Commercial Open Source: Can It Scale?
I recently came across a blog posting on InfoWorld by Savio Rodrigues of IBM that claims that open source (so called "pure open source" business models that give away a product and attempt to monetize it by selling services to some portion of the user base) does not scale; that in order for an open source company to move beyond $100 million it must move to the so-called Fedora model, in which access to certain products is unavailable except by payment. This refers to the fact that Red Hat insists that you sign up for its subscription if you want access to RHEL; if you're too cheap for that, you can use the free, unsupported Fedora.
Savio proves that pure open source doesn't scale by analyzing (in another blog posting) some stats from a blog posting by Matt Asay on CNet, in which Asay asserts that pre-acquisition JBoss identified 7000 viable leads per month, which it could pursue via low-cost internal direct sales. Savio calculates that JBoss must have been closing only 3% of its qualified leads, which is, in his words, incredibly low, since an inside sales rep should close 25% of leads. From this, Savio infers that open source business models don't scale beyond $100 million. He goes on to quote, approvingly, Marc Fleury (founder of JBoss) that the only true viable software model is the proprietary model.
What this strikes me as is a failure to really come to terms with what open source means to IT, and, by extension, the software industry. It is judging open source by the standards of proprietary software business models and finding it lacking, instead of thinking through what business opportunities are made possible by the new mode of software distribution.
First off, it is an unsupported assertion that there aren't enough customers to take an open source business beyond $100 million. Where is that written? Given that open source can be easily adopted by a far broader user base than proprietary software - made possible by cheap distribution and no financial barrier to user trial - why is it not possible that the potential market is far, far larger? His argument fails to fully comprehend the power of price elasticity, richly explored by Clayton Christensen; simply put, reduced prices don't mean less money is spent on a given item; reduced prices increase the customer base by more than enough to increase total market spend. Except for luxury goods, elasticity has been demonstrated in every good and service market invented by humans. Therefore, asserting that the potential sales opportunity for an open source company is limited to $100 million seems, to me, to be flat wrong.
Second, the assertion that a 3% close rate for direct sales is too low to run a successful business - again, where is that written? At a minimum, one has to understand what a viable lead is. I think the open source industry puts the bar for qualified pretty low - it can be someone who registers during download or signs up for a webinar. So, asserting that a 3% close rate reflects failure of the open source business model is, at best, unsupported. An open source lead base is a mix of real prospects and many "I have to register for the webinar, so I'll register" users. Segmenting the lead base will find a far richer vein of prospects, of whom a much larger percentage of leads can be closed.
More troubling about this assertion is the estimation that 3% is an inadequate response rate (response in the sense of ultimately becoming a purchaser). Generating a lead list by offering a webinar is analogous to creating a direct mail list - and in the direct mail business a 3% response rate is definitely high enough to build a successful business. Land's End, to cite one example, sells about $1 billion through its mail catalogs. I think most open source companies would be delighted to achieve Land's End results.
With regard to Red Hat's approach, one might think that Fedora is not their community edition, but that CentOS is. CentOS offers the same functionality of RHEL but lacks commercial support.
What is important to recognize about all of these open source businesses is that they follow a publishing model and sell subscriptions. And, like all subscription-based businesses, they scale slowly, but are, nevertheless, able to grow to very large scale and can be phenomenally profitable. In their heyday, newspapers achieved 20% net profit margins, very close to what the most successful proprietary software companies achieve. By contrast, licensed software companies can grow much faster, but tend to have problems later on when it becomes difficult to find new customers to fork over large upfront signing fees.
I think Rodrigues' plaint is more that open source companies seem to be puny compared to the most successful proprietary companies - from the perspective of the companies. If you evaluate open source companies by the same criteria traditionally used to judge proprietary companies, of course they seem like laggards. To my mind, however, this is somewhat akin to Charles I declaring democracy a poor replacement for divine right; from his perspective, democracy sucked, but for thevast majority of English, it was a far better governing system (bringing to mind Churchill's quote: "It has been said that democracy is the worst form of government except all the others that have been tried.")
The key to successfully scaling a software business (and this is true for both open source and proprietary software businesses) is to generate sufficient leads and to efficiently manage them well enough to create a viable business, defined as sufficiently profitable on a sufficient revenue base. An efficient sales process, applied to the very large lead base made possible by price elasticity, can certainly achieve success on even a 3% close rate. The question to my mind is "do open source companies have sufficiently inexpensive ways to sort through a torrent of leads torealize sufficient return on capital?"
By the way, if you're noting Rodrigues' argument explores the same territory as my recent "second chasm" newsletters, you're right. The key issue is not whether a 3% close rate is too small. The key issue is how open source companies can offer enough value to motivate users to become buyers or motivate buyers to spend more - in fact, addressing the latter issue could be far more productive in making commercial open source economics better.
Open source companies themselves may be wearing proprietary blinders; as I remarked before, they're selling insurance, so following the traditional per processor/per user model of proprietary software may be throttling theirpotentialrevenue streams.
Open source is moving through a transition period from proprietary licensed-bit coercion to something new - and what that something new will look like is not clear. A good analogy is the PC - at first derided as inadequate replacements for minicomputers ("toy" was often used to describe them), then adopted as less expensive ways to perform traditional tasks (preparing memos and creating financial analysis), and ultimately transforming the way we do business and live our lives. No one could have foreseen that most people would be buying portable computers, that they would be entertainment platforms, that a big issue would be getting enough shared storage into the home to make data accessible to the four or five dispersed throughout the home -- none of these things could be foreseen, yet they are logical outgrowths of personal computing.
We haven't even begun to see the potential for commercial open source (or for that matter, community open source). It will so transform the IT landscape that, in the not-too-distant future, we'll look back on the bit-coercive proprietary software worldand marvel that it existed at all - and that it was so small.
