(FORTUNE Magazine) – ''How am I doing?'' We ask ourselves that constantly. What seems odd is how seldom most of us -- the conspicuous exception being New York Mayor Ed Koch -- put the question to other people, even when our career progress depends on them. This may be changing a bit, however, at least among corporate managers. Nobody suggests that upward appraisal -- having subordinates formally evaluate the boss's performance -- is bursting like a bombshell above the business landscape. A few companies have done it for years. But lately the still thin ranks of corporations embracing the process appear to be growing. Professor John Bernardin of Florida Atlantic University surveyed 140 companies in 1984 and found that about 10% had tried it. When he did a follow-up last year, the figure was nearly 20%. Brian Davis, a senior vice president of Personnel Decisions Inc., a Minneapolis consulting firm, reports ''mushrooming'' demand for his company's Management Skills Profile, which combines self-evaluation with appraisals from boss, peers, and subordinates. You don't have to be a deep, revolutionary thinker to appreciate the reasons why. ''Who better to tell you what kind of manager you are than your subordinates,'' asks Donald Swan, who went through the drill a couple of times at Monsanto and now is instituting it as chief operating officer at Waldorf Corp., a paper products company. George Brennan, an internal consultant on organizational development at Massachusetts Mutual Life Insurance, which has begun such evaluations of its senior executives, sees other benefits: ''Going through this kind of exercise is a much more meaningful way to learn the importance of developing your people. It beats attending some three-day program off-site.'' If you don't believe it, compare your response to each of the following propositions: We'd like to talk to you about management skills. We'd like to talk to you about your management skills. ''People at the highest level in a company typically don't get much feedback,'' Brennan concludes. ''Criticism gets filtered out on the way to them. The biggest surprise I've had is how much our top management group has gotten into this. They're hungry for feedback.'' On a slightly less inspiring note, Babson College professor Neal Thornberry waggishly points out that upward appraisal ''takes the onus off the boss to be the bearer of bad news'' to a subordinate being reviewed. Let's just hear what the troops have to say. Even fans of the practice admit that it's tricky. Don't try it, for instance, in a primitively authoritarian organization, one being downsized, or anyplace with minimal communication up and down: It will only feed the general paranoia. Administered incorrectly, the process may leave subordinates open to reprisals from you know whom. And just what use do you make of the results? The highfalutin version of this dilemma: Do you treat the appraisal exclusively as a developmental exercise for the manager, or integrate the findings into his regular performance review? The more down-to-earth version: Do you tell his boss what they said? IF YOU REMAIN unfazed and want to go ahead with the process, know that upward appraisal comes in two models. The Deluxe entails bringing in an outsider -- someone from the human resources department, say, or a consultant -- and having him interview subordinates, boil down what they say, and play it back to the manager. Potential problems with this: Underlings may fear that the outsider will tell the boss who said what. If the person doing the interviews isn't well trained, he may pick up on comments in the first two or three and develop biases that cause him not to hear what people say in the rest of the interviews. The biggest drawback, though, is the cost -- not only the consultant's fee but the time of everyone involved. Which is why the Standard model is more popular. In it subordinates complete an anonymous questionnaire, rating the manager on different aspects of his % performance. The results are collected, tabulated, and communicated to the manager, typically in a written report. You still need an outsider -- to handle the paperwork and, if you have any sense, to go over the findings with the person reviewed -- but his role isn't as big, or as costly. Sometimes, if the Standard model uncovers a boss in deep kimchi, the Deluxe model should be brought in for a more searching look. If you choose the Standard model, try to customize the questionnaire for your organization, soliciting suggestions from raters and ratees alike. You will, of course, already have explained the purpose of the exercise and maybe even allowed everyone to debate the pros and cons. If you have decided to use it just as a developmental tool -- which, almost all the experts agree, is the way to begin -- you will stress that fact. Emphasize that nobody will be canned based on the findings, however disappointing this may be to some of the troops. Then draw up a list of what each manager is expected to do, a salutary and instructive exercise in itself, particularly if you have never before provided such guidelines. Don't be content here with mushy managerialese -- ''facilitates communications with subordinates on relevant matters.'' No, nail down the specific behavior, or, as the experts like to say, behaviors, required: ''Returns phone calls quickly,'' and ''Holds frequent meetings.'' This will help remind all concerned that what's being appraised is the boss's performance, not his personality or worth as a human being. REMEMBER, TOO, that subordinates may be the worst people to judge certain aspects of what their exalted leader does -- for example, his ability to cut payroll costs on orders from above. To the extent that you let him have a say in what goes on the list, you forestall him from dismissing the findings with a curt ''Oh, they looked at all the wrong things.'' Circulate the resulting questionnaire, asking subordinates to rate him on each aspect of his performance and to indicate how important each aspect is to them in doing their jobs. The process works best if you can survey at least four immediate subordinates per manager. Otherwise, you run the danger of what the experts euphemistically term reciprocal leniency: ''Rate me high,'' the boss insinuates, ''and I'll do the same for you on your next appraisal, hee hee.'' Even worse, with three or fewer, you might compromise the anonymity of the participants. Yes, in an ideal world, subordinates would feel free to raise their criticisms in everyday conversation with the boss, and he would welcome such contributions. A few organizations may even approach this ideal. But, say the experts, the vast majority of companies do not, and there can be hell to pay if a weak-sister manager divines precisely who gave him those terrible ratings. What not to do when the questionnaires have been scored and the results tabulated: Hand a written report to the boss with a snappy ''Well, here's how ya did.'' In the first place, as consultant Richard Dunnington points out -- he carried out IBM's employee surveys for years -- the raw data require a lot of interpretation. Even more crucial is how you break the news. It's here, in providing what the experts call structured feedback, that the consultant or person from h.r. really earns his pay. In going over the results with a manager, a true pro will make as much of the strengths uncovered as of the weaknesses: The more you give the reportee that he can agree with readily, the more likely he'll be to buy into the other stuff. When it comes to what the Personnel Decisions people label ''development needs,'' stress the fact that these are problems the manager can do something about, areas he can work on improving. Here it helps to suggest possible reforms, the more specific the better: ''These are three things you might do to give more guidance to subordinates when they're doing something new.'' Throughout the discussion, the manager should have the chance to say what he thinks and to raise questions. If he feels he's not being listened to, he's less likely to listen himself. What does the boss typically learn from an upward appraisal? The results are as various as bosses themselves, the experts say, but a few generalizations are possible. Notes Babson's Thornberry: ''I've yet to find a manager who didn't rate his people skills as one of his strengths.'' Alas, subordinates may have a different idea. Other experts observe that the boss often thinks he's doing a better job of communicating than he really is. Also that in these uncertain days, subordinates frequently wish he would give them more advice on what they should be doing to further their careers. WHATEVER THE FINDINGS, manager and consultant should then work together to devise a plan that will build on strengths and address, uh, development needs. It may be a good idea to have the manager share this plan with his boss, even if he doesn't share the appraisal results, to ensure that both agree about what he should be working on. A good manager will also let the troops know that he hears them. Ideally, says Personnel Decisions' Davis, he will discuss the findings with them in general terms, thank them for their trouble, and tell them what he plans to do to improve matters. Then he'd better do it. If he and other managers in the company who have had the experience follow through, the overall improvement can be dramatic, says William B. Werther Jr., a management professor at the University of Miami: ''It has a ratcheting effect. Say that across the organization, the average rating on communications is four on a scale of ten. Everybody below the average takes action. Next time you do the appraisals, the average moves up.'' Pretty soon you've got yourself one of those trendy, continually improving organizations. But woe betide the outfit where managers just sit on the appraisal results. ''You have raised expectations,'' notes Stewart Black, a professor at Dartmouth's Tuck School. ''By doing nothing you dash those expectations.'' Indeed, you may end up worse off. Says Black: ''Better for employees to think the boss doesn't care than for them to think that he betrayed them.''