Google launches early and often in small beta tests before releasing new features widely.  >>> 

"There are two different types of programmers. Some like to code for months or even years, and hope they will have built the perfect product. That's castle building. Companies work this way, too. Apple is great at it. If you get it right and you've built just the perfect thing, you get this worldwide 'Wow!' The problem is, if you get it wrong, you get a thud, a thud in which you've spent, like, five years and 100 people on something the market doesn't want. Others prefer to have something working at the end of the day, something to refine and improve the next day. That's what we do: our 'launch early and often' strategy. The hardest part about indoctrinating people into our culture is when engineers show me a prototype and I'm like, 'Great, let's go!' They'll say, 'Oh, no, it's not ready. It's not up to Google standards. This doesn't look like a Google product yet.' They want to castle-build and do all these other features and make it all perfect. I tell them, 'The Googly thing is to launch it early on Google Labs and then iterate, learning what the market wants – and making it great.' The beauty of experimenting in this way is that you never get too far from what the market wants. The market pulls you back."

Don't use "I like" in meetings, push staffers to use metrics.

"When I meet people who run design at other organizations, they're always like, 'Design is one of the most political areas of the company. This designer likes green and that one likes purple, and whose design gets picked? The one who buddies up to the boss.'

Some companies think of design as an art. We think of design as a science. It doesn't matter who is the favorite or how much you like this aesthetic versus that aesthetic. It all comes down to data. Run a 1% test [on 1% of the audience] and whichever design does best against the user-happiness metrics over a two-week period is the one we launch. We have a very academic environment where we're looking at data all the time.

We probably have somewhere between 50 and 100 experiments running on live traffic, everything from the default number of results to underlined links to how big an arrow should be. We're trying all those different things."

Give people a vision, rules about how to get there and deadlines. 

Provide something simple to use and easy to love. The money will follow.

"I used to call this 'Users, Not Money.' We believe that if we focus on the users, the money will come. In a truly virtual business, if you're successful, you'll be working at something that's so necessary people will pay for it in subscription form. Or you'll have so many users that advertisers will pay to sponsor the site."

There's always a kernel of something good that can be salvaged.