I assume that by "increase signups," you mean that when people view the landing page, a higher percentage of them signup. But what would worry me is the stat that such testing does not capture: the number of people who did not share the site with others because it's not easy to play with. And that kind of sharing is advertising.
In other words, it's possible to have a lower signup percentage, but more total signups.
In other words, it's possible to have a lower signup percentage, but more total signups.