Ask the pros which comes first, marketing or building a product/service and of course they'll resoundingly answer "Both".
For those willing to listen, Steve Blank created an in depth series of posts on the Customer Development model. Paul Graham and the folks at YCombinator have preached early release as the best way to learn about user/customer needs.
As startups we're guided to get a minimum viable product released and iterate improvements fast, as a means of growing a business. The simple set of release features reduces the chance that bugs or design conflicts will creep into our product or service at an early stage. Business leaders strive to create a communication and action loop between product iterations and user/customer feedback. Now the two are not necessarily the same thanks to freemium business models, and their input should be handled differently.
In order to develop a customer base, we need to build something worthy of their attention. Our product must differentiate itself within competitive markets, or be so "full of win" that it creates a new industry around it. Examples of "full of win" breakthroughs are Google Adsense, the Internet, computers, automobiles, the wheel, and going way back the planet Earth (one helluva market ;)). Products like that take time and energy to create. As a startup we have to initiate this process with little or no capital (chicken or the egg).
Certainly the design decisions can be made by the leaders of a business, but being aware of to the true value points of a user community is the key to making hard resource decisions. The opposite end of design, emergence (I love this method), grants users maximum flexibility for customizing your product. As an example in web software, APIs and protocols allow for rapid extension of a service. By opening design control to a user base, your organization expresses respect and value to the end user. In return our users and customers help explore the vast value potential of our creations.