How does getapp.com or g2crowd.com succeed in terms of customer retention?

I'm the CTO of https://3dagogo.com a marketplace of proven to print 3D designs. We look at the two sides differently. There's not a single customer. In our case you have designers and purchasers ( sometimes the same person can be both ). Cost and methods for acquiring designers are very different than those to attract purchasers. I would clearly separate the sides and come up with separate cost structures. In my opinion when you're looking at the marketplace from the purchaser perspective, the other side's acquisition costs can be seen as fixed marketing costs.

Your support software should cater to your needs, depending on how your business operates. Fiver uses Vanilla forum and Zendesk. Thumbtack uses Zendesk. Not sure about AirBNB, their help center seems to be custom. Depending on how well funded your are, I would recommend starting with a free plan with one of the help desk SaaS products, or even using open source ticketing platform. Then, as your needs grow and you need integration with your marketplace, there's no reason you can't scale and migrate.

There is no valuation until you sell something. An idea or a company is only worth what its sales are. Once you have your initials sales, sales strategy and forecasting length (ie 9 months from first customer lead to close) then you have a formula for valuation. Valuation for start-ups is generally 3.5 x last years sales model should be the growth factor. When you are looking for investors, you will want to have atleast 9-18 months of SALES, not just pipeline and they will be looking at 5x revenue for a 3-5 year payback.

Focus on the more difficult side of the marketplace. For instance, if you think it'll be easier to get suppliers, then focus first on getting buyers - always be working on your toughest problem (aka your biggest risk). You'll find some great blogging on Marketplace and Platform topics here http://platformed.info (read the ebook too!)

The best way to solve chicken and egg problems for marketplaces is to prove market need on each side independently first with a low-cost MVP-type test. Once you've proven the market on both sides with metrics it is much easier to leg in supply and demand with a strategic or enough funding to match a market on a local or niche level to ensure liquidity. For a deeper analysis, here is a post on medium that I wrote... http://bit.ly/1k2vYbY Also, feel free to schedule a call with me if you'd like to dig deeper.