“We were throwing things against the wall,” says Justin Kan, founder of Justin.tv, Twitch, Exec, Socialcam, and now a Y Combinator partner. (He’s also a power-user on Snapchat, where he’s putting out some great stuff for startup founders.)

“We thought that instead of talking to [customers] and finding out what they liked and didn’t like about our existing service, we would continue to invent new features that we thought would be used even more than our existing features,” he said.

Only often, they were wrong.

For instance, they created a feature for Justin.tv called Justin.tv Challenges. The idea was that people could vote on what they wanted to see, like maybe a specific skateboarding trick.

“And the biggest ideas would get performed by someone live,” says Justin. “So we spent a couple months building this feature. It was the future of the company, Justin.tv Challenges. And we launched it. And it turns out that only one person ever completed a challenge.”

And that one person was Justin. Ouch.

“So I’ve lived the experience of building things that people didn’t want, and eventually finding out how to build something that people actually want,” he says. “The way that you actually build something that people love is by building incrementally.”

Justin says the process goes like this:

Step one: Gather data on customer problems.

Step two: Form a hypothesis about how to fix that problem and build something that you think addresses it.

Step three: Test your hypothesis by giving what you’ve built to your potential customers and getting feedback.

“Then from there, you figure out the problems, and go back to step one,” says Justin. “If you do that over and over and over again, if you complete that cycle over and over again, I don’t believe that you can be unsuccessful building a product.”

Here’s what else you’ll learn in his Mixergy course:

  • Got a wishlist of features for your product? Justin says you should burn it if you want people to buy your final product…
  • Why Justin grows his startups by focusing on a single most important metric, so that he can focus on growing it consistently week over week…
  • …and more!