If a particular track is broadcast on a TV or radio station, or simply played in a cafe, bar or shop, the not-for-profit PPL (or Phonographic Performance Limited) is there to collect royalties on behalf of those who hold the rights to that track and those who perform on it. In total, PPL collected £212.1m ($273.6m) in revenues in 2016, up 8% on 2015.
And in 2016, the most played track by UK businesses was Justin Timberlake’s Can’t Stop the Feeling!, while Coldplay were named as the most played artist, PPL recently revealed.
Getting insight into these metrics makes for a huge data management challenge, according to Matt Phipps-Taylor, head of insight and innovation at PPL. The organization has 90,000 members and 11 million recordings sitting on its databases, he says. Every week, it receives 35,000 new recordings.
Even more complex is the data about when and where those recordings are played to the public, he says. PPL provides licenses to some 380,000 businesses that play recorded music to the public in the UK, and collects royalties from 84 organizations that perform the same role in other countries:
A particular challenge for us is the complexity of our data. It’s well-structured, to be fair, and we’ve got pretty good MDM [master data management] processes in place, but it’s still far from simple. In a classic BI environment, you might be analysing data on orders, or online shopping carts, or mail outs - all kinds of things. But music is harder to quantify. There are different data structures around different rights holdings in different countries, different rules. It all gets quite complicated, quite quickly.
To add to the complexity, this data is stored in different systems across PPL, too. In the past, Phipps-Taylor explains, access to that data by business users within PPL came in the form of different tools, installed on their desktops, with frequent silos of information created, duplication of data presenting a problem and a need to often reconcile data across different tools.
Putting it all together to generate insight via end-to-end processes has been quite difficult. We saw the need for a more product-based, on-demand way of providing analytics that fit neatly into the business processes that employees here perform every day. And we really didn’t want to embark on yet another enormous data warehouse initiative. I feel that the days of being given the time and the budget to go away for six months to a year to develop something like that are long gone.
Instead, PPL has implemented a cloud-based analytics platform from Infor-owned Birst to bring together multiple sources of data into analytical ‘products’ that are developed and managed by a dedicated product team and are then consumed by different users working within PPL. Since PPL went live on Birst in September 2016, it has rolled out four such products, with three more planned by the end of 2017.
The most recent to go live, for example, focuses on the collection of royalties internationally on behalf of members, from those organizations overseas that perform the same role that PPL does in the UK. This international performer claims process involves collecting data from the corresponding overseas organization, matching it to PPL’s own data and then issuing a claim to that organization. But producing the report needed in order to make a claim involves cross-checking data held in six different PPL systems. Birst pulls all the relevant data together in the cloud and allows business users to review the claim, check its contents, apply business rules to identify any potential issues and make amendments before the claim is issued.
The Birst implementation at PPL combines its cloud-based analytics with Amazon Redshift, the AWS data warehouse service. As Phipps-Taylor explains:
The fact that Birst itself is cloud-based is great - that leads us to some nice economies in itself, in terms of cost and the ability to scale up and down. But the storage element is important too. We have a distribution engine within PPL that ingests the information on what got played, who owns the music, who performed on it, what the rules are around payment and so on, and then produces very fine-grained calculations that say, for example, ‘This performer is due X amount of money for this play, on this channel, on this data, because of this reason.’ This is what allows us to give really good insight into where the money came from that we pay out to members - but it also produces huge volumes of micro calculations, billions of rows each time.
We obviously don’t want to throw that stuff away, but neither do we want to pay over the odds for storage or have our data center keep adding new hard drives, so being able to have Birst with an Amazon Redshift back-end was a really attractive combination for us.
From projects to products
At present, he adds, PPL’s approach to getting new products up and running and in the hands of business users has been quite project-oriented, but over time, the organization wants to focus more on managing and evolving an established range of products:
The wonderful thing about data and insight is that as soon as you start sharing it, people start changing the ways they work in very positive ways. If you don’t keep up with them, you’re no longer supporting them and the new opportunities they’re finding for your organization. We’re very passionate about using analytics to drive processes and we’re finding a lot more opportunities to do more of this, so our ‘to do’ list right now is pretty long.