How the Bell Curve Cheats You
By Michael Schrage

(FORTUNE Magazine) – Would you ever tell a lie about one of your best employees, a lie that threatened that worker's very career? No? But what if your boss asked you to? What if your boss absolutely insisted? Suppose your bonus depended on it?

Many of the world's best-managed companies don't hesitate to present their managers with this quandary. They even do it with a smile. Of course, they don't believe that they're asking anyone to lie; quite the contrary--they think they're pushing managers to be brutally honest. Nothing personal, just good business.

Hard-charging global giants are notorious for insisting that their people get rated and ranked against one another. Up or out. Survival of the fittest. The companies typically grade their folks on the strictest of curves; it's statistically impossible for everyone to be above the median, let alone in the top 10%. They passionately care about objectivity--and not just to avoid discrimination suits. They honestly believe they can draw a bright managerial line between "better" and "best." But can they? More important, do the organizational benefits of drawing that bright line outweigh the costs?

Lots of firms make it extraordinarily difficult for individual managers to declare more than, say, two or three of their best employees "outstanding" come evaluation time. The firms want to reward only the best of the best, the creme de la creme de la creme. Now, everyone can't be an "A" player; so to make absolutely sure, a few companies have made use of the cruel but useful twist of requiring managers to list an "F" player for every "A" they name. The underlying bell curve logic is impeccable if harsh. Then there are high-tech companies like IBM and Cypress Semiconductor, which have been famous for using computers to track and challenge those managers who think too highly of too many of their people too often. You might be a terrific manager, but there's no way that fully 20% of your people deserve "A" ratings. After all, "evaluation inflation" in organizations can corrupt corporate integrity and morale as surely as academic grade inflation undermines classroom integrity. Only 10% of your direct reports are in the top 10%. Deal with it.

These issues of integrity and morale cut both ways. After a McKinsey, General Electric, or ABB has culled the corporate herd for a few years and built strong teams in key business units, just how meaningful do the distinctions between "superb" and "outstanding" become? Beyond a certain point, diminishing returns set in. It becomes ever harder for top management to honestly distinguish one excellent performer from the next. When a manager has five crackerjack people working for her but is told that only three are eligible for the bonus pool, what criterion permits her to prefer one over the other? By any metric of honesty, that employee rating system requires the manager to be dishonest. She is being told to lie.

Organizations intent on rigorous self-improvement and its measurement inevitably confront an evaluation paradox: The more successful they are in developing excellent employees, the more trivial and inconsequential the reasons become for rewarding one over the other. Perversely, truly effective objective employee-evaluation criteria ultimately lead to personnel decisions that are fundamentally rooted in arbitrary and subjective criteria.

To make the problem even worse, such systems designed to reward top performers must inevitably alienate them. It's one thing to have reviews that demotivate average performers; to demotivate one's best performers is the height of self-destructive folly. The coup de grace occurs when the top employees are all told that they must collaborate better with one another even as they compete in this rigged game of managerial musical chairs.

These perils of performance-rating pathologies have not gone unrecognized. "We do not artificially limit the number of outstanding or exceptional employees," reports a GE spokesman. "Our appraisal system is flexible enough to ensure that truly outstanding employees are recognized appropriately." However, a number of GE managers privately acknowledge that, while the 340,000-employee giant has indeed become more flexible in recent years, corporate is relentless about making sure that "evaluation inflation" doesn't creep into the process.

The irony is that in their zeal to recognize and reward the best of their best, many of the world's leading firms are sowing the seeds of tomorrow's internal discord. "They are creating the very human capital market they are afraid of by encouraging a 'free agent' role of knowledge workers," asserts Tony Tjan, a co-founder of Zefyr, an aggressive young Internet firm that recruits from leading professional service firms.

The problem is not that life is unfair or even that the best-intentioned employee-review system is unfair. The problem is that a system explicitly designed to encourage honest and fair evaluations of hard-working people inherently requires dishonest and unfair evaluations by management. That kind of deviation shouldn't be allowed to become a standard.

MICHAEL SCHRAGE is co-director of the MIT Media Lab's e-markets initiative and author of Serious Play. Reach him at