Microsoft's 'stack ranking' made employees compete against each other — and stifled.

For years, Microsoft used a performance-review system known as 'stack ranking' (or the 'vitality curve'): managers were required to rank their employees into predefined buckets — a fixed percentage at the top, a fixed percentage at the bottom — and compensation and survival were tied to the bucket. The bottom bucket faced being managed out, regardless of absolute performance.

The effect on culture was corrosive. Because the rankings were zero-sum, employees were pitted against one another. As Microsoft executives later acknowledged, 'superstars' avoided working alongside other top developers for fear of being hurt in the rankings, and employees were more concerned with making sure their peers or rival projects failed than with proposing new inventions. The company became, in one famous description, 'a collection of non-cooperating fiefdoms, unable to catch on to many technology trends.'

The practice drew intense internal criticism and, after Kurt Eichenwald's 2012 Vanity Fair article 'Microsoft's Lost Decade,' significant public scrutiny. In November 2013, following the announced departure of chief executive Steve Ballmer, Microsoft's HR chief Lisa Brummel announced the company was abandoning stack ranking. The episode became a cautionary tale about how a performance system that forces employees to compete against each other can destroy the collaboration and innovation a company needs most.