Updating the mechanism for counting plays

Hi everyone,

Daniel and I have been busy hacking away on the apps, and should have something to preview to show off soon.

And as a part of the apps, we’ve been working on a new infrastructure for counting plays:

That is, at what point after clicking play on a track, should we count it as a play?

We’ve identified some problems with the current way it’s done, and would like to share our proposed solution to get feedback.

Current issues

Plays (sometimes called streams on other streaming platforms) are used to calculate which artists should your subscription go to and by how much.

Currently, a play is recorded the moment the play button is clicked or track is loaded, irregardless of how much of the track was listened to.

This has a handful of drawbacks, namely that when browsing around the site and playing multiple tracks it generates a lot of “unintentional” plays, which can end up diluting your money from going to the artists that you actually listened to, and presumably want to support.

Secondly, these unintentional plays end up influencing the charts, and so tracks that are skipped over will show up in the trending pages.

Lastly, it creates a lot of noise in your recently played list, and you might find tracks in it that you intentionally skipped, or just don’t remember listening to at all.

This isn’t to say that we want to reduce the number of plays reported or pay less royalties. The amount of royalties paid will be the same thanks to the user-centric model. The aim of this is to improve the accuracy of the royalties.

Proposed solution

What we propose is to use a less boneheaded approach and instead record plays after 30 seconds of listening to a track, or 50% of the track duration, whichever is shorter.

This can be any 30 seconds: If you scrub to the last 30 seconds of a track and start playing from there, it will count.

Repeat plays, when we get round to adding a repeat mode on the player, would also count as plays.

The main benefit of this would be that casual skipping through tracks would not result in a play being recorded, and so you can browse tracks a bit more freely without worrying that it will affect your distribution at the end of the month.

As a bonus, the technical infrastructure we’re planning on setting up for this would allow for detailed analytics about your tracks, e.g. how much of a track is listened to on average, percentage of skips, heatmaps, etc. :exploding_head:

This is pretty on-par with what Spotify, Apple Music and Soundcloud all do, but we’re interested to hear what you think about this.