What’s your backstory? How did you end up launching Entertainment Intelligence?
I’ve spent about 30 years in technology working across different industries including banking and finance. I co-founded the ticketing company CrowdSurge in 2008 with Matt Jones and we did tours for pretty much everyone from Paul McCartney to Foo Fighters. When I left the company I wanted to address one of the biggest frustrations I had during my time there, which was getting information in one place. Pulling in tour information, artist information, images, and everything else to do with a tour was very cumbersome. So the original plan for Entertainment Intelligence was to address issues with tour planning and data management.
I had also started offering tech services to other parts of the music industry, and that’s how I started working with Essential Music on their distribution system. When I was there I met my future business partner, Erik Gilbert, and it became apparent to us that streaming data was going to be the biggest headache, so Ei pivoted into addressing that and that’s when we officially formed as a company. When Essential sold to Sony Red and eventually became part of The Orchard, Erik and I went independent. That was January 2016 and we’ve been doing this ever since.
Most DSPs have analytics platforms. What added value does Ei provide?
Platforms like Spotify for Labels are great services. We’re not scraping data or taking data that’s already publicly available – we integrate into Spotify, Apple, Deezer, Google Music, YouTube, Pandora and Amazon and we collect all of the granular data directly into one place. And when I say granular data it’s every listener, age, gender, location, the device they’re on, the source of the stream, playlists they’re listening to. It’s about 170 million rows of data a day.
From there you can start doing correlations. For example, you might have a label that’s doing very well in Sub-Saharan Africa but not so well in Asia. Or if YouTube is doing well, is it translating into other services? If Facebook’s got a lot of social messages going on, is it resulting in an increase in streaming or video views? Then you can start to do things like cohort analysis, which is what retailers have been doing for decades. What is a super fan? Who is influencing other people? You can do that once you’ve got all the data in one place and you can understand it at a very granular level. We’re helping to create a rich portrait of how users around the world are interacting with music.
“We’re helping to create a rich portrait of how users around the world are interacting with music.”
Who specifically is using your platform?
The majority of our users are people generating reports for artists and managers, as well as general management reports within their company where really granulated data is needed. We work primarily with independent distributors like Zebralution, Naxos, Secretly Canadian, Xelon – they have thousands and even hundreds or thousands of label clients, but the data feed is all done through them. We collect the data and manage it on their behalf and they share it with their clients. The level of access depends on the type of user, so labels and managers will get access to the data associated with the artists on their rosters. If they’re working with a PR contact, they can give them access to the data and stats surrounding only that specific release. Data sharing is very powerful. Version three of our platform is coming up very soon and that will be aimed at managers and business owners who need top line metrics.
You have a new feature called Indie Benchmarks (iB) that offers labels and artists actionable insights on par with internal tools used by the majors. What does that involve?
We looked at the sort of data that major labels are able to use internally and we realized that independent companies can’t hit those benchmarks because they don’t have such high volumes. So, we spoke to our clients and said, look, if we anonymized your data and gave it back to everyone as indicators of how things are performing, would you be cool with that? And thankfully, so far every single client has said go for it. For example, if you look at how many followers a playlist has – that’s not actually an indication of how popular that playlist is because those followers could have all joined on Valentine’s Day and never returned. Very few people unfollow playlists or artists, so what you need is an average number of daily streams for a single track on that playlist.
It’s all of those types of benchmarks. What’s the average audience? What’s the average age and demographic? What are the top locations? Which territories are over-indexing? One playlist might be really popular per capita in Chile, for example. You’re always going to see America, Germany, Japan, UK, etc. at the top but when you distill it down and take those out of the mix the spikes in smaller or emerging territories could be enormous. Those are the type of benchmarks that add flavour and help you to understand value. It’s a percentage game. I came from stocks and shares and portfolio management, and music is now like share trading. When you’re down to 0.0006 of a cent for every stream, moving the needle half a millimetre could mean another thousand dollars in your pocket. People need to know how to keep moving that needle and get those small increases.
“I came from stocks and shares, and music is now like share trading. When you’re down to 0.0006 of a cent for every stream, moving the needle half a millimetre could mean another thousand dollars in your pocket. People need to know how to keep moving that needle.”
It’s great to see the independent community working together and sharing data
Yeah, and it can happen more. There are still silos and certain groups that keep data to themselves because they think it’s a competitive advantage. But actually, if they were to collaborate, they would get better data in return. Data can help generate more income for the community if we all work together.
Can you tell us about the heartbeat feature?
It’s a feature that tells us when an older track is picking up steam. It could be a track that’s been flatlining for years and suddenly starts generating a heartbeat. In this case, the Ei platform will notify the catalog owner to let them know something’s happening. Often it’s the result of a big sync or playlist placement, and it gives our clients a unique opportunity to capitalize on movement within their catalog. It just happened with our client Epitaph. One of their smaller bands triggered a heartbeat because they’d been added onto a big Green Day playlist and are involved with their tour. Previously they wouldn’t have seen that until the statements had come in a month or two later when they’d have looked and thought, damn, we could’ve done something about that. If you know about it at the time, you can react.
There was another similar heartbeat example with the folk singer Peter Oren and Secretly Distribution’s label Western Vinyl. He suddenly had a 300% increase in streaming in a 48-hour period because he’d been added to one of the big folk playlists. Secretly were able to go straight in and push him onto several other playlists. Our client Africori noticed a massive streaming spike which came from the re-release of Miriam Makeba’s iconic album Pata Pata, which has gone crazy in America. That’s now an opportunity for them to look and think, let’s start making playlists about Miriam Makeba, with local talent seeded into that playlist.
You’re starting to apply machine learning to uncover relevant audience patterns. How does that work?
It’s early days for machine learning, but it’s really around the concept of looking at every single stream (there’s 170 million a day going through the system) and being able to work out which ones are lean forward and lean back. Which ones are super fans, which we categorize as users who have listened to at least 80% of an artist’s catalog more than 10 times. So, when a new single comes out, you can see how many superfans stayed with it afterwards and how many new fans came on board. Then you get into good skips and bad skips. Everyone gets really twitchy about skip numbers and completion ratios. You need about an 80-85% completion rate on streams before playlist curators and radio pluggers will look at a song. If someone who has already listened to a song 10 times or already has it saved and then skips it, that’s not a bad skip.
All of this stuff is very hard to do in code or query language. It’s just too much crunching. Machine learning can start to pull out those nuances and identify skips like that as a neutral skip. It’s not a good skip, but it’s neutral. Don’t panic about it. If someone skips the song the first time they hear it then that’s a bad skip. Eventually you’ll be able to weight them from very bad to don’t worry about it – they’re a super fan and they’ve got the song saved. Those sorts of subtleties are very important. Without them you end up with vanity figures or people panicking and throwing thousands of dollars into advertising.
You have access to so much valuable data. What trends are you noticing? Are playlists still as important?
Playlists are still a good way of collecting stuff together. We’re able to see what devices people are using, so we’re noticing a lot more smart speaker activity as well as smart TV and games consoles because now you can play games and listen to music at the same time. We’re even seeing digitally enabled car radio slowly starting to increase. The big thing with that is that playlists help you to ask for something quickly. Try asking Alexa to play !!! – the thing goes into meltdown. Ask it to play the band The The. I’d really love to see the royalty statements for The Band because I bet they’re getting a lot of their music played by mistake! With playlists you can just say, Alexa, play eighties playlist or whatever.
We’re seeing a lot of activity in emerging territories and mobile phone contracts. People used to say Africa wasn’t worth bothering with, but there’s millions of people there listening to music and now with mobile phone contracts they’ve got new levels of access. The emerging markets are going crazy so it’s knowing how to tap into that and knowing how to cross pollinate with appropriate music. You can speak directly to a kid in Lagos now that you couldn’t before. I think that’s stunning.
“We want to give indies the same level of power and utility as the majors, but that’s only going to work if people actually work together. So that’s our vision – to be the network that joins everyone together.”
What’s your plan for the future of Entertainment Intelligence?
Our aim is to be almost like a utility. There’s lots of innovation and cool stuff coming out but we would rather be the utility that puts the pipes down and creates that network that people can build great services on top of. Someone has to maintain the pipes and make sure they run as fast and efficiently as possible. We’re starting to work with companies for advert placements and advances on payments. There are so many amazing services but they’re all in isolation and should be working together. We want to give indies the same level of power and utility as the majors, but that’s only going to work if people actually work together. So that’s our vision – to be the network that joins everyone together.