July 12, 2016

The Limitations of Using Streaming Data to Pick New Music

In our last post, we outlined three reasons you should examine streaming data to help pick new music. Before fully embracing streaming data as a new secret weapon, here are five limits to what you can learn from streaming:

1) Streaming data only shows the positive, not the negative. There is a fundamental difference between radio listenership and on-demand listening that’s easy to overlook:  In radio, playing a song people hate hurts you because it can cause listeners to turn off your station. For on-demand streaming, however, haters have no impact—all that matters is how many people love a song enough to play it. Idina Menzel’s “Let It Go” was a #2 smash on streaming in 2014, but went nowhere on radio (much to the relief of Frozen weary parents everywhere). Silento’s “Watch Me” was the #3 most streamed song in 2015, but didn’t even make the Top 75 in radio airplay. In contrast, a song like Walk The Moon’s “Shut Up And Dance” was #2 in radio airplay in 2015 but only #31 in streaming: Fewer people might love “Shut Up And Dance” enough to play it repeatedly on YouTube, but enough people do enjoy hearing it when they happen to come across it and relatively few hate it, so it was a successful radio song. In streaming, you can have a huge hit with a song many dislike. In radio, however, ignoring the haters can cost you listeners.

2) Streaming data sometimes misses when people are getting sick of a song: In theory, when listeners are sick of a song, they stop streaming it. While this theory often holds true for CHR hits, there are other cases where songs remain among the most-streamed songs even though a lot of people are burned out on them. This phenomenon is particularly acute in Country, where a select few megahits remain relevant with listeners months and even years after they’re new. Sam Hunt’s “Leave The Night On”, Luke Bryan’s “Play It Again”, and Brantley Gilbert’s “Bottoms Up” remained among the Top 10 most streamed Country songs for well over a year. While many listeners did indeed still love these titles, Country stations with custom new music research knew that listeners were also burnt out on these songs. Stations wisely kept airing these titles, but as recurrents, not currents. Even in the pop music realm, novelty songs, songs people stream for the video, and songs popular with young children remain strong in streaming long after many people are sick of them.

3) Streaming data doesn’t tell you if a lot of people are playing a song, or if a few people are playing it over and over and over.  Streaming data doesn’t tell you how many people like a song—it tells you how often a song gets played. Ten listeners playing a song once count the same as one person playing that song 10 times. Anyone who has a young child knows how obsessed they can become with a song.  Was Silento’s “Watch Me” the third most streamed song in 2015 because a lot of people love it, or because eight-year-olds played it over and over and over? (I’m pretty sure my daughter single-handedly got “Let It Go” in the Top 10 back during the height of “Frozen”). In radio, however, we need songs that lots of people like, not just songs a few people passionately want to hear repeatedly.

4) Streaming data isn’t inherently ahead of the curve. There are a few cases where songs become hits on streaming before they’re on radio’s radar. In 2015, Fetty Wap’s “Trap Queen” broke on streaming before radio fully caught on. In 2014, Tove Lo’s “Habits (Stay High)” broke on Spotify months before radio embraced it. These songs, however, are the exception, not the rule. Overall, the typical Top 10 biggest songs on the radio are between 9- and 20-weeks-old. In comparison, the Top 10 biggest songs on the on-demand streaming services are also between 9- and 20-weeks-old. While this fact is an advantage for determining what songs listeners like most right now, it doesn’t provide an early read on songs’ hit potential.

5) Streaming data is subject to record labels’ launch strategies. Why do Taylor Swift songs not rank as highly in streaming as they do in airplay? Because her record label famously pulled her material off Spotify. Since labels would rather sell CDs than earn streaming royalties, withholding on-demand access to superstar artists is a strategy they employ to maximize revenue. Labels are increasingly allowing new releases to on-demand subscription customers before allowing ad-driven “freemium” customers to hear them. The labels call it “windowing”. It’s designed to force consumers to pay a premium for access to brand new music. For up and coming artists on the other hand, some labels intentionally try to break a song on Spotify to create a buzz about it before pitching it to radio. That’s how Tove Lo’s “Habits (Stay High)” got noticed. Without knowledge of the label’s windowing strategy, a song can appear more popular or less popular than it actually is. Unlike Taylor Swift’s highly publicized pullout on Spotify, most songs’ windowing strategies don’t make front page news.

While it’s important to keep these five limitations in mind, the additional information examining streaming data can provide about which songs listeners are playing most when they’re in control of their music makes streaming a valuable tool to have in your toolbox. When you examine streaming alongside sales, Shazam, MScore, airplay on similar stations, and your own new music research, you’ll have a complete picture of how your listeners are engaging with new music.

Where to find streaming data

Streaming data is compiled by Nielsen and published each week in the Streaming Songs chart, which includes a wide variety of audio and video streaming services such as YouTube, Spotify and Slacker (but not Pandora), and the On-Demand chart, which focuses on audio-only on-demand services such as Spotify.  You can access these charts at Billboard.com or from the Nielsen’s BDSRadio service.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>