Wise words against my views on open source sales and marketing

I received an email back from a good friend about my recent post on open source sales and marketing. He strongly disagrees with my view. I'm going to include some of his comments here, because I think they reveal the depth of his experience and accumulated wisdom. They also point to an optimal way to start an open source company.

He suggested some ways to build an open source business:

1. Fire everyone in sales.
   
   (Asay: This is difficult for a company like Alfresco, as we don't actually have anyone in sales besides an inside sales person, and he's kicking tail. I'd never consider firing someone who paid for himself in the first month on the job.)
   
   
2. Set revenue expectations for the next 12 months to zero.
   
   (Asay: I think this is very wise counsel, but it's hard for any venture-backed open source company to actually do. I don't know any VCs that are patient enough to wait the 5-10 years for a JBoss, Red Hat, or MySQL to develop. I speak from some experience, having tried to raise money to build a business around Apache Cocoon. There have been a few VCs who had the foresight to invest early in these projects, but few to none that did so before there was serious community traction. My friend is essentially suggesting that the community should come before VC cash, and this is excellent counsel. It's just not practical for most would-be startups.)
   
   
3. Sales for the next 12 months should be handled only by the CEO and VP business development, and should be limited by their bandwidth to handle incoming inquiries.
   
   (Asay: Again, excellent counsel, and almost precisely what we're doing today. Back to his comment....)
   
   Steps 1-3 are designed to make sure you are focused on building the product that the market wants and will come to you for, and not about pushing software. (It's about pull!) The limited bandwidth of the CEO and VP Business development to handle sales inquiries will cause you to focus only on the highest likelihood accounts. i.e those you can close quickly.
   
   (Asay: Great advice. The difficulty for Alfresco has been that we have quickly been pulled into large, Fortune 500 accounts. It's a good problem, but I readily admit that it can be a problem. Larger deals require customer handholding. No one pays $100K without talking to someone, no matter how good the software is (and ours is exeptionally good). Ask MySQL, Red Hat, etc. if their experiences have been different. They haven't.)
   
   
4. For the next 12 months focus solely on building a product that is downloaded for free and that is installed and actively used. Your goal during this period is to drive up the installed base of active, useful internal installations of the product. It is NOT to close revenue and sign up customers. REVENUE IS THE WRONG METRIC TO TRACK DURING THIS INITIAL PERIOD. Tracking revenue too soon will cause you to push software sales too early, and you won't develop the pull model that will give you huge leverage 12 to 24 months out.
   
   (Asay: Good counsel, but see #2 and #3 above. It's hard to set revenue expectations to zero when customers want to pay for it today. Alfresco isn't alone in this. Mule, the leading open source ESB/EAI platform, is being used in big installations today, as is Xen (XenSource). I think it would be hard to tell Peter Levine (XenSource) and the Mule people to turn that cash down. :-)

My friend had other comments, but these are enough. Excellent counsel. They don't apply in all cases, but they're generally true. If you can afford to spend a year working for free, or know a VC that will fund you to take that time, you will do well with the counsel above. I believe, however, that SugarCRM has illustrated that a commercial open source community is possible, and can start first with venture money and an excellent product and drive near-term revenues.

Open source is a marathon, not a sprint, but I think it's a marathon that can be run at a brisk pace.