The Washington PostDemocracy Dies in Darkness

With Steve Ballmer’s departure, a look at Microsoft’s flawed system of performance reviews

Placeholder while article actions load

In the aftermath of Steve Ballmer's announcement Friday that he intends to retire as CEO from Microsoft within the year, we've been reminded of the various reasons the software giant faltered under his leadership. He failed to move the company beyond PCs, some said. He missed the tablet revolution and other major technology trends. He had terrible timing.

And, as several have noted, he embraced a controversial type of performance review known as "stack ranking," in which employees and their teammates are forced into groups of top and bottom performers, distributed over a bell curve. In such a system, also known as "forced ranking," "forced distribution" or "rank-and-yank," the highest rated are typically promoted and given raises, while the lowest often don't get bonuses or are shown the door.

Kurt Eichenwald reported in Vanity Fair a year ago that the practice led employees at Microsoft to avoid working with top developers, not cooperate on projects, and remain focused on short-term achievements rather than long-term goals. (In a first-hand account published on Monday on, a former employee called it "as toxic for innovation and integrity and morale as media reports made it out to be, and then some.") The program was reportedly still around as of July--in a recent interview in the Seattle Times, Ballmer said there had only been "minor changes to the system."

But for all the attention Microsoft has gotten for stack ranking, it's actually still a common practice in other organizations. Forced ranking is often most identified with Jack Welch, who popularized it when he led General Electric in the 1990s, and roughly 60 percent of Fortune 500 companies still use the performance evaluation system or a form of it, according to the Wall Street Journal. While that may be less than it used to be--and many call it by more euphemistic names--30 percent of employers, according to a 2010 survey, specify some distribution of performance ratings that managers should use to separate their stars from their also-rans.

Why is there still adherence to this system? Is it really that good? Research seems to show otherwise. As far back as the 1960s, researchers found that bias plays a role when managers are forced to grade people along a curve. A 2005 study by researchers at Drake University and North Carolina State University discovered that while forced rankings can boost employee performance immediately after the system is introduced, the gains fall over time, with people more focused on competing with each other than collaborating. That makes sense, of course: After getting rid of the worst performers, those who remain are bound to be more competitive with each other.

Two years later, another study found that forced rankings discourage cooperation and hurt productivity. And more recently, research by Wharton's Iwan Barankay has found that evaluating employees in such a way can make the top performers complacent and the bottom performers uninspired. Another one of his studies, currently under review, found that removing feedback about an employee's rank actually increased the performance of furniture salespeople by 11 percent. No wonder only 23 percent of H.R. execs are happy with their performance management system--a number that's surely even lower for the rest of us.

So why do companies persist in such a questionable practice? Simple: It's easier to change the system than it is to change people's behavior. The reason many organizations began using stack rankings or distribution curves in the first place is that they believed managers were handing out too many As, avoiding tough discussions with their poor performers and failing to appropriately recognize their real stars. But rather than take the effort to train or teach managers how to make the hard decisions about poor performers, or do a better job of not hiring mediocre workers on the front end, companies put in arbitrary curves that could have a toxic impact in the long run.

Moreover, the downside of forced rankings can go beyond complacency and ugly office politics. In their strictest form, forced rankings don't allow managers to make their own calls about their own people. In other words, the system can prevent leaders from practicing--and improving--the most basic skill of management: their judgment.

Jena McGregor is a columnist for On Leadership.

Read also:

Everything that's wrong with performance reviews

The corporate kabuki of performance reviews

Like On Leadership? Follow us on Facebook and Twitter.