Our Co-Op Spotify Playlist

I’d been kicking the idea of a co-op playlist around in my head for a long time. Around the beginning of the year, a friend prodded me to finally give it a go.

Here’s what I proposed:

Each artist pays $200 per month for four tracks, which can be swapped out monthly and/or on release day

In lieu of paying, I’ll curate the playlist and manage the ad campaign

We’ll start with 5 friends I know and trust = $1000 per month in ad spend

After the first month I’ll invite another friend or two, as long as the ads continue to scale

At the beginning and every 3 months, I’ll create or the artist supplies 4 ads = 24 ads total in the ad group

The song from the top performing ad goes in the top slot of the playlist

The rest of the songs will be evenly spread out among tracks from bigger artists

I’ll add three outside tracks for every one of our tracks, so 96 tracks in the playlist to start

I won’t accept submissions

I’ll write about it in my blog at some point, but until then, let’s keep it a secret!

Unfortunately the meritocratic approach didn’t pan out. The ad campaign produced no clear winning song, or more precisely, different winners on different days. Even when an ad performed well, some songs produced more playlist followers than others.

So I made the executive decision to use my song “Crystal” in the ads and at the top of the playlist. Selfish perhaps, but it had already proven itself across multiple ad campaigns, and it allowed me to track the growth of the co-op playlist relative to my Vocal Synthwave Retrowave playlist. The two campaigns are mirror images of each other, and cost per conversion has moved in lockstep.

Here’s the top-performing ad:

I somewhat balance out my song’s top position by placing my next song in the 25th spot, after everyone else’s second song. There are currently 90 tracks in the playlist, with the last co-op track in the 55th spot.

My goal is to create a playlist that fans of 80s synthpop will genuinely enjoy. I update it every month by swapping in some of my favorite new releases. It’s available on Spotify, Apple Music, and Deezer.

Other than not swapping out the ads, my initial playbook has mostly held. We’ve stuck with just 5 members, but there’s no reason we can’t open it up to more, assuming the quality is there and the genre is a match.

For the most part, things have gone smoothly. At the beginning of each month, I post a status update on a dedicated Facebook group for the co-op.

One artist bowed out after six months for financial reasons. Everyone else confirmed that they were pleased with the results and eager to continue. I was willing to cover the missing $200 myself, but fortunately I found someone to fill the slot right away.

That’s not to say there haven’t been any hiccups! Last month I rebranded the playlist to what you currently see: an image of no artist in particular, though he reminds me of a young Morten Harket from a-ha.

Prior to that I was using a variation on Depeche Mode’s Violator cover, with the trademark rose over a bed of stars:

I figured it would catch fans’ eyes, and perhaps it did, but it also captured their ire. Every day, people commented to point out that Depeche Mode wasn’t in the playlist, or that I shouldn’t be using that image.

So I tested out three different covers in ads, and Morten Harket won. The ad cost settled back to where it had been, so apparently the rose wasn’t especially eye-catching after all.

After 6.5 months, we’re up to 5344 followers on Spotify. If there’s a way to track playlist growth on Apple Music or Deezer, I’m not aware of it.

Meta results (click to enlarge)

We generated 29,738 conversions (clicks from the landing page to a music service) for $6488.23, for an average cost per conversion of $0.22.

15% of clicks were to Apple Music and 3% to Deezer. Here’s how that looked over time:

You’ll notice a gaping hole in the middle. That’s when I switched from Hypeddit to SubmitHub Links, which has a lot going for it, on paper anyway.

After switching, our cost per conversion decreased, but so did follower growth, until it became obvious that something was wrong.

You can see that the number of daily followers (in green) declined until the middle of May, when I switched back to Hypeddit.

As for why, I’m not entirely sure. Sometimes ad campaigns go off the rails and Meta starts optimizing for the wrong thing, like bots instead of people. That’s not necessarily a knock against SubmitHub Links, but I wasn’t going to take any chances with other people’s money.

Currently, $10 of our daily $33 budget goes to Brazil and Mexico, entirely on Instagram. The remaining $23 goes to “high paying” countries, mostly tier 1, with $15 to Instagram and $8 to Facebook.

Could we get cheaper streams? Absolutely, but it would defeat the purpose of the playlist, which is:

To reach legitimate fans of the genre who may actually make a purchase at some point

To teach Spotify who those real fans are and encourage more algorithmic streams

In our case, they’re probably not in India, Indonesia, or The Philippines.

Three artists have already seen big algorithmic successes since joining the co-op, though of course I can’t say for sure that the playlist is responsible.

Another artist has several playlists of his own, and is sticking with the co-op because it outperforms them on a cost-per-stream basis.

I suspect the remaining artist would see more algorithmic growth if his co-op streams weren’t dwarfed by those from third-party playlisting services.

Here are listener and stream totals for the three artists I have Spotify for Artists access for:

My numbers are about 50% higher, but not for the reason you might think. Two artists selected Color Theory remixes to include in the playlist. Having the first track certainly helps, but not as much as having two extra tracks!

While everyone is happy with the way things are going, there’s certainly room for improvement.

Adding a couple more high-quality artists would boost our ad spend and improve the ratio of co-op versus outside tracks. Meta routinely nags me to raise my budget, so our cost per conversion shouldn’t increase significantly.

What do you think of our little experiment? Got any advice? Let us know in the comments!