Given the popularity of online communities, it's always useful to hear from those experienced at building them talk about their failures and the lessons they learned from them. One of my favorites is the "Lessons Learned" interview in which Mark Pincus, CEO of Zynga, talks about why his first social network startup, Tribe, failed to be a succes for investors. His lesson: Fail fast and test often.

Craig Sherman shares very similar lessons from both his days at Gaia Online and Ancestry.com. Craig was COO of Ancestry.com before he became an Entrepreneru-in-Residence at Benchmark and eventually CEO of Gaia Online in 2006. He's since left the teen-focused-virtual-world site.

But before he left, I managed to catch up with him and ask him about some of the lessons he'd learned from building online communities. 

At Gaia Online, one game took 2-1/2 years to build, he said.

“Looking back, we should have built that on one specific part of the site. We should have done it in two or three months and gotten data from our users really quickly and pivoted then, if it worked or didn't work," said Craig. “investing two or three years in something – big mistake – you should be able to figure it out in a couple of months."

What's the trick to creating engagement? I asked.

Craig says that listening to the customer or user helps a company learn what product or service or feature will be the most engaging. For example, Craig joined Ancestry.com when it was losing $5 million a month and had half a dozen product lines. He spent six months speaking to about 400 to 500 users either online or on the phone. By the time he had been done conducting his market research, he had realized that the key to Ancestry.com (which at the time was called MyFamily) was enabling users to find at least one ancestor. To that end, resources were stacked up against helping users find at least one ancestor.