The game has changed. Traditional marketing for web 2.0 means viral features and a huge dependency on word-of-mouth. Meaning, you better build a kick ass product that is unique enough to make the first page of digg.com. You most likely won't get there on your first attempt so you better be ready to iterate.
With the dramatic explosion of web services, and an unleashing of creative efforts surely there will be huge successes and monumental failures. Giant bets are made every day on build it and they will come. The dream is herds of web surfers relentlessly punishing the servers with request after request, traffic builds so quickly that you have to stay up all night racking servers to keep up with demand, the network is saturated, and your nightmare is server too busy messages. Your systems architect smile widens as you seamlessly plug in server after server and the system scales. This is what we call a good problem in the web 2.0 space.
The problem with all the services launching everyday is rising above the noise. Most of us will make bets on what we think users want. The best and most agile web 2.0 companies will make several bets. The key to crushing competition and putting other companies so far in the review mirror is creating an architecture that allows for rapid releases, extending the platform, and iteration at a speed that you blow past your competitors as they are trying to figure out how to scale the backend and unjoin those last few tables...
The key is architecture. It starts at the beginning of development, and is built by engineers that are master at systems, extensibility, and understand that most features are bets that require flexibility because few get it right the first time. Some features will never get used and a few will make you great. You want to tune and tweak existing features, not redesign. You want to extend an existing platform, not redesign. You want to scale databases, not redesign. You want to scale web servers and application servers, not redesign.
I believe that several companies will crush their competitors before a single pixel of UI is displayed. These companies will have the site architecture that allows them to innovate at a speed their competitors will never reach and continue to innovate quickly as traffic grows. Rapid innovation, many failures and few successes will be the working plan of record for the most successful web 2.0 companies. Each tweak and new feature is a result of a great architecture. This allows for doing more of what is working and cutting losses on stuff that doesn't. The more time spent innovating vs scaling is directly proportionate to the distance of your competitor in your review mirror.
Get a great system architect is what you want to play in the big leagues of web 2.0.