To say we work with a lot of data now is an understatement. Multifaceted media organisations have hundreds of data sources and are processing massive amounts of data daily. As the new music economy moves ahead we’re coming up against challenges that will only multiply as our industry grows.
What challenges are facing data science and music? Two of the most glaring challenges are disparate, disjointed data sources and also then how best to filter the insane scale of data to create actionable insights for our teams.
Gathering data from disparate silos is a headache; what causes migraines is the effort it takes to connect all of the sources in a coherent way. The music industry is particularly tricky in this regard, much of the incoming data comes from sources across the internet and with only loose associations.
It’s a challenge to make strategic creative and business decisions with this disjointed data. These data sources are all interlinked in the real world but the real value is in how to build relationships to make insight-lead decisions. For instance, how does a Facebook Ad campaign effect direct sales on Bleep (Warp’s online store), or to Spotify Streams?
An insight-led label takes a pragmatic approach to blend quantitative data with the qualitative knowledge from teams. The big caveat: understanding the need in separating correlation and causation. Just because overall streams for an artist has a correlation to Twitter impressions doesn’t mean one is directly affecting the other.
One way to take advantage of the relationships between all of the data sources is by applying insight from one platform to actions on another. When an artist has an opportunity that brings spikes of new listeners through a tour or television performance it’s important to be able to re-engage those listeners. Analysing the tremendous data from streams allows you to see who the new fans are from these opportunities and where they’re coming from.
The most interesting part comes when you can segment only the brand new listeners. With just that information, a campaign can map out highly targeted audiences for digital advertising. By running those targeted campaigns for your new listeners, you have additional opportunity to turn casual listeners into long time fans. By cross linking your demographics across platforms you’re adding new value to existing opportunities.
The other huge challenge, with data science and music, has been how to extract actionable insight from billions of lines of data. In order to wade through the noise it’s important to develop strong, clear Key Performance Indicators ahead of your campaigns.
In the previous example you’d use returning listeners as the KPI to determine whether your campaign was be a success. It’s one thing to have a clear view of everything that’s going on – it’s quite another to have a clear view of what you’re going to do once you’ve got the data. I was a massive proponent of “asking the right questions” back when I originally spoke on data in our industry – and we’ve certainly come a long way. Identifying major analysis metrics to help artists and campaigns allows for clearer goal setting and more insightful planning.
KPI’s help the success of goals at any particular part of a music campaign. Work toward the Primary Indicator (Streams, Sales, Revenue), and use tactical Secondary Indicators (Number of Playlists, Source of Stream, Returning Listeners, et cetera) to guide the overarching plan. Finally, use granular facets (Demographics, Age, Location, Context, et cetera) to really see if you’re moving the needle.
In setting KPIs, another interesting example is helping along the promotion of a new artist that has unexpected demographics. Combing through the data science helps to bring to the surface specific metrics to work on – increasing the number of listeners in those outlier demographics. This demographic could be very active on certain video sharing sites and niche blogs. Set your secondary indicator for getting increasing promotional activity on those sites / blogs, while having the primary indicator being to increase audience size and ultimately long time fans.
If we’re able to find the right relationships through all of our data sources and continue developing resilient KPIs to wade through our massive amounts of data we’ll come a long way in tackling our data challenges. Taking an investment in the application of data analysis is important in adding value to every part the new music economy.
This is the latest in a series of posts from key industry influencers from the world over, whom you’ll be able to meet at Midem 2017.