My Writing

How we signed up 10,000 paying subscribers in just over a month

For about a year, I’ve been working with the “Catalyst” team at Envato to design, develop and launch a compelling new subscription for graphic designers, marketers & creatives. Our new product, Envato Elements, beta launched on August 1 and attracted around 10,000 paying subscribers in just over 1 month. That’s a very fast start for a new subscription!

Reflecting on that experience, we identified five (5) factors that were critical to this success. See the article on Medium to find out what they are and why they matter.

Read the full article “How we signed up 10,000 paying subscribers in just over a month”

Pioneering Subscriber Share for Contributor Payouts


Most subscription services that use third-party content – such as music streaming services – use the so-called “big pool” method to work out how much money to pay individual contributors. That method has come under fire for being unfair to indie artists and for being vulnerable to click fraud. But it’s far and away the simplest and most common method:  i.e. put all the money to be shared into a big pool, and divide it based on the number of uses.

For contributor payouts on Envato’s new subscription for digital assets – Envato Elements – we chose not to use big pool, but instead a novel method called ‘subscriber share‘. We think it’s much fairer to contributors, and should also lead to a more diverse and interesting content library for subscribers.

Exploring subscriber share

I first became aware of this approach in a great article by Sharky Laguana called How to Make Streaming Royalties Fair(er).  He kindly responded to my email, answering various questions and explaining that there was considerable interest in his proposed method. However, there did not seem to be any large-scale implementation by a major subscription service, so for many it remained simply a good idea but not really a practical option.

An Associate Professor, Arnt Maasø, also wrote about the method (calling it ‘user centric’ instead) in his article A Better Way to Cut a Cake and Eat It Too. Interestingly, he published a detailed paper that analyzed data from two months of data for a Norwegian music streaming service (WiMP). As far as I can tell, that’s the only analysis based on real-world data looking at the impact of big pool versus subscriber share on contributor earnings (for now!)

After learning as much as we could about the proposed approach, we explored some scenarios of our own. It quickly became clear that the subscriber share method had significant advantages over big pool. It was significantly fairer, more resistant to fraud, and aligned everyone’s incentives. So we decided that we wanted to use it for Envato Elements, if it was feasible in practice.

A proof of concept

Before committing to using subscriber share, we had to be confident we could implement it at scale. After all, our new unlimited download subscription was expected to rapidly attract tens of thousands of subscribers. After a few half-joking suggestions that “we’ll just work it out in a spreadsheet”, I eventually set about building a payout calculator using the statistical programming language R (often used by data scientists with large datasets)

My first step though was to map out an algorithm to implement it, based on data we’d have about subscribers, contributors and usage events. This was harder than it sounds, and it quickly became clear it was a good plan to use a sophisticated tool like R, given the nature of the computations required.

With the algorithm in hand, we moved on to a proof of concept using sample data generated for this purpose (since we didn’t have a live subscription service yet!). After reassuring myself, my colleagues, and then finally our finance department, we finally committed to using subscriber share for Envato Elements contributor payouts.

Implementing subscriber share

The payout calculator has evolved significantly since that initial proof of concept, with several major re-factoring efforts, as well as hooking it up to use production data. We enlisted the support of our finance development team for unit testing of the functions involved, and also conducted integration tests by manually calculating (painfully!) the expected output from a given set of input data.  Finally, we wrapped it all up into an interactive application using the Shiny library for R, so that a non-technical user can select the time frame for analysis, the data inputs, and a few other relevant parameters.

Coding aside, our implementation also involved a lot of effort around communicating what subscriber share is, why it matters, and how it works.

This is vital for our contributors, since it plays such a large role in determining their earnings from participating in our subscription service. But it’s also important for subscribers, helping them understand how their behaviour influences contributor earnings and encouraging them to behave responsibly with their unlimited download access.

This educational effort for subscriber share includes the following resources:

We’re also making subscriber share a big part of our PR efforts in launching the new subscription. We think it’s an important part of our offering, as well as a newsworthy innovation in its own right.

What’s next?

Our new subscription – Envato Elements – had its beta launch on August 1st, and official public launch on August 29th. At the time of writing, it’s still early days so we haven’t yet done a payout run. But in the coming months we’ll look very carefully at the actual data on subscriber usage and contributor payouts.

Ultimately, I’d love to prepare a paper that systematically examines the impact of our choice to use subscriber share, and compares the distribution of payouts for subscriber share versus big pool with real-world data. I feel that this would be a useful contribution to the world’s understanding of the subscriber share method, and perhaps encourage other subscriptions (such as those in the music streaming industry) to consider adopting the method.

In the meantime, I encourage you to go read the article on Medium