Microsoft was famed for its ability to iterate its way from knockoff products to market dominance. Everyone was too scared of Microsoft copying their products at top speed. The first and second versions of Microsoft Windows were unsuccessful attempts to copy Apple’s Macintosh operating system; the third version, while still inferior to its inspiration, was good enough, and Microsoft unleashed a marketing blitz for follow up versions such as Windows 95 and Windows NT that carried them to dominance. Microsoft later repeated this strategy with its Xbox business, which evolved from the Xbox, to the Xbox 360, to today’s Xbox One.

<aside> 💡 Mark Zuckerberg credits speed for the success of Facebook. In an interview for Reid’s Masters of Scale podcast, Mark told us, “Learn and go as quickly as you can. Even if not every single release is perfect, I think you’re going to end up doing better over a year or two than you would be if you just waited a year to get feedback on all of your ideas. That focus on learning quickly is the focus of the company

</aside>

The ideal is a tight OODA loop—observe, orient, decide, act—over and over again. Speed really matters, and launching early lets you climb the learning curve to a great product faster.

Iterate, Iterate, Iterate, Iterate.

I learned this lesson the hard way when I was running my first start-up, SocialNet. I didn’t want to be embarrassed by our first release, so the approach we took was to complete the entire product before we pulled back the curtain and let people sign up. This approach delayed SocialNet’s launch by a year, and when we finally did launch, we quickly realized that half of the features we’d painstakingly implemented weren’t important, and half of the important things that our service would be useless without were missing because we hadn’t thought of them. While there were other reasons why SocialNet failed, not launching early and iterating based on market feedback was probably the main cause of death.

Shortly before launch, we started worrying about whether LinkedIn would be useful without a critical mass of profiles. If a user logged in to LinkedIn, how could we make it useful even if none of that user’s friends had signed up yet? We decided that what was missing was a Contact Finder, a version of search that would let a LinkedIn user find potential vendors. For example, if you needed a consultant to help you figure out how to globalize your service, you could use Contact Finder to find Minna King. Our engineering team estimated that it would take us a month to build this feature.

<aside> 💡 We were presented with a difficult choice—delay the launch by a month, or launch without a feature that we thought might be essential to our success.

</aside>

Operating on the embarrassment principle, we launched without Contact Finder. And quickly we discovered a far bigger problem: Unlike users of personal social networks like Friendster, which were growing explosively as new users invited their friends to join, LinkedIn users weren’t sending any invites. Our user growth was stalled. Our baseline product was embarrassing because no one was using it! If we had delayed the launch a month to build Contact Finder, there still wouldn’t have been enough people hanging around to use it, meaning that we would have lost a month building a feature that didn’t address the core problem. We estimated that we would need at least one million users before search (and Contact Finder) would be useful, and solving that problem was the top priority.