The Rise of OPEN Innovation: The 3P’s for Building a Durable Open Software Company
Accel recently hosted more than 100 of the top open source executives and thought leaders for a full-day event covering the next wave of great companies building atop OPEN technologies. This article highlights some of our learnings from the event around the future of open software development and how to build sustainable business models around it.
In the mid 80s, Richard Stallman ignited the “free” software movement, advocating that users of software have the freedom to manipulate their software as they saw fit. Years later, the open source software movement extended the scope of Stallman’s dream — redistributing power from tech giants to software developers.
The first generation of open source companies was less about providing free software than pushing forward a new development methodology for software. It introduced an exciting new world of opportunity, where developers were no longer tied to proprietary vendors who dictated the pace of innovation and how to deploy technology. This shift from a vendor-centric view of the world to a user-and-customer centric view of technology is a profound one and is the reason why open development is here to stay.
An OPEN Innovation Framework for the New Generation of Software
First let’s examine the reason why the first generation of open source software (OSS) businesses failed to reach venture scale (with the sole exception of Red Hat).
The initial impact of OSS offered customers a wider buffer of options, at lower prices. Ubuntu, for example, offered a Linux-based operating system as a substitute for Windows. MySQL offered a database alternative to Oracle (which Oracle later acquired). But ironically, greater control in the hands of the customer decreased market opportunity. Because these businesses were 100% open source and essentially offered feature parity, the first generation of OSS were only able to charge for support and services. Under the ideology of “pure open,” few of these companies were able to reach true scale.
Not only did this business model have inherently lower financial margins than normal software products, it also created a destructive loop that de-incentivized innovation. Since profits relied on customers needing support, companies had no motivation to innovate and improve upon their product. In a support-only model, ironing out wrinkles could only lead to smaller profit margins. This, in turn, could only further handicap their ability to invest in research and development, especially compared to incumbent software vendors.
Today, we’re witnessing the rise of a completely new generation of open source companies. The current generation of open source companies — like Databricks, Docker and Elastic — each bring novel functionality to the market. And in doing so, these new open source technology companies are expanding the market opportunity far beyond what incumbent vendors serve today.
We’ve seen hundreds of open source companies, and have invested in dozens. To a large extent, open software development has gone from the exception to the rule. The majority of new infrastructure software startups are now ‘open companies’ and almost all application companies are, at minimum, building atop the open software stack.
Though there were few success stories in the first generation of OSS, a handful of modern OSS companies are already reaching meaningful revenue run-rates and a framework for open-businesses is finally starting to emerge. The solution for creating valuable and sustainable OSS businesses stems from progressing through three company-building phases.
The Three Phases of Open Software — The 3P’s
As open development companies grow and mature, they pass through 3 stages, the 3P’s, that are formative for the business. Companies that operate within this framework, sequence the phases correctly, and nail all three, can build a large business. Those that don’t adhere to this framework will have a harder time enduring.
I’ll highlight the 3 phases in this article, and dive into each in greater detail in future posts:
- The Project phase — developing an open source project and a vibrant community around it.
- The Product phase — laying the groundwork for packaging and deploying a monetizable version of the project.
- The Profit phase — doubling down on a core customer-base and builds the business to scale.
Phase I: Project — Community
From the start, the open source movement has been bolstered by vibrant communities of software engineers and developers trying to solve problems. Open businesses don’t start with a product — they start with a project.
For traditional software development, the question from the outset is “will people pay for it?” In the project phase of open framework business question is: “Can I build something that someone uses and cares about?” It’s a fundamentally different question, and one with a radically different set of decisions to be made.
In this phase, the sponsoring company should be striving to establish the voice of the brand and proliferate organic, word-of-mouth buzz and adoption through the open source community. Rather than validating a product by seeing if people are willing to pay for it first (i.e. developing software behind closed doors), project-market fit is about achieving validation through the community and building a core base of contributors who believe in the project, build on top of it, and ultimately propel it forward.
Apache Spark and the Rise of Big Data
Databricks, for example, is developed completely on top of the Apache Spark project, which began as research project in 2009. Like the best open source projects, Spark began with a problem to solve. It became incredibly obvious that data-intensive companies like Netflix and Twitter — needed a faster way to be able to crunch data at near real-time speeds. Some of the existing applications built with Apache MapReduce often took all night to run and were clunky and difficult to use.
This was the pain point Spark set out to soothe. And it did — Spark’s in-memory and iterative computing capabilities are easier to code and can improve performance by orders of magnitude for a variety of streaming/interactive analytics and machine learning use cases. It created a powerful new functionality that the open developer community immediately seized upon and could mobilize behind.
Spark leveraged the existing Hadoop community and initially targeted users of MapReduce. Zaharia hosted community meet-ups to discuss Spark, published educational materials, and allowed external contributors to help develop and build Spark outside of UC Berkeley. As the project’s popularity has grown, it has gradually helped to expand the broader Hadoop community and functionality set.
The community presents real-use cases, applications, and experiments of the project that can be ideated in phase II.
Phase II: Product — Definition and Packaging
In the project phase, you’ve built software around a community’s real need, but this doesn’t necessarily mean that you have a sustainable business. In the product phase, open businesses need to find ways to ‘productize’ the open source project at healthy margins.
With free software available, first generation businesses struggled to sell their product, which led to unhealthy models that discouraged innovation. The new generation of open companies are taking a different approach. They are continually innovating — both at the outset to create products that provide novel functionality, and over time as they add more functionality to both the open platform and proprietary add-ons around it. This is a win-win as the community gains a more consistent value stream thanks to pricing models that enable the vendor to package and deploy commercialized products that deliver value to customers above the core open source project.
At the end of the project phase, an open company might be 80% services and 20% product. By the end of the product phase, this should flip.
Docker’s Product Deployment
Docker began as a Platform-as-a-Service (PaaS) provider called dotCloud. The company had built powerful container technology that allowed applications built on a laptop to run universally in different environments. In 2013, dotCloud founder Solomon Hykes decided to completely pivot and open source the container technology, which meant creating a totally new product.
While keeping the open source project free, Docker is moving toward meeting real business problems through its commercial product, and is developing an increasingly sophisticated set of tools to meet these needs.
While Docker’s core open source project is free for anyone to use, its monetization strategy revolves around making its software easier to manage and secure. For now, the majority of Docker’s revenue comes from offering support and services in the traditional open source model. Most enterprise customers, for example, run thousands of different applications, and typically need to operate them behind firewalls. While you can do this with the open source version of Docker, paid products like Docker Engine and Docker Trusted Registry make it easier to securely run applications from anywhere — it’s a convenience that the enterprise both needs and is willing to pay for.
Phase III: Profit — Scale and Go-to-Market
By phase-three, an open source company will have built a roadmap to monetize its product. Now the business must scale by growing sustainable revenue streams beyond venture dollars. But how do you go from a tight-knit community of early adopters and software engineers to reaching a large net of paying customers?
This is the fundamental question open companies need to answer in order to step out of the red and into the black. While each user may be a lead, only a small percentage are likely buyers. This requires a radically different sales and marketing approach than traditional software vendors.
Cloudera: The First Open Enterprise Data Platform
Cloudera began around the core Hadoop projects and was founded by engineers from Google, Yahoo, and Facebook — the same engineers who planted the seeds for the original Hadoop project.
The majority of Cloudera’s revenue comes from its enterprise data hub deployment. The enterprise version of Cloudera offers proprietary software to handle data management, system management, governance, security and access control features. It also includes search, faster processing, data optimization, and predictive support & maintenance. For the fiscal year of 2015 (which ended in January 2015), Cloudera hit a milestone of $100M in revenue, with around 100-percent year-over-year software revenue growth.
When it came to turning a profit, Cloudera strategically set its sights on the enterprise market, and built a wide-range of partnerships to bolster this goal. Intel has partnered with (and heavily invested) in Cloudera, along with Accenture, IBM, Cisco, and Deloitte. This robust systems-integrations ecosystem leads to increased trust in Cloudera’s enterprise product by signaling the market’s larger belief in Cloudera’s long-term viability — while expanding the functionality of its product.
It’s an upsell model — give away the basic infrastructure and open source platform away for free, and scale the business profitably by offering proprietary products, add-ons, and strategic services. The money Cloudera makes creates a virtuous cycle — it pours profits back into developing the broader ecosystem around Hadoop — and makes Cloudera one of the most important voices in the community. It invests heavily in and around large number of open source projects, including Kafka, Hadoop, Impala, Kudu and Spark, as well as its own proprietary layers.
The Open Software Future
The realm of possibility is endless for open technology. OPEN innovation powers progress at some of the largest and fastest-growing companies in the world — hundreds of billions of dollars of equity value have been built on open foundations. Companies like Airbnb, Facebook, Google, Netflix and Twitter all actively leverage and contribute to open projects.
With OPEN innovation, software developers around the world are able to innovate collaboratively at a truly massive scale, without autocratic lock-in. The new generation of open companies have learned from the mistakes of the previous one, and instead of de-incentivizing innovation they encourage it every step of the way. Each of the three phases reinforces another, implementing a flywheel effect that brings to the table a completely new set of possibilities.
In this short period, ‘open’ has grown from the “new” model to the “only” model. If you’re part of this community of entrepreneurs and technologists, we’d love to hear from you.