I picked up on a post by Elizabeth Dunn about decision trees for which type of social network - large network, nice audience network, or roll your own. Kevin Gamble has more describing a limited test of two efforts - one on Facebook and one on elgg.
This is absolutely the right approach -- start with small test before investing further.
Kevin used the white label social networking solution Elgg (an open source sonet with facebook like functionality) to set up a community.
So we embarked on getting some people to help us kick-the-tires on Elgg. We "invited" our advisory committees. People we knew well and that we had easy access. And we invited, and we invited, and we invited. We couldn't get even 50% of our advisory committee members to even try it. They never once logged-in. Mind you, they didn't even have to create an account. All they had to do was login using their existing identities. We also tried to work with some of our select, and mostly progressive communities to give it a try with zero uptake.
Meanwhile, here's what happened with his group on Facebook:
With almost no effort at all our groups exploded. We saw growth that we never could have imagined. I can't begin to tell you have easy it was to create functioning and engaged communities. It was so successful that I have considered closing down some of our old ways of reaching these people. I don't think we need them anymore.
Building your own communities is hard work. It might be next to impossible. White label or big box? Our experience has been that you'll have far more success going to where the people are already assembled.