On Spotify’s Train Wreck Recommendations

Back in secondary school, resting on one’s laurels would earn the dreaded “Must try harder!” remark from the teacher.Spotify recommends the Black Keys

Fast forward to 2014 and that’s exactly the treatment that hardcore music listeners should be giving to Spotify for its frankly appalling email artist recommendations.

I enter for the court’s consideration, exhibit A:

“Because you listened to: The Gaslight Anthem

…. drum roll…

The Black Keys!!

Hooray!”

 

Exaggerated for effect, sure, but a train wreck recommendation for a regular listener if ever I saw one.

And before I come off as some elitist nerd music buffoon (too late), here’s a raft of reasoning as to why this is just a boring, basic recommendation with no thought for user experience:

  •  The Gaslight Anthem are popular, but The Black Keys are another level entirely. They crossed into mainstream awareness a few years back and can sell out Madison Square Garden. If I know about The Gaslight Anthem, there’s very little chance I’ve never heard The Black Keys.
  • This isn’t an isolated incident. In every recommendation mail I give another chance, Spotify will send a handful of fairly well-known artists, most of whom I’ve listened to before. There is no discovery happening here.
  • There’s no way to adjust the recommendation settings to tell Spotify’s algorithm I’m okay with it getting a bit more adventurous. I get that users with less listening history than me may want popular recommendations, but this is a service dedicated to all kinds of listener and with 20 million tracks at its fingertips… we need some control!

Black Keys listens on my LastFM profile

  • Finally – and most importantly – Spotify has my listening history, so it knows exactly how many times I’ve used its service to listen to The Black Keys. I can’t access the exact number in their database, but most of my plays on Last.FM are sourced from Spotify, so a good proportion of those 331 spins are on Spotify’s books. How the hell are you recommending a band to me that’s almost at the top my listening count?!  This is the antithesis of discovery!

Getting Recommendations Right

I get it, this is a minor gripe in an otherwise excellent service, but it’s just because I’m rooting for you, Spotify.

Apple, Google and Amazon are all lining up for our streaming dollars as part of a much wider master plan, but I at least credit you guys with a passion for music… for satisfying listeners. With that in mind, I have the following suggestions to get music discovery right for every listener, every time:

  1. Cross-check recommendations with listening history and filter out artists with a significant number of spins. Better still, allow users to import listening history from services like Last.FM so that speccy nerdy music geeks can get a head start on discovering gems in your vast archives.
  2. Set up easy-to-use filters that let us customize the recommendations. Safe to adventurous. Similar styles to new genres. Closely related artists to loosely connected. Any of these and more would help users to tell you what they want to see.
  3. Have a clear recommendation notification sign up from the get-go. It’s been so long since I first signed up that I forgot what I agreed to, but guiding listeners to a variety of ways to receive suggested new music (recommended playlists, email notifications, app notifications etc.), combined with those filters we talked about, would make for a much better user experience.
  4. Make it easy to completely opt out of all recommendations and discovery. Some people like what they like and don’t want to be bothered with anything else. I don’t understand them in the slightest, but we should respect their desire for musical ignorance all the same.

What else would you add? Have you actually found an artist you’ve never heard of but now love, thanks to Spotify’s suggestion emails?

Let me know in the comments here or on Twitter/Facebook. Together, we can discover better music.

 

(And for anyone who truly hasn’t ever heard The Black Keys, here’s what the fuss is about…)

 

From Listen to Live Show, Music Streaming Just Got Smarter

Spotify has undoubtedly made my music listening life infinitely better, but there remain several areas in which it can improve. Recently the guys and gals in green made a huge stride in one of these areas, namely that of applying data to discovery.

Spotify Discover Function

From listen to live show, can apps like Songkick drive up show attendances?

I have plenty to write about on music discovery via algorithms versus the human touch, but here I simply want to shine a light on the elegant simplicity of integrating local concert listings into this broader recommendations channel.

 

From the Stream To the Show

Who makes up your music audience?

Image Credit: Anirudh Koul

This may seem like a natural extension – indeed, one that was already available by visiting the existing Songkick app tab – but the significance of bringing the show alongside song recommendations should not be overlooked.

Many users look for guidance on what to play when they first open a listening platform such as Spotify, meaning that the Discover page will be a highly visited area. Throw in what appears to be a much refined recommendation engine, one that has thrown up some genuinely intriguing unknowns for me in the last week, and you have the potential for a lot of eyeballs perusing these listings.

If even a small percentage begin to show an interest in the concert element of the page, it seems like something that all parties involved would benefit from developing further.

Personally, I see an increasingly valuable place for services like Songkick in both becoming a go-to source when I want to browse gig listings and delivering concert news to me. Combining my online listening history with that service helps to filter and improve the latter, making both services even more useful and raising the likelihood that I can be persuaded to purchase a ticket.

 

Next Steps

From integrating Facebook data to recommend shows based on the upcoming events of friends, to converting fan follows and listener likes into information that artists can use to better target their marketing, there are a great many extensions of this move that may bode well for music makers.

At a time when streaming services are regularly under fire for simply making money off the backs of the creators whose content fuels their business model, it is heartening to see moves being made to use the vast data sets they collect to pull fans further into the music.

Whether or not such connections actually drive up sales and attendances remains to be seen but, as any marketer will tell you, visibility, relevance, and a compelling call to action are key. Functions like Spotify’s Discover begin to solve the first two elements, but there will be a great deal of tweaking and dealing on the third before we begin to see a truly valuable connection between the listen and the live show.

 

Google+