Heroic Labs' Journey to Open Source and 5.3M Docker Downloads
November 21, 2022
This article was originally posted onHackernoon
Many companies start out with an open source product and later add proprietary offerings, but what if this path is taken in reverse? This is a question that Heroic Labs came face to face with. They began as a SaaS company but eventually recognized the benefits of pivoting to open source in their quest to deliver gametech tooling and infrastructure that empowers game studios and publishers to develop real-time, multiplayer, competitive, and social games.
On The Hacking Open Source Business podcast, CEO Chris Molozian and Head of Developer Relations Gabriel Pene at Heroic Labs elaborate on this shift in more detail. This blog post is a summary of their conversation.
Why open source?
For many startups, their top priority is growth at all costs. For a startup in the open source space, the idea of boosting adoption now and worrying about monetization in the future likely becomes even more magnified, although other key metrics such as revenue, profitability, and ROI must eventually be considered. The initial mindset might work for a B2C company such as Facebook, which productizes the individual and leans on advertising to pay the price for growth, but the B2B world is different. The popularity of a project might advocate for product market fit, but successful commercialization is much fuzzier. You might have a large developer audience operating with your technology, but convincing them to see the value in paying for it directly is the challenge.
Nevertheless, open source opens up the opportunity for developers to give to the world something that they have created. It is a chance to contribute skills to a community as part of a bigger cause than just your own project. Plus, if your product resonates, converting even a small percentage of users to paying customers would present a huge win.
Proprietary model: The starting point
Prior to going open source, Heroic Labs joined Y Combinator in 2015 and assumed a proprietary model. At that point, they had already built out the first version of their gametech using Erlang and Elixir. They sought to provide a mobile backend as a service via Parse- or Firebase-like capabilities for the game development world, and their specialization would center on game APIs. In other words, you would get your developer dashboard token, drop it into your SDK, and make API requests to store your data. After Y Combinator, Heroic Labs concentrated their efforts on helping game studios adopt and integrate their proprietary technology.
Open source model: The turning point
Over time, however, questions among customers increasingly arose: What would happen to the technology if Heroic Labs were to disappear one day? How would I support my game? If the infrastructure were no longer available because Heroic Labs ceased its tech distribution, what protections would exist? Given that Heroic Labs stewarded core data, the frequency of such concerns grew, and Heroic Labs needed to rethink their distribution model.
After four months of internal deliberation, Heroic Labs concluded that their particular set of technologies was more appropriately categorized as core infrastructure. Their technology stood at the heart of businesses and served as their lifeblood, so it needed to be just as dependable as one’s database technology. In line with the precedent set by database companies, which rely on an open source model to achieve the standards necessary to protect and assure their customers, Heroic Labs followed suit. Furthermore, they rewrote their technology in Golang to reach more potential contributors and adopters.
This approach appeared not only to address concerns about the continued availability of their technology but also best cater to the needs of their target developers. They wanted to make game development so accessible that an established studio backer would not be required. They believed that anyone has the potential to develop a hugely successful game, and as it grows, so would the need for the product.
4 open source models to consider
As part of the process, Heroic Labs surveyed the four current main open source business models:
- Support: A classic example of this is Red Hat. You can use the technology for free, but to troubleshoot any problems or challenges that you encounter, you pay for support.
- Cloud: This includes pretty much every database company (e.g., Elastic and MongoDB). When the complexity of your core tech is high, you build a managed service so that developers can use it without operational overhead.
- Consulting: An example is Ruby on Rails, which generates revenue through a strong network of partners, consulting companies, and digital agencies that encourage people not only to use the technology but also give back to it.
- Open core (aka commercialized open source software or COSS): You offer some parts of the technology for free but other parts are premium, such as a dashboard or enhanced monitoring. These are components that larger organizations will want but only with the backing of a trusted commercial entity.
Heroic Labs opted for a combination of # 1, 2, and 4, ensuring that they offer something for almost every potential customer:
- They built a support model for production users who may need hot fixes or want to request prioritization for new features
- They built a managed service via Heroic Cloud for studios that prefer not to operationalize the technology but have someone else do it for them
- They have a licensing model for the clustering technology that sits outside of the core code base, enabling various game features, including messaging, multiplayer chat, and matchmaking; they also offer clustering technology that enables a singular player base at scale
Because of open source, Heroic Labs can now provide the exact same technology for everyone—from the largest game publishers worldwide to independent developers in their garage. Open source democratizes Heroic Labs’ software, lowers the barrier to entry, and enables access to previously costly software.
Open source metrics: Measuring adoption
Leveraging an open source approach is not without its challenges. The line between commercialization and community can appear murky. You want to foster transparent relationships with users but need to determine success with numerical metrics: conversions, revenue, and the like. One thing is for certain: Marketing within the open source space is different. Gating and pressuring users to sign up, register, and provide personal information simply go against the open source grain. Thus, metrics tend to prove tricky.
Fortunately, the Developer Relations team at Heroic Labs found a metric that works for them by looking at pure adoption in the form of Docker downloads, even if it does not reflect unique users perfectly. A download from Docker may trigger the corresponding registry to note it, but it is what we are used to seeing with the internet, which matches well with people’s expectations. Heroic Labs currently sits at 5.3 million Docker downloads and counting, and they use Scarf to help with more advanced metrics. In light of growing sensitivity to how companies capture and use data, it seems wise to shift the emphasis from embedded telemetry to distribution line.
If there is anything that you can gather from Heroic Labs’ journey to open source, here are some key takeaways:
- Open source provides end users with a hedge against your business failing or changing and leaving them without something that they rely on
- It’s never too late to switch to open source–it is easier to go from closed to open than from open to closed
- Although you can drive open source adoption through various avenues, there are some simple directionally correct ones, for example, “real" downloads (i.e., packages or Docker images), but you have to look at unique downloads or downloads from a single source
- You don't have to stick to a single open source model; often companies focus only on cloud or building a managed service, but plenty of large companies want otherwise
If Heroic Labs took the open source route from the very beginning, they believe it would have accelerated their work. They could have leveraged the maturity of the product that the open source community already built and progressed from there. Still, even after two years, it was not too late and they made a successful transition!
To learn about it in further detail, you can watch the vodcast below or listen to the audio-only version, where additional topics such as, “Is there a trend toward open source in the gaming industry? If so, why now?” are discussed. You can also read more case studies on our site.
Latest blog posts
Tools and strategies modern teams need to help their companies grow.