Licensing is packaging

A couple of weeks ago, I had a conversation on X with Chef co-founder Adam Jacob about software packaging. I was explaining how going source-available had driven Keygen Cloud's 2x growth over the last year. Adam argued that the growth probably wasn't due to becoming source-available but rather the change in packaging, specifically around offering a freemium self-hosted variant of Keygen.

He suggested that staying closed-source would have yielded the same success. I disagreed. My lived-experience of running a closed-source software company for 7 years before going source-available let me see first-hand the key role that licensing plays in packaging.

Packaging is, as the name suggests, how a software product is "packaged" into a cohesive offering and presented to users. In other words, it's how a software business communicates its product's value and positions itself in the market — through brand, features, pricing, distribution, and yes — licensing.

All of these are levers that let you control where you sit in the market; at the end of the day, it's all about directing how customers perceive your product and ultimately how customers use your product.

In packaging, licensing has first-order effects, namely how it influences perception of your product — a source-available product is perceived differently than a closed-source product, and Open Source can be perceived differently than source-available. It also has second-order effects on how customers adopt your product.

This is why Adam's argument doesn't make any sense to me. Licensing is absolutely a part of your packaging because it directly affects the perception of your product. It plays a significant role in building customer trust and reducing friction in the buying process. And it has a non-zero effect on how customers adopt and use your product.

The key driver of growth for Keygen over the past year wasn't merely the addition of a freemium self-hosted option. According to customers, the real driver was the reduction in fear of vendor lock-in. This was achieved by offering a clear continuity path, made possible through our move to a source-available model. This friction is especially high when you have a bus-factor of 1.

Customers understood that if anything happens to Keygen — whether Keygen LLC goes under or the product takes a wrong turn — they had options. They could self-host Keygen, and even outright fork it for internal use, similar to what Oxide is doing with CockroachDB.

Continuity was the main driver for Keygen recently going Fair Source — relicensing from the ELv2 to the FCL — to curb the remaining friction by answering the burning question, "What happens to Keygen EE if Keygen LLC goes under?"

Fair Source addresses this by undergoing full delayed Open Source publication, even the self-hosted enterprise features, offering yet another clear continuity path — directly addressing the customer concerns around continuity of Keygen EE.

The value of options can't be understated. When you reduce enough friction, users go from evaluating you against the competition, to evaluating your product against your product.

I believe this move, like the last move, will be instrumental to Keygen's growth over the next year and beyond.

Startups fail all the time, so reducing friction around adoption is paramount, especially when it comes to business-critical software like Keygen. Offering a clear continuity path and ensuring longevity is a subset of packaging because it directly affects the perception of your product, and this is accomplished through licensing.

Closed source can't do that — it has no continuity path, by design.

Licensing is packaging, but it's also a distribution model. Over the past year, I've seen a 10x increase in overall adoption of Keygen. A source-available product, much like Open Source, has a much easier time saturating a market, regardless of incumbents.

Users read your code and learn from it, they self-host it for personal projects, get hooked, and then spread the word to peers, which trickles down to coworkers, and ultimately to bosses.

There's an obvious virality to it, not just through the network effect. You also get users creating content for your product: writing blog posts, talking about it on socials, building sample apps, contributing code, plugins, and even maintaining SDKs. All these second-order effects, driven by users for their own benefit, drive the growth flywheel.

It's a powerful distribution model, one that's perfect for Keygen and other businesses looking to drive grassroots growth.

Closed source can't do that — at least, not to the same extent.

But as David Cramer, founder of Sentry, has said — licensing isn't a business model. You can't just flip the Open Source switch without a plan and expect to succeed. You can't rely on charity. And you can't rely on people being good. You have to proactively protect yourself and — especially for a bootstrapped business like Keygen — you can't tolerate unnecessary risk.

Licensing is packaging, and there are a variety of paths to explore — Open Source, Fair Source, source-available, and closed-source. And while I love Open Source and believe it's fundamentally better than closed-source, it's not always the best path for every project.

Consider Elastic, who originally chose the Open Source path, but later hopped onto the source-available path to protect themselves from AWS. Yet after 3 years, they've returned to Open Source. I suspect they encountered friction under the ELv2, as I did, and are using licensing to adjust their packaging to reduce that friction.

The move reintroduces unnecessary risk, and I'm not convinced this will work out as they hope. History may repeat itself. We've seen it play out many times throughout the history of Open Source sustainability, even recently with Plausible's move to an Open Core model.

Open Source is a risky choice for a core product, even under AGPL. Again, you can't rely on people being good.

This is why I think Fair Source is such an innovative model — it offers the distribution benefits of Open Source without the risk, while eventually contributing to Open Source via DOSP. I believe it's a better model than source-available, and a much better model than closed-source. It's also a safer model than Open Source.

Licensing is a powerful lever in packaging, with first- and second-order effects shaping perception, adoption, and overall usage.

Let's stop proliferating closed-source. Instead, let's embrace Fair Source and, in doing so, proliferate more Open Source.

Until next time.