Some years ago a human resources manager at a Silicon Valley computer company offered managers free tickets to San Francisco Giants games if they completed their subordinates' performance reviews on time. When David Russo headed up human resources for software maker SAS Institute, he earned employee cheers for a bonfire celebration that burned appraisal forms and ended annual reviews.
These two examples reflect a broader reality: Managers don't like giving appraisals, and employees don't like getting them. Perhaps they're not liked because both parties suspect what the evidence has proved for decades: Traditional performance appraisals don't work. But as my colleague and fellow Stanford professor Bob Sutton and I pointed out in our book, Hard Facts, Dangerous Half-Truths, and Total Nonsense: Profiting from Evidence-Based Management, belief and conventional wisdom often trump the facts. And when it comes to performance evaluations, companies ranging from HR consulting firms to providers of software that automate the process have a big stake in their continued use.
The most basic problem is that performance appraisals often don't accurately assess performance. More than two decades ago research done by professor David Schoorman showed that whether or not the supervisor had hired or inherited her employees was a better predictor of evaluation results than actual job performance. Employees hired by people doing the reviews got higher scores because of the greater psychological commitment managers have to the people they put themselves on the line to hire. That there is rater bias in performance reviews is consistent with the evidence showing gender and race effects on reviews. Similarity is an important basis of interpersonal attraction, and so people who are "different" get lower ratings, other things being equal.
Involve More People
When work is difficult to assess objectively, performance reviews mostly reflect how well employees can ingratiate themselves with the boss. One straightforward recommendation is to reduce managerial discretion in doing ratings. Make criteria more explicit and objective and have more people involved in the ratings process, so that one person's perceptions and biases don't matter so much.
A second issue is that reviews occur too infrequently to provide meaningful feedback. In return for getting rid of the appraisal form, Russo told SAS managers to provide more regular, ongoing feedback through frequent conversations with their people. Once-a-year reviews suffer from short-term memory loss: Managers remember more recent events and forget things that happened longer ago. If you are serious about feedback and helping people improve, do it all the time.
Next problem: Those receiving the reviews invariably believe they are above average—and defensively resist being told that they aren't. This "above average" effect has been widely replicated in numerous studies considering everything from sense of humor to appearance. "Forced rankings" require half of the people be rated below average. And that poses a threat to employees' self-esteem. As a result, people discount the ratings, making performance appraisals unlikely to improve performance.
Forget Colleague Comparisons
A fourth hurdle to productive reviews is the peer comparisons often required. Ranking someone against their colleagues creates competition and, consequently, a reluctance to offer help or collaborate—a big problem when so much of the way we work is interdependent. As lots of research in educational settings shows, the best assessments compare people with their own past performance. The assessments ask whether people are getting better or worse, not forcing comparisons with others that can cause people to give up (because they can't hope to exceed their peers) or to coast (because they know they don't need to improve given the competition). The lesson? If you're going to do performance assessments, at least don't force comparisons among people on some curve.
Possibly the biggest issue, however, is that performance appraisals focus managers' attention on precisely the wrong thing: individual people. As W. Edwards Deming, the father of the quality movement, taught a long time ago, company performance often results more from variations in systems than from the individuals doing the work. One of the reasons Toyota Motor (TM) has been so successful for decades—even as leaders have come and gone and the automobile market has changed—is that the fundamentals of the Toyota management system, which emphasizes quality, continuous improvement, and standardized tasks, provide the advantage. By focusing on the presumed deficiencies or strengths of people, individual performance reviews divert attention from the important task of eliminating the systemic causes, such as inferior technology, behind poor performance.
Even as companies and employees complain about performance appraisals, they do them because "everyone else is," and because they believe in the importance of individuals in boosting company performance. It is time for management to focus more on facts and evidence and less on benchmarking and unexamined conventional wisdom.