What open source-based startups can learn from Confluent’s success story

It’s common these days to launch an enterprise startup based on an open source project, often where one the founders was deeply involved in creating it. The beauty of this approach is that if the project begins to gain traction, you have the top of the sales funnel ready and waiting with potential customers when you move to commercialize your business.

In the past, this often meant providing help desk-style services for companies who appreciated what the open source software could do but wanted to have the so-called “throat to choke” if something went wrong. Another way that these companies have made money has been creating an on-prem version with certain enterprise features, particularly around scale or security, the kind of thing that large operations need as table stakes before using a particular product. Today, customers typically can install on-prem or in their cloud of choice.

“A key aspect of these kinds of technology-developer data products is they have to have a combination of bottom-up adoption and top-down SaaS, and you actually have to get both of those things working well to succeed.” Jay Kreps

In recent years, the model has shifted to building a SaaS product, where the startup builds a solution that handles all the back-end management and creates something that most companies can adopt without all of the fuss associated with installing yourself or trying to figure out how to use the raw open source.

One company that has flirted with these monetization approaches is Confluent, the streaming data company built on top of the open source Apache Kafka project. The founding team had helped build Kafka inside LinkedIn to move massive amounts of user data in real time. They open sourced the tool in 2011, and CEO and co-founder Jay Kreps helped launch the company in 2014.

It’s worth noting that Confluent raised $450 million as a private company with a final private valuation in April of $4.5 billion before going public in June. Today, it has a market cap of over $22 billion, not bad for less than six months as a public company.

Last month at TC Sessions: SaaS, I spoke to Kreps about how he built his open source business and the steps he took along the way to monetize his ideas. There’s certainly a lot of takeaways for open source-based startups launching today.

Going upmarket

Kreps said that when they launched the company in 2014, there were a bunch of enterprise-size companies already using the open source product, and they needed to figure out how to take the interest they had been seeing in Kafka and convert that into something that the fledgling startup could begin to make money on.

“There have been different paths for different companies in this space, and I think it’s actually very dependent on the type of product [as to] what makes sense. For us, one of the things we understood early on was that we would have to be wherever our customers had data,” Kreps said.

In order to be able to connect to that data, which often lived in complex legacy environments, they had to be able to install the software wherever it made sense to the customer. In 2014, that usually meant on-prem.

“In a sense, the more legacy, the more disparate the systems are, the more environments, the more lines of business, the more complexity — the more important that connectivity layer becomes. And the harder it is to do it without real infrastructure that helps. So as we started talking to these companies, we realized, you know, this is a huge opportunity that we hadn’t even really realized,” Kreps explained.

They also recognized that a lot of these customers weren’t even necessarily tech companies and certainly didn’t have large engineering teams that could deal with all of the complexity associated with downloading and supporting the open source product as early adopters had done. They had to provide something for them that was more packaged.

“So we started with a software offering that was sold as a subscription to companies that services — and still services today for Confluent — on-premise environments,” he said.

Turning to SaaS

Kreps said that as the company developed over the next several years, he realized that many companies couldn’t deal with the on-prem installation overhead and they needed to develop a SaaS version of Confluent.

“Obviously, [some customers] need much more of a packaged product that they can take and be successful with like a cloud service … that’s easy to use, that can help them just get started [because] they can’t hire 100 Silicon Valley engineers to try and do it internally,” he said.

The shift to the cloud required capital and engineers skilled in building a cloud solution around what Confluent was doing. But Confluent was a small startup, and this new approach presented some resource challenges. Compare that to Microsoft, which was shifting to the cloud around the same time. As Microsoft’s Jared Spataro told us in an interview at the same event, the company was a large enough company to build cloud products in parallel with on-prem and boxed software without batting an eye:

“The context for Microsoft had been our ability to develop multiple, very large businesses that ran in parallel. So this idea of like, we had multiple billion dollar plus businesses [like the] Windows business and Office businesses … even a server business associated with productivity.”

This was not the case for Confluent. It was still a pretty early-stage startup with limited resources, and everyone was advising Kreps to keep his focus on the on-prem side of things. “I would go around to different gatherings of CEOs, many of them more experienced [than me]. And of course, one thing everybody says is keep focus. Don’t try and do a second thing,” he said.

Kreps understood that building a second product was not what you’re supposed to do, but he felt it was imperative to try, even if it meant dividing his limited engineering resources. “The challenge for us was that we had a software offering with very large customers with lots of demands, and we had to [build] a cloud offering across all the different clouds while still servicing that [existing] customer base. Growing the existing business and building something new are both pretty hard problems, so that was the big challenge for us.”

The bet is paying off

A recent report by Battery Ventures suggests they made the right decision as cloud infrastructure vendors like (and including) Confluent had SaaS businesses that were growing substantially faster than their on-prem business. For Confluent, that difference was 46% growth for on-prem business versus 180% growth for the SaaS side of the house.  Of the five companies presented on the chart, only GitLab’s cloud business grew faster. You could argue that they would have been in trouble if they hadn’t made that move.

Cloud growth blows away other revenue growth according to research from Battery Ventures.

Image Credits: Battery Ventures

Part of the reason the company experienced this kind of SaaS growth was a decision in 2019 to open up a free tier, which much like the open source version earlier in the company’s history, provided a new way to get people interested in the product.

“A key aspect of these kinds of technology-developer data products is they have to have a combination of bottom-up adoption and top-down SaaS, and you actually have to get both of those things working well to succeed.”

Kreps says the free tier also forced the company to support this lower-level user in the interest of one day turning some percentage of them into paying customers.

“One of the challenging things for us was we were effectively adding [the free tier] later [in our company evolution]. Whereas most companies start with self-service, then layer on light sales and eventually go enterprise, we were moving almost downmarket.” He said in some ways it was going against the grain because the company had to switch focus from bringing in revenue to focusing more on user volume.

In the end, Confluent’s approach of growing from open source to commercial product has worked out well for them, and today they are a successful public company. Startups taking a similar route can take heart that the company defied conventional wisdom and still ended up thriving. As Kreps pointed out though, each situation is different and you have to figure out the right strategy for your company.