We developed a learning matching plattform that compares offerrings with needs. Since we approach a new market we do not have a DB with "offerrings" and no DB with "Needs". So the old known chicken&egg causality. Its not like Match.com, were two sides have the same need its more like "Car dealers" matching "Car buyers". Problem during ramp up: Buyers find no/less dealer and deales find no/less buyers!? Any prove strategie approach is highliy appreciated.
Find a channel that already carries one side and, find out if you can 'borrow' their content or access to people.
For example you can scrape yellow pages listings for car dealerships ( filling up the seller side ) and make posts of thier pages in your site on Craigslist (getting some buyers to know about you). Make sure your listings have top notch SEO built-in and don't worry too much about the little traffic that Google brings initially, it'll grow.
Make it super easy to share the one action that sets you apart with their friends.
All the while start marketing heavily to the sellers side. Do manual, non scalable work here, it doesn't matter at this point. Promise that you'd post to Craigslist for then, run Facebook ads of the pages in your marketplace, etc. Baby this new customers, learn from them, become their friend.
In my experience it's always easier to find 'brave' sellers willing to try something new. If your product is truly differentiated and you provide the necessary social tools and feedback loops, users will slowly find he site, have great experiences and start talking to about them with their friends and one day... You get critical mass and ball stars rolling on its own.
Good luck. Give me a call if you'd like to brainstorm more.