want to do the curating, we’ve assembled all these things for you, and all you need to do is go ‘that one!’ and get the session going.”
TRACKING THE PATH OF SUPERFANS
That extends to the way Spotify is defining sub-categories of fans within its Artist Insights analytics dashboard.
“Fans” are people who’ve listened to an artist several times in recent months and have saved their music to their collections. “Streakers” have listened to an artist every day in the last week; “Loyalists” have listened to an artist more than any other; and “Regulars” have listened to an artist the majority of days in the last month.
The aim is to help artists understand how many superfans they have, rather than simply listeners. “As someone building and nurturing a product like Discover Weekly, the hardest thing to quantify is discovery,” says Ogle.
“We can say ‘Oh, users spent this many minutes on average in their Discover Weekly playlists this week’ and watch that go up or down as we experiment, and we can track how many saves happen every week. But people used to ask me ‘How many people discovered something new they really loved through Discover Weekly since it launched?’.”
“The cool thing about starting to create these definitions like superfans is now we can do studies like ‘Alright, over six months how many new superfans did Discover Weekly create and for how many artists?’ and you can quantify why we do what we do. It’s new for us.”
Ogle is clearly relishing life within Spotify, and while he’s heads-down with his duties, he does still keep one eye where possible on the external ecosystem of music-tech startups, as new teams take on similar challenges to his from This Is My Jam.
Are we in a fallow period for music startups, though? A lot of interesting companies have shut down or been acqui-hired – bought for their teams rather than their technology – and there are worries about whether funding will dry up.
Ogle has a more optimistic view, drawing on his past experiences. “Immediately after a consolidation or a bunch of shutdowns, which absolutely the industry’s been in for the last couple of years now, there always seems to be a resurgence,” he says.
Last.fm, for example, emerged in the wake of Napster’s shutdown, at a time when the online economy was not yet booming after the dotcom bust at the turn of the century.
“I feel like it’s just a natural cycle. I also feel that for better or worse, music motivates people and moves people beyond all our reason and sense and rationality,” says Ogle.
“There will always be people who could either create some new kind of B2B app or could create a music discovery product, and they’ll do the latter even though it makes no economic or business sense!”
Ogle hopes that some obstacles for music-tech startups will resolve themselves in the longer term – for example, the issues referred to by entrepreneur Brenden Mulligan in another Medium post: where startups with tools for grassroots artists find that the latter can’t afford to pay for them.
Ogle does have another doubt, though. “We still see people trying, for example music messaging, over and over. It’s died down mostly now but there’s still a lot of direct-to-fan, ‘What’s the new MySpace, what’s that thing where artists talk directly to it, wouldn’t a social network around music be cool?’ thing going on,” he says.
“I wonder whether the general-purpose social networking / communication / publishing tools gotten so good that when Beyoncé or some indie artist wants to talk directly to their fans, they just do it on Instagram or Snapchat?”
“Have we evolved to a point where we actually no longer need purpose-built music experiences in most categories, because the general-purpose tools can be used so effectively by artists? Even with Jam, it was ‘How is this better than just sharing that YouTube link on Facebook or Twitter?’.
IF YOU CAN’T BEAT ‘EM, WORK WITH ‘EM
Ogle adds that even Spotify is “taking a fresh look” at how it interacts with those large social platforms.
“We don’t want to own all the conversation around music, all the messaging, all the et cetera. So how do we just make sure that as seamlessly as possible we can be a part of experiences on those other platforms?” he says.
“That’s something we’ve tried very hard at in the past. It’s something that we’re thinking of doing more on now.”
Spotify’s recent partnership with Facebook Messenger may be just the first of many such integrations. In the meantime, Ogle says that Spotify is working hard on ensuring more hacks like Discover Weekly can make their way into its consumer product.
“A lot of the efforts underway internally right now are all about how we can lower the cost of experimentation. As soon as running an experiment becomes internally costly, it’s the worst thing for product people,” he says.
“Because what that means is you never release anything, or it’s going to take so long to release anything that you’re like ‘It has to be perfect before i can release it!’. And then you start talking to other product people, then you have more ideas and it’s back to the drawing board to add this feature and that feature.”
“Discover Weekly was just a hack from the team. We want to make sure that we can have 100 more of those in future, but you need to get the conditions right internally. That’s always hard, but it’s fun to figure out how to do it.”
Tools: platforms to help you reach new audiences
Tools: Kaiber
In the year or so since its launch, AI startup Kaiber has been making waves,…