Looking at the open source software industry from the outside, it's often difficult to tell what is really going on. To use a string of clichés, it is hard to peel back the onion, to look behind the curtain, to perceive "Das Ding an sich" (German for the "thing-in-itself"; an idea made famous by the last Enlightenment philosopher, Immanuel Kant. http://en.wikipedia.org/wiki/Immanuel_Kant).

Part of this perceptual problem may be due to the good old fashioned human tendency toward denial. No one wants to exclaim at the dawn of a new era: "Here comes the new boss, just like the old boss."

But unfortunately, perhaps even tragically, any serious inquiry into the business models of the open source industry has to circle back to the same depressing answer: Open source business models don't look much different from the traditional software industry business models.

Conversion - Enterprise Server - Software Sales
There appears to be an evolutionary pattern in the development of a typical open source software company. First, it launches a GPL version of its product that causes an initial burst of interest. This phase of the development of an open source company is fun. Everyone is interested in the "new kid on the block," interviews are easy to get, and everyone is hungry for a glimpse of a fresh, new, and exciting idea. Venture capitalists are knocking at the door.

The second phase is also extremely exciting and full of romantic idealism. Against all odds and obstacles, the new company is able to produce a commercial version of the product and launch it through an online shop. Sales take off immediately and, through the miracle of the Internet, on a global basis. Life looks easy. "Money for nothing; chicks for free." Venture capital funding is secured.

Then harsh reality sets in during the third phase, the "rubber meets the road." Direct, viral sales through the Internet bump up against normal limits. Evil competitors pop up out of the weeds and spread malicious rumors. The initial customers run into technical problems. Because they have paid good money for the product, they dare to expect the company to fix them. Development becomes overwhelmed with support and with fighting fires. Most of the problems are related to things beyond their control. Something doesn't work in a Linux distribution. Someone changed something in a Linux distribution without telling them. Every customer has a different configuration and requires a tweak. Development of all those fresh new ideas grinds to a halt. The venture capital investors are not happy.

The next phase revolves around "passing the trash." One morning everyone wakes up and says, "Hey, this is not what we signed up for. This is no fun, getting yelled at every day by strangers. Let's offload the customer-facing and technical support issues to a partner network." So off the company goes and recruits the partner network. Recruiting a partner network is not hard; getting a partner network to produce consistent revenue, that is the tricky part. Initially, building the partner network is fun and easy. Revenue breaks through the natural limit of the Internet-direct channel, average deal size goes up, and everyone gets to fly around and have lots of meetings and expense-account meals. The venture capital investors start counting their money.

Then the "rubber meets the road again." The partners don't really know how the product works. They really can't provide technical support. They have problems training their sales force to sell it without overpromising. And they get a lot madder than customers because their livelihood is on the line.

The company enters the final evolutionary stage, direct sales. "Back to reality, whoops there goes gravity." Usually this phase is not entered into consciously. A big deal pops up and someone says, "We can't lose that one." The company decides not to pass on the lead to the partner network. Someone from development is recruited to serve as a temporary sales engineer on the account. The customer is happy. The deal closes and whoosh the clock is set back to IBM System 360 in 1966. The venture capitalists start planning an IPO.

In case you think that this scenario does not contain a grain of truth, let's take a look at MySQL AB, a Swedish company with U.S. operations in Cupertino, CA, close to their marquee venture capital investor, Benchmark Capital, in Menlo Park. According to Zack Urlocker, VP of marketing, MySQL will achieve revenue of $20M in 2005 and $35M in 2006. MySQL is a very successful open source project. It is ranks fourth on www.freshmeat.net out of 44,000 open source projects in terms of popularity. MySQL is downloaded 40,000 times per day - that's right per day - or 4,600,000 per year. Hard to believe but it's confirmed repeatedly by the company. Yet for all of that activity, a small amount of MySQL is generated through Internet direct sales through the MySQL online shop; my guess is that the figure is less than 3% - just a guess.

MySQL also has a well-developed network of 107 partners. The partner program is professionally run, with different levels based on precious metals and the obligatory pyramid (http://solutions.mysql.com/program/). (Whoever invented the precious metals analogy and partner program pyramid should have gotten a Nobel prize in economics, because everyone has copied it.) But for all the press releases and lists of partners, another data point indicates that the partner network is not able to carry MySQL as a channel. With only four persons allocated to the partner network, MySQL is hinting that "a vast and strong ecosystem of partners complementing its offering and relying on MySQL Network to build their businesses" is not bringing home the bacon. In Table 1, I made up the Sales per Employee number to support my theory about what is going on at MySQL - sorry for taking the liberty but it is unlikely that the company would release such a statistic.

Which brings us to the eureka moment, the epiphany, and Das Ding an sich. MySQL really earns a living through direct sales to enterprise customers - hey, we all have to earn a living and feed our kids or at least pay the electric bill to keep our computer running. My guess is that MySQL is achieving a respectable level of revenue per sales employee at more than $600,000 per annum. Though this level may lag behind the more established software companies like Oracle and IBM who aim for more than $1M per sales professional, direct sales is clearly the way forward for MySQL.

A quick look at MySQL's customer list further reveals that its customers are the usual suspects, I mean prospects (www.mysql.com/customers/).

My guess is that to make its $35M target in 2006, MySQL will need to hire more direct sales persons, so dust off your résumés.