Cumulative Percent Streamed as a Metric

Currently there are two (non platform specific) metrics for gathering song stream data:

• Number of streams

• Minutes streamed

I would like to propose a third option; Cumulative percent streamed.

Instead of counting minutes streamed, count the total proportion of a song streamed. As far as I can tell, this eliminates a number of drawbacks of counting either minutes streamed or number of streams.

Technical Yap (optional read)

Number of streams has the drawback of still counting skipped songs in full. That means if a song gets automatically queued (i.e. by shuffle), it will still count.

This is easily resolvable by weighting each stream by how long it was listened to for. Thus we obtain the minutes streamed metric. Unfortunately, the minutes streamed metric over weighs longer songs. It’s still a useful metric, but doesn’t tell the full story.

Thus, the cumulative percent metric. Instead of weighing songs by the number of minutes streamed, weigh them by the percent of the song streamed.

As an example, the metric would work as follows:

If someone listens 2 minutes through an 8 minute song, the cumulative percent streamed would increase by 25% (or one quarter).

If someone listens 2 minutes through a 2 minute song (i.e. the whole way), the cumulative percent would increase by 100%.

This is mathematically equivalent to dividing the number of minutes streamed for a particular song by the length of that song. It’s also equivalent to the number of streams multiplied by the average percent of the song listened.