Open source business : lessons learned

Initial thoughts on Opencell’s business and service model.

Published Date: 20 Dec 2015
By: Ethan

The world’s most innovative and successful companies are increasingly turning to open source models. Last month, Google announced that it was opening up the code to Tensorflow, its AI software engine.  Two weeks ago, Facebook announced that it was open sourcing the design of its AI hardware.

For many, it will probably seem strange, or at the very least counter-intuitive, to think of building a company that would make its intellectual property available to all.

And, when we started Opencell nine months ago using an open source business model, I must confess I had my initial doubts as well. Having spent many years working at companies where extracting every last cent from our intellectual property was a company ethos, it wasn’t easy to understand how a startup could give away its intellectual property for “free” and how we could build a viable business model based exclusively on providing support and professional services to users of our software. At one point, I thought close-sourcing certain parts of the source code or features would probably be required in order to achieve the minimum level of monetization projected in our business plan.

After benchmarking other open source software companies and testing our model on prospective clients, we did however finally decide to launch Opencell Software with a standard open source business model:

  1. We adopted a dual licensing model. Our community edition source code is published under the “standard” AGPLv3 licence that contains strong copyleft provisions. In addition, we provide an entreprise edition that contains exactly the same source code and features under a commercial software license. This gives enterprise users to avoid the requirement to make eventual source code modifications available to the community.
  2. We decided our support model would be primarily indexed on platform billed revenue. Professional services for training, configuration etc. would be charged an hourly basis and would only be available to clients under a support contract.

So far, we’ve been able to execute according to plan. Our revenue and margins are on target and we’ve been able to finance a big increase in our R&D spending. In a nutshell, we’ve found that:

  1. Our customers are actually willing to pay for support. Since our software is a critical application, the support model has good traction. Our clients want to know that they can call someone in the event they run into a problem.
  2. Our business model works… but not exactly the way we expected. For the moment, our support revenue has been smaller than expected since many clients are still in a project implementation or ramp up phase but, overall, we’ve hit our targets for our first financial year since we’ve had greater demand for professional services and custom development. Over time, we still expect that the portion of our support revenue will represent our main source of revenue. 
  3. Our product roadmap is financed collaboratively with our clients. When there’s a feature gap in our software, we’ve found that our clients are willing to co-finance development in order to have these features added to the Opencell core as quickly as possible. This was something of a surprise but it seems a powerful trend. Increasingly, I think we’ll see Opencell, and other open source specialized enterprise software companies, operating at the center of a network of clients that will actively contribute to financing our software development, in order to get the functionalities added to the core as quickly and as effectively as possible. 
  4. Avoiding any distinction between the code contained in the community and enterprise versions was the right way to go. During commercial discussions, prospective clients invariably want to understand how committed we are to open source. We often get compared with other “open source” companies that aren’t really open source anymore because too many key features have been reserved for the enterprise edition. Being uncompromisingly committed to maintaining all core functionalities in our community edition makes our value proposition compelling and easy to understand. 
  5. Open source is a great model for a startup that targets larger, institutional clients. We increasingly meet a lot of companies that are fed up with the vendor lock-in and forced upgrades imposed by traditional vendors and that are willing to listen to open source software options proposed by software “dwarves” like Opencell. By choosing  an open source model, startups can compete with the proprietary giants since your prospective client knows it will be able to take over the code if your company fails or gets taken over.   
  6. Open source gives you great marketing traction. A Gartner study shows that software companies spend an average of 10.2% of revenue on marketing expense and top spenders like Salesforce dedicate a whopping 53%. At Opencell, we spend very little. Our leads generally start with IT-savvy techies trying our product. 
  7. Open source lets you reduce your R&D costs significantly. Most open source startups are genuinely “… dwarves standing on the shoulder of giants”.  By being open source,  we leverage the software commons without needing to reinvent the wheel each time.
  8. Some things don’t change. Last but not least, great software companies, open source or not, will be judged on the quality of their features and roadmap decisions but also on the quality of the code and the documentation they deliver. We understood this all along… but we constantly remind ourselves that being open source doesn’t let you cut any corners.

In conclusion, we’ve been excited about the progress so far. We love the fact that open source gives us the ability to take on the big guys. We’re focused on our customers and hope that they’ll help make us successful.