Three years ago, shortly after he became president and chief executive officer of Crystal Tissue Co. in Middletown, Ohio, William Seymour noticed some strange-looking figures in the company compensation budget. The wages paid to his 240 hourly employees were in line with what others in the lightweight specialty paper segment of the paper industry were paying, but some of the salaries paid to the company's 60 salaried employees seemed out of kilter. It took a closer examination of the company's salary-administration plan before Seymour discovered the source of the problem.

"First, there was no way to measure how our [white-collar] pay stood with our competitors. Second, we didn't even have a way of comparing salaries in one part of our operation to those in another part of our operation." In other words, Seymour says, "there was no plan." To a newcomer charged with the responsibility of turning around a foundering $30 million company, these were troubling revelations. How could Seymour attract and hold talented employees if he couldn't be sure he was paying them what they were worth? How could he control his human-resources budget if each department continued to set and administer its own pay policy? Seymour called in a consultant. Together they evaluated and ranked Crystal's various job categories, meshing the consultant's expertise with the sensibilities of Seymour and his officers. When they were finished, they had 16 salary levels and accompanying salary ranges with which to compare the existing pay system -- or lack of it.

"We found inequities," he says. "Some people turned out to be underpaid, and others turned out to be overpaid -- a few by as much as 35%. In general, we found we had been too liberal, rather than too stingy." Seymour blames the disparities on salaries that were set arbitrarily and then allowed to march upward -- without regard to job description or individual merit -- at the same 6% to 9% annual rate that hourly wages rose.

Since then, some salaried employees have seen what Seymour terms "dramatic" pay increases. Others have had their pay frozen. Overall, Crystal isn't saving much money, Seymour says, "but we're spending it more effectively. Now we're in a position to reward good performance and take action against lousy performance. [Management has] gained credibility. Employee morale is better, too. And it's all because we've finally got something that's defensible. It's not loose anymore. We've got structure."

For the first few years of a company's life, an entrepreneur can usually maintain some semblance of a salary structure on the back of an envelope. It is likely that the entrepreneur has hired each employee and set each salary. But when the responsibility for hiring and compensating people is farmed out to department heads, matters can get out of hand.

"Pretty soon, somebody who works for manager Joe is making $10,000 more than somebody who works for manager Pete," says Howard Coate, a compensation consultant with Hewitt Associates in Lincolnshire, Ill., the firm that came to Crystal's aid, "and if just one situation like that gets around, employees lose faith that they're being fairly compensated. That's why you need a tool to guide managers in setting appropriate pay levels for the jobs that report to them."

That tool is a salary-administration plan. It can be as simple as Crystal's 16 pay grades -- established by ranking selected jobs for which market prices were known, in conjunction with internal analysis of jobs for which there were no data. This job-evaluation technique is known as ranking-to-market. Or, the plan can be the result of more painstaking job-evaluation techniques that involve breaking down jobs into clusters of skills and duties -- known in consultant's lingo as compensable elements. One method, called a point-factor evaluation system, arrives at a ranking of jobs by awarding points to these compensable elements; other techniques skip the mathematics and use more qualitative comparisons to establish value. But no matter which technique is used, what results is a document that reflects the particular goals of each organization. It sets forth what various kinds of work are worth to that business and provides a framework to deal with a company's changing employment profile.

None of this was invented yesterday. But, traditionally, small and growing companies have been slow to discover the merits of a salary-administration plan. Consultants say executives frequently tell them to go away -- that the company is too busy developing products and establishing markets to spend time revamping its pay structure. But during that time, compensation inequities are often building. A company in a mature industry, such as Crystal, may not see these inequities for decades. But in a very aggressive market, where the competition for talent is fierce, they can show up in a year or two -- particularly in recessionary times, when margins are thin.

"Here's a classic example," says Jim Finkelstein, of the Baltimore office of Meidinger Inc., another consulting firm. "I know of a Fortune 500 giant that started very small, not so long ago. They tried to get by with the simplest of job classification systems, even though they were rapidly emerging and things were changing every day. The bottom line? Over five years, keeping the head count constant, their increase in payroll averaged 21% per year. That's outrageous. That's out of control.

"You've got to look at your spending patterns as you grow, and make sure you're getting the biggest bang for the buck."

For a 50-employee public relations firm in Washington, D.C., that meant moving money from one part of the personnel budget to another. The company's expenditures for temporary clerical help had escalated to $15,000 a year -- a figure the comptroller deemed excessive. Turn-over was obviously on the rise, but the comptroller didn't know why until she saw the company's new job-evaluation results. Clerical workers, she found, were underpaid. So, she took that $15,000, added another $1,000, and hiked salaries for each of the company's eight secretaries and file clerks by $2,000 a year. More than a year later, turnover no longer seems to be a problem. Costs have gone up, but to this company's way of thinking, paying for good full-time help that stays around is vastly preferable to pouring money down a temporary-help drain.

A major part of the time and effort that goes into the establishment of a salary-administration program is spent in the mere selection of which job-evaluation technique to use. That decision is primarily based on two criteria: whether all jobs within the company are to be evaluated, or just one collection of job categories; and what sorts of jobs those are.

"Ranking-to-market fits best where the jobs are highly market-driven and cover broad responsibilities," says Coate. "I'm thinking mainly of executive-type positions. But this approach has problems on lower levels, where there are unique combinations of duties and responsibilities. You can't find survey data for them, and you end up making very subjective judgments of where they should fit in the structure." The Crystal Tissue Co. was able to use ranking-to-market because only salaried positions were being evaluated.

"A point-factor system " Coate continues, "is more popular in traditional manufacturing industries where many of the jobs are technical in nature. To them, just the aura of the numbers brings some face validity to the results."

A more qualitative job-evaluation system is usually deemed best for the youngest, smallest, and most entrepreneurial organizations -- companies in which most employees wear many hats. Instead of trying to fit those jobs into a market-based hierarchy, or rating them on the point value of the varied skills and duties within each job, most consultants recommend that the company develop its own job values and definitions.

To see how these job-evaluation techniques differ, consider the work of a secretary. The ranking-to-market approach would involve deciding whether secretarial duties in a particular company are worth more or less than what the average secretary does for the average salary elsewhere in the market. In a point-factor evaluation, the duties might be described as "answering telephones and preparing standardized reports and correspondence using a typewriter," and would likely be worth several hundred points. But in some of the more qualitative methods -- which are labeled in various ways by various consulting firms -- the job could be categorized, in descending order of importance, as "secretarial science," "secretarial procedures," or "clerical procedures." It is then up to the company to determine whether, say, "secretarial science" is worth as much as an accounting position.

"You have to be careful not to rely too heavily on market data," warns Meidinger's Finkelstein. "In markets where you're fighting to get and keep the best people, you may tend to bid up the prices if you pay too much attention to what other companies are paying. You may also hold your salaries down by perpetuating any inequity that may exist in the market."

But you can also skew your results from the inside. In identifying and weighting the compensable elements of a job, there is the possibility of giving unfair advantage to some workers. For example, too much emphasis on manual dextenty favors women. Too much emphasis on physical strength favors men. It may not always be right to place academic education above vocational training; and overseeing a function is sometimes a tougher job than supervising people.

Both consultants and their clients say many of these potential pitfalls can be avoided by including employees in the job-evaluation process and making sure that they understand why the development of a salary-administration plan is necessary.

"The best advice I can give is [for you] to spell out ahead of time what you're going to do with the job evaluation when you're done," says James Burns, administrator of Bartlett Memorial Hospital in Juneau, Alaska. The municipally owned facility recently did a total redesign of its compensation system and found that some employees -- particularly nurses -- were underpaid by as much as 20%, while the salaries of some department heads were as much as 36% too high.

"You have to decide whether to raise the low salaries right away or over a period of time," Burns says. "You also have to decide whether to lower or just freeze the ones that are too high. My advice is, get those answers early, and make them public. We didn't warn our employees, and we've lived to regret it." Burns is a hero with underpaid staffers who got their adjustment in one fell swoop, but to those whose salaries have been frozen, well, he isn't winning any popularity contests.

By all accounts, the development of a salary-administration plan is tedious, tiring work -- but worthwhile. Each company that takes on the task rapidly discovers that there is no such thing as an objective job-evaluation, just ones that are a little less subjective. Personnel costs are seldom cut, but are usually controlled.

"The real payoff, however, is that we've determined what the work we do is worth," says Crystal's Seymour, "and we've got a fair structure that's flexible enough to take us into the foreseeable future."