There’s More Than One Way to Skin an API: The Analytics Arms Race
The tale will be a familiar one, pick up any industry trade magazine, or open their email newsletter (it’s 2018 after all and well.. the internet), and you’re certain to come across an article talking about the launch of a new analytics / insights portal / tool / app from a label, aggregator, distributor, label services company, artist AND label services company, tech start up etc etc etc. This new tool you’re reading about (which sounds awfully familiar to the other new portal you read about last week) promises to revolutionise how artists and managers interact with their fans, it’ll process millions of rows of Spotify and Apple Music data in an instant, it’ll tell who is listening to your music, it’ll tell you where they’re listening to your music, it’ll tell you how they’re listening to your music and it’ll tell you what they had for breakfast (not quite.. but you can guarantee whatever it was they listened to Coffee + Chill while eating it).
It all sounds very impressive. In fact, to be honest, it is very impressive. All of the data that is now available to the industry is indispensable so having a tool that can help labels, artists and managers make sense of it is incredibly important. Despite my thinly veiled sarcasm I couldn’t be more of an advocate for these tools, Hell for the best part of half a decade I’ve built, rolled out and championed them myself. I’ve lived through the music industry’s digital revolution and seen first hand how transformative access to this data has been, particularly for the independent sector. It’s helped level the playing flied ever so slightly and empowered smaller artists and labels with the same insights the majors have access to.
My issue isn’t with the tools themselves, but rather with the amount peacocking and one-upmanship that goes on in the media and press releases between the organisations trying to use access to this data as a USP. Kobalt led the charge with access to data and transparency, and they quite rightly take a lot of the plaudits, but since they launched their first data portal we’ve seen competitor after competitor try to compete with similar offerings, sometimes inferior, sometimes superior, it doesn't really matter because ultimately, and here’s the secret, the data Kobalt has access to is identical to the data Believe have access to, which is the same as the data FUGA have access to.. and PIAS.. and EI… and Beggars.. and Domino.. and Universal.. and Sony.. and us here at the Mushroom Group. It’s all exactly the same.
Spotify’s new API, available to master rights holders, has around 30 unique fields. These include user ID, territory codes, gender, age, skipped flags, playlist name — nothing overly groundbreaking. There are only so many ways to present this data; some people like to see data visualized in a table, or a bar chart, or a pie graph – how “good” a system is can often come down to the personal preference of the person using it and how they like to absorb data. Ultimately as long as the system you’re using allows you to overlay each of these data points (so streams by territory by gender by device etc) then you stand a pretty good chance of being able to derive maximum value from it, and really user it drive engagement in your campaigns.
Yes there are huge challenges associated with big data and analytics projects these largely lie in and around having the right infrastructure to process data at volume and speed. Interpreting this data, not to undermine those who work in this space (including myself), can often be the easy and fun part.
What prompted this blog was an article I read recently where a well known operator in the digital space was talking about their label and artist facing analytical portal. I’m going to be vague as to not reveal who I’m talking about but the article included a line that read something like “our tool, unlike our competitors does X. That allows us to do Y and Z, and we’re the only people doing this”. The thing is… X is not unique to this company. In fact it is a well known and widely adopted, I’ll go as far as saying almost universally adopted, process used by anyone operating a music industry, big data system. The comment annoyed me because it was misleading, and in an industry where our data literacy skills aren’t yet where they need to be, many people find the technical aspects of data and analytics intimidating and confusing. If, as an industry, we’re pushing for greater transparency shouldn’t this extend beyond just the raw numbers themselves?
Maybe I’m being too puritanical. After all I’m a numbers guy not a marketing guy.
What I feel we should be doing, particularly in the independent sector, is rather than compete with each other to essentially “sell” the same system just with a different logo and colour pallet, is to pool our data to maximise the insights we’re able to derive from it. I’m not talking about giving another label your detailed streaming data, I’m not saying tell a close competitor how many streams a specific artist or track of yours has had, but rather sharing top level, anonymised data such as demographic data for specific playlists. I’ll explain..
The Spotify playlist ‘Hot Hits Australia’ currently has 50 tracks and 949,410 followers. Everytime a track from this playlist is streamed the age and gender of each user (well call these metrics user demographics) is capture and fed back to the label that owns that track. The more data we have for each of these 50 tracks will help us build up a picture of who a typical ‘Hot Hits Australia’ user is. This information can then be supplied to the label marketing and promo teams to help them know which of their artists they should be pitching to Spotify for inclusion in ‘Hot Hits Australia’ vs ‘A1 Hip Hop’ or ‘Indie Arrivals’. The problem is the only people who know exactly who a typical user of ‘Hot Hits Australia’ is is Spotify themselves, as they’re the only ones who have access to data from all 50 tracks in the playlist. As of today the 50 tracks in ‘Hot Hits Australia’ are owned by 11 labels, the 3 majors and 8 independents. Universal have 19 tracks, Sony 14 tracks, Warner 9 tracks and the 8 indies have a track each. Universal, with 38% ownership of the playlist have access to the most streaming data for this playlist, so have a pretty good chance of being able to figure out the demographics of a typical ‘Hot Hits Australia’ user. That’s compared to the indies who each have a 2% ownership of the playlist.. nowhere near enough data to be able to get a reliable take on user demographics. But what if all 8 of those indies pooled their streaming data for this playlist together? Again, not detailed streams by track but purely the number of streams reported to them for ‘Hot Hits Australia’ by gender and age. All of a sudden they go from having 2% of the picture to 16% of the picture. Not quite Universal’s 38% but still a much bigger, much more reliable, data set than their individual pools. This is something our friends Greg Delaney and his team at Entertainment Intelligence having been working on with their Playlist Benchmarks feature and we’re loving it so far.
One of the fundamental building blocks for any big data system is to not silo your data. Try where you can to keep everything in one place so you can pull it apart and analyse everything together, but as an independent community, by not sharing our data, by keeping it in our own silos, we’re not getting the most from one of our most valuable assets.
Lets get sharing.