Building marketplaces from scratch is tough, and labor marketplaces are especially hard. All the same issues exist with convincing both supply (sellers of labor) and demand (buyers of labor) to come and play, but none of the tricks from other types of marketplaces apply. In goods marketplaces, for instance, the marketplace owner can fake demand by buying stuff from sellers herself, and then selling them on eBay to recoup costs, or hustling to fulfill buyer’s request, faking as a seller, to retain that buyer and spark word of mouth. The dynamics are different when dealing with people and their services.

Those doing it successfully, like oDesk or Elance, took a long time to get there. Others that have achieved early liquidity, as defined by Simon Rothman in this article, such as CustomMade or Scripted, had their roots in other services that gave them initial traction in attracting sellers. CustomMade started a paid listing service for cabinet makers, and Scripted as Hollywood screenwriting Web-based service (previously Scripped). Both leveraged their respective initial businesses to spool up supply, gained momentum with buyers, and became credible marketplaces in their categories.

Those getting in the game now will have to figure out a way to hack liquidity or become another casualty of the proverbial chicken and egg problem. However, the casualties will leave behind valuable lessons for the rest of us. A recent discussion on Hacker News offered up some insights about Workfu, the job and talent marketplace that announced on Twitter at the beginning of August it would be shutting down. But per the announcement on their website and fervent user requests, the company still remains in play. This company’s experience offers the freshest set of insights into sequencing the key actions required to build a talent marketplace today.

1. Monetize early.

At their core, marketplaces are transactional businesses. They benefit from the flow in the exchange of goods or services for cash. The core function of discovery, i.e. matching buyers and sellers, is in itself hard to do and of value to the one entity (like oDesk charging 10 percent fees to the buyer) or both (the Ladders charging buyers and sellers). Accordingly, the business models that power talent marketplaces are generally straightforward – pay to post (Monster, StackOverflow, Dribble), pay per click (Indeed) or pay a percentage of the transaction (oDesk, Elance).

In Workfu’s case, launching swanky profiles, which attracted a loyal following of talent that wanted to look good, garnered them considerable traction. Additional features, such as matching applicants with jobs, along with the “FU Score” (not sure if pun was intended), were also offered for free. At no point did the site monetize the interactions of sellers and buyers. Early lack of a monetization model denied Workfu the insight of what users were willing to pay for and how much.

2. First, get sellers. Second, third, fourth, and fifth, get buyers.

Simon Rothman gave this useful tip last May at a Greylock marketplace gathering. And as Workfu showed, getting early adoption is an important first step. After signing up an initial set of sellers on your platform, the hard work of getting buyers begins. From that point on, all resources must focus on getting more buyers for your sellers, or in this context, more jobs for your talent. With jobs, you can attract more talent, with talent more jobs, and voilà! a profit engine is born. Another way is to go out and get sellers first, a la Zaarly, but it is unclear that this approach works without massive PR.

3. Do not run out of cash.

It took oDesk — the most successful work marketplace on the Web with revenues the size of Elance, Guru, Freelancer, and others combined — over two years to get to $1 million in total transaction volume. Now it grosses $78 million per quarter. This type of growth curve is typical for labor marketplaces. It is not clear what happened to Workfu, but running out of cash is a potential culprit. If you embark on creating a marketplace, be ready to be revenue poor for a couple of years. Raise a good round, lower your living expenses, or both, because it will take a while before you can live off what you kill.

Once a marketplace has traction, it tends to be very profitable and almost impossible to kill. (For instance, how many Craigslist killers have been started to date?) Such rich dynamics compel startups to try to up-end incumbents with novel approaches. However, marketplace creation is a long, hard slog, and anyone who gets in the game will need the cash and stamina to stick with it.

[Illustration by Hallie Bateman]