But be careful—even if you have a cofounder, there can only be one CEO. And if you pile on the cofounders, you’re asking for trouble. Having two founders works well. Three can work sometimes. I’ve never seen it work with more.

I remember one startup we worked with that had four cofounders. Every decision was made by consensus, which meant every decision took forever. They’d never started a company before, so even the most basic questions were endlessly debated—hiring, product changes, who to take money from, and how to structure the agreement. If they couldn’t agree they would hem and haw, trying to be nice, trying to be reasonable, watering down their opinions until the company fell behind the competition, ran out of money, and the board had to swoop in, remove some founders, and change the whole team around.

The Kauffman Foundation study also illustrated that almost 86 percent of companies that succeeded in the long term did not take VC money. Why? Because a company’s interests may not always align with the interests of its backers.

Launching isn’t a onetime, singular event, but a continual process of launch, measure, adjust, repeat. The cofounder of LinkedIn, Reid Hoffman, has said that if you aren’t embarrassed by the first version of your product, you’ve launched too late. It’s ridiculous to believe that every company grows out of a founder’s fully formed and unchanging idea, especially since most wildly successful companies achieved their place only by course-correcting, changing entirely, or iterating their way to greatness.

Experimentation creates tension. It carries a risk of failure. Moreover, when such experiments succeed, and companies innovate, people have to integrate change. The potential of failure and the need for change can terrify people. It can feel like the conflicts from their childhood that folks were programmed to avoid.

My Harvard colleague Thomas “Tom” Eisenmann, an entrepreneurship expert, finds that many start-up failures are caused by the skipping of basic homework. For example, Triangulate, an online dating start-up, rushed to launch fully functional offerings that didn’t fit any market needs. Eager to launch fast, founders skipped the research— customer interviews to probe for unmet needs. Giving short shrift to that crucial preparation, the company paid the price. Tom attributes this common failure, in part, to “the ‘fail fast’ mantra,” which overemphasizes action, shortchanging preparation. Moreover, while this might seem self-evident, once you’ve done the homework, you must heed what it’s telling you.