Cyanamid’s New Take on Performance Appraisal

In one form or another, performance appraisals have been with us more than 50 years. Few would quarrel with the notion of linking pay to performance, thus rewarding good work and (the employer hopes) giving mediocre performers an incentive or a goad to do better next year. If those who fail to perform quit, the organization benefits. Still, performance appraisal systems remain endlessly controversial. For one thing, many (if not most) achievers are motivated at least as much by pride or the desire to excel as they are by the hope of better pay. Critics also claim that poor ratings tend to demotivate inferior performers rather than to spur their improvement. Moreover, these poor performers often don’t quit; many hang on for years and years. Then there’s the sometimes thorny question of how to measure performance. Finally, many managers regard with distaste the obligatory, one-to-one meetings with subordinates to talk about results versus expectations. At American Cyanamid Company, a performance appraisal system called “progress reviews” had been in operation for more than a decade when, in 1984, the company decided to take a hard look at it. The reviews were part of the merit-based salary administration procedure. The evaluation grew out of an effort launched by CEO George J. Sella, Jr. to change the company’s culture. In a survey, its executives had used these words to describe Cyanamid: conservative, bureaucratic, and not sufficiently people oriented. While giving it high ratings for integrity and dedication to research and development, they had wished it had “a looser structure,” stirred “more creativity,” and exhibited more “willingness to take risks.” In response, Sella set $ 1 million aside for an “innovation fund” to support projects that might improve the quality of life for the company’s employees. A project that got under way at the same time was a different approach to the much-criticized performance appraisal scheme. Undertaken in the Medical Research Division, it was adjudged a success and a similar plan has been adopted for all 11,500 salaried Cyanamid employees in the United States. In effect companywide only since 1986, the new tack is still on trial. But there’s no question of its success among the knowledge workers in the Medical Research Division. This alone may have great significance in view of the growing numbers and importance in organizations of workers who make their living mainly by interpreting information. (About one out of eight people in the labor force is now in that category, and their proportion is growing fast.) Satisfying the desire for recognition and advancement of these professional or quasi-professional employees often conflicts with the preference of the organization’s bureaucracy for uniform and easily administered procedures, like the performance appraisal. Acute Dissatisfaction At Cyanamid, under the old system, salaried people whose performance was judged to be best received an O (for “outstanding”), which entitled them to the highest raises. Most others received one or the other of two rankings: X (“excellent in several major areas”), or R (“achieves expected results”). Approximately 20 % of the employees in a given unit were supposed to be rated O; 40 % , X; and 40 % , R. Subdividing these categories were plus and minus qualifiers, for example, X+ and X– on either side of X. There was also a fourth rating, N (for “needs improvement”), rarely given. Thus there were ten possible ratings. Division administrators of the plan customarily awarded ratings more or less in keeping with the 20–40–40 guideline. This also affected salary increases, which were set according to a range in each category. The biggest raises went to employees who won O ratings but who stood low in their salary ranges. The fixed distribution of rewards tended to flatten out the pay raises of those who had earned the same ratings for a period of time. Scientists in Cyanamid’s Medical Research Division, asked for their reactions to the system, offered acid comments, including these: Encouraging results from surveys of the two groups “The progress reviews do illustrate some aspects of how my superior rates me, but I know damned well that it’s also influenced by quotas.” “My work could be outstanding year in and year out, but if one other person in my group was outstanding, I would be denied such a rating.” “Until it’s my turn, I can’t get a better rating no matter what I do.” “I don’t take it seriously. It’s a Mickey Mouse numbers game. My superior and I just laugh at it, and we get through it as quickly as we can.” “Evidently there is an effort to fit the people within a section to a bell curve so they can dole out raises and not go over the budget. But as a scientist I know that a bell curve should be applied only when a population is large enough; and these departments are very small.” Among Cyanamid managers, the procedure had advocates and detractors. The advocates held that although it suffered from many of the usual flaws inherent in performance appraisal, its central purpose—to reward and encourage excellence—was simply too important to be sacrificed. Others thought that too often rewards seemed correlated with education and length of service. And further, that the system created apathy and a sense of futility among employees receiving lower ratings whose performance, if measured against their job descriptions rather than against each other, was entirely satisfactory. To the critics, the worst aspect was the demotivating effect of an R rating, which was generally regarded as a stigma. Cyanamid’s personnel department often heard complaints from managers who felt uncomfortable handling the progress reviews. Supervisors were required to identify and discuss aspects of an employee’s performance that needed improvement. They dreaded this task. They often felt forced to dwell on trivial shortcomings that ordinarily would be overlooked or on habits that were unlikely to change. And managers shrank from the duty of comparing an employee’s performance with that of others in the same department. Many complained of having to face the resentment of employees who got Rs. Clearly, the company was ready for a change. Divisional Trial The Medical Research Division carried out the experiment in two of its sections, each employing roughly 70 people, over a two-year period. These sections were chosen because they were the most comparable in the division with respect to size, mission, job characteristics, demographics (especially education and length of employment), and salary distribution. Both sections do experiments and laboratory analysis; both include employees in about the same proportions at the bachelor’s, master’s, and Ph.D. levels, as well as managers at these levels. The first step was to establish a baseline of attitudes and perceptions for both groups by means of a questionnaire. This survey found that the two groups were comparable in their almost uniform dislike of the progress reviews, particularly in their antipathy toward the predetermined distribution of ratings. One section now became a control group and the other an experimental group. While the control group continued under the old system, the experimental group was introduced to a new plan that had these features: It had only three possible performance ratings. It assumed that most employees, most of the time, perform somewhere within a broad range of acceptability, and that while fine distinctions within this range are possible (like above or below normal expectations), making them is neither necessary administratively nor desirable motivationally. This broad range of expected performance levels was designated G (for “good”). It assumed that occasionally, but not inevitably, someone’s contribution so greatly surpasses the usual range of performance that no reasonable observer would include it within the normal expectations for the particular function. This is not simply a matter of improving on the performance of one’s peers but of exceeding the demands of the job itself. Such performance is often a matter of both luck and pluck, of not only being in the right place at the right time but also of seizing opportunities and exploiting them creatively. This category of performance was called E (for “exceptional”). Finally, it assumed that performance is sometimes unacceptable and that unless an employee makes a prompt, pronounced, and lasting improvement, he or she should no longer be considered qualified to hold the job. This is not a mere matter of being the poorest performer in one’s group but of failing to fill the requirements of the job. This category of performance was designated U (for “unacceptable”). It called for no recommended distribution. Since the E and U categories had narrow, stringent definitions, few (or even, occasionally, no) employees would get either rating. By far the greatest number would ordinarily be rated G. The distribution depended on how many employees’ work fit the three descriptions. The system simplified the salary procedure. All employees rated G and E were awarded raises based only on their positions in the salary range. The Es also received lump-sum bonuses equal to fixed percentages of their salaries. Those rated U received no raises. In the revised progress review interviews, supervisors were expected to compliment subordinates on their strengths and accomplishments, and refer to shortcomings only if they were serious (as in the case of the Us) or were clearly within the employees’ power to change through improved effort or attention. The purpose was to make the meeting an ego-supporting instead of ego-damaging experience. The theory was that workers who felt good about themselves would perform better, or at least view their circumstances in a more favorable light, than those who had to defend themselves against what they might see as attacks. The new system simplified the managers’ task. Determining ratings was easier and, more important, the communication of the rating decision was easier. No longer were some employees stigmatized when their work was actually acceptable. But the program also had its costs. The better performers among the Gs (and for that matter the worse performers) were lumped with the others in that big group. Moreover, employees who excelled were not necessarily rewarded differently from their peers. Division officers feared that these features would cause those whose work had been superior to lose their drive to excel. One purpose of the experiment was to determine if this would occur. Happy Results Following expectations, the vast majority of the members of the experimental group scored G in the appraisals. The distribution of performance ratings in two years was this: In measuring the reactions of the two groups over one- and two-year intervals, the division used the same survey as the one that established baseline attitudes toward the progress reviews. The results showed a much more favorable attitude developing toward the experimental procedure. Supervisors noted that their subordinates were complaining less. The responses of the experimental group to a series of statements, four of which are shown in the insert, indicate a growing perception of the performance review system as being less arbitrary. This contrasts sharply with the control group’s attitude. Can the responses of the experimental group be attributed to a “Hawthorne effect”—the brief surge in favorable attitudes that often follows changes in long-standing procedures? We doubt it. Such an attitude would not last two years. The experimental group became more favorably disposed later in the trial period. Nor is it likely that the negative trend in the control group reflected envy of the experimental group, since the existence of the study was not widely known outside the experimental group. These results imply that views about compensation among comparably paid groups of professional workers depend less on salary levels than on the perceived fairness of the system for determining pay. They may also be a reaction to a simpler, more easily understood way of assigning salary increases. A happy benefit for managers was a reduction in time spent preparing for the interviews with subordinates: an average of about six hours each under the old review system compared with about three hours for the experimental group’s supervisors. One reason, no doubt, was the change from sessions full of negative comments to anticipation of more pleasant meetings. Whereas, for instance, some 80 % of subordinates were reported by their superior in 1984 to be critical of the matching of performance ratings to the recommended distribution, the experimental group’s supervisors reported that 85 % of their subordinates expressed satisfaction about the absence of such distributions. After two years of exposure to the new system, the most favorable views toward it came from the Ph.D.s—despite the advantages they had enjoyed under the previous system. This reinforced their claim that they were motivated more by professional pride than by money. One of them wrote in response to a survey question: “For me, the work ethic, my professional pride, and drive are stronger motivating factors than salary increases. But a poor salary increase could be a demotivator.” The Medical Research Division groups were small and unrepresentative of the company as a whole, so it is necessary to interpret the results cautiously. But they are internally consistent. They imply that, at least with highly educated workers, performance appraisal systems that are designed to support rather than deflate egos bolster motivation and promote acceptance of a salary administration system, at no greater financial cost whatever. Cyanamid Buys It Without waiting for the end of the experiment in the Medical Research Division, and in a sense anticipating its results, Cyanamid introduced in 1986 new progress review systems for all its salaried workers domestically. The new systems have these characteristics: There is no recommended distribution. There are three possible ratings, S (“superior”), Q (“quality”), and N (“needs improvement”). Specific objectives are emphasized, with sign-off by both employee and supervisor on annual performance objectives. The employee is involved in the performance planning process, particularly in setting up a training and development plan. The experiment in the medical group confirmed management’s judgment in designing those features. But in fairness it must be said that the company would have made these moves even if the trial had not confirmed their worth. Cyanamid’s experience with its new system has been largely positive, but of course it is still new. Early reactions indicate it is considerably more effective and helpful for supervisors and subordinates than the previous system. Some personnel managers have expressed a preference for further differentiation within the broad Q rating; but for the present, top management has decided to keep the system intact while acquiring more experience with it. So far, available evidence indicates that it is broadly applicable to salaried workers of all kinds, not just scientists, and that it represents an improvement over the previous system for all such groups. The new performance appraisal system has not been without its critics, either in its trial mode or now. Some thought that the Medical Research Division program, however popular with employees and supervisors, was overly egalitarian and would ultimately discourage the most capable people from striving to excel. Others feared that the lack of emphasis on employee shortcomings during progress review interviews would gradually entrench sloppy or inefficient habits. The experiment furnished no evidence to support these fears; but if they would eventually prove to be true, they would have to be reckoned as costs to be weighed against the benefits of the system. At first glance, the different method of administering salaries may seem to have broken the link between pay and performance (and to have done so without damage to the organization): within the broad group of Gs—87 % or more of the total, as we noted—raises were based not on performance differences but on the position of one’s salary within a range. The traditional system, however, may actually have broken that link more than the new one has, by rating workers’ performance against each other rather than against the expectations implied in their job descriptions, and by making finer distinctions between acceptably performing people than can be justified. The experiment recognized that in companies that hire selectively, promote on the basis of superior performance, and reassign or dismiss on the basis of inadequate performance, the distribution of performance does not follow a bell curve but is skewed disproportionately toward superior performance. Therefore, expectations must be similarly skewed; presumably, most people will turn in results that fall within a broad range of acceptability, and only a few will either exceed or come up short of that range. By eliminating the implied stigma of inferiority that inevitably follows when most others are rated superior, the new system has restored the perception of a link between pay and performance for those whose performance had in fact been quite acceptable. It has not created a new stigma for those formerly rated as superior because they too preferred being rated against the expectations for their jobs rather than against other workers in jobs not necessarily similar to theirs. Two main lessons have emerged thus far from the Cyanamid experience. First, motivation is more important than administrative convenience. Second, professional pride can be a powerful motivator and should not be discounted in considering how changes in performance appraisal and compensation systems will affect the productivity of knowledge workers.

Cyanamid’s New Take on Performance Appraisal

Partner Center