(FORTUNE Magazine) – Steve Kerr is the "chief learning officer" of General Electric, a company known for its executive development programs. An expert in reward and measurement systems, Kerr has long focused on matching incentives to goals, a practice at which many corporations are surprisingly inept. Before moving to GE, he capped a long academic career as the dean of faculty at the University of Southern California's business school and consulted for companies such as Citicorp. At GE he has played a key role in designing the famous Work-Out program, which trains GE employees to prune bureaucracy and to think about work in new ways. He met with FORTUNE's Strat Sherman at GE's Crotonville school.

GE has been touting stretch targets for years, yet you say they can be destructive. What gives?

Most organizations don't have a clue about how to manage stretch goals. It's popular today for companies to ask their people to double sales or increase speed to market threefold. But then they don't provide their people with the knowledge, tools, and means to meet such ambitious goals. We all agree that generally you get more output by committing more input, but now corporate America seems to be trying to get more output just by demanding more output. Ask them to explain the incongruity, and they say, "We're smarter now. We're not going to give you more people, or money, or physical space; we're not going to give you more of any resource, so your solution is going to have to be to work smarter, get out of the box, and be creative."

So what happens?

To meet stretch targets, people use the only resource that's not constrained, which is their personal time. I think that's immoral. People are under tremendous stress. And that's what I'm seeing all around this country, people working evenings, working Saturdays, working Sundays to achieve these stretch targets. Americans, in fact, now work longer days with fewer vacations than people in almost any other developed country.

Therefore we have a moral obligation to try to give people the tools to meet tough goals. I think it's totally wrong if you don't give employees the tools to succeed, then punish them when they fail.

Why bother with stretch targets in the first place if they're going to harm workers?

Well, if done right, a stretch target, which basically is an extremely ambitious goal, gets your people to perform in ways they never imagined possible. It's a goal that, by definition, you don't know how to reach. You might, for instance, ask people to cut costs by half or reduce product-development time from years to months. Stretch targets are an artificial stimulant for finding ways to work more efficiently. They force you to think "out of the box."

There's plenty of evidence that if you don't find dramatically new ways of doing business, you're not going to be in business. And if you don't intrude artificially into what's going on, you probably won't come up with radical, out-of-the-box ideas. So clearly some intervention is needed.

What's the right approach?

No. 1, don't set goals that stress people crazily. No. 2, if you do set goals that stretch them or stress them crazily, don't punish failure. No. 3, if you're going to ask them to do what they have never done, give them whatever tools and help you can.

Most employees already feel overworked today. How do you persuade them to accept such ambitious goals?

When you first start out, I think you have to give people confidence, because the first reaction always is "You've got to be kidding." Generally speaking, I think it's necessary that the stretch target be seen as achievable. But you're walking a thin line. Because if you already know how to get there, it's not a stretch target. And if you set a goal that's way out of line, you become an object of ridicule or fear. You can offer people any amount of money to fly around the room, but they still won't be able to do it.

It's not the number per se, especially because it's a made-up number. It's rather the process you're trying to stimulate.You're trying to get people to think of fundamentally better ways of performing their work, to cut out unnecessary work.

So in a way, you need to sell people on the notion that we use only a small percentage of our creative energy, that we have an infinite capacity to speed things up or an infinite capacity to improve things. Once you do that, you don't need to waste time justifying every stretch target by proving that someone else, either in the company or elsewhere, has proven it can be done.

The late Norman Maier, a psychology professor at the University of Michigan, did some great research on problem solving. His conclusion was that groups stop too soon. He would give people targets, and they would bring him a solution, and he would say, I don't accept this solution. And invariably, the next solution they brought was markedly better than the first.

There's a story about Henry Kissinger. Apparently he asked for a report on some country, his assistants handed it in, and he wrote in the margin: "Is this the best you can do?" So they panicked. They went and did it over. And again, he wrote something similar in the margin. The third time, he wrote another note, and they responded, "Yes, this is the best we can do." Kissinger then said, "Now I'll read it." I think he should have read the earlier drafts, but you get the point.

But you can't really just pull a goal out of thin air. No, of course not. You've got to find a middle ground. If you set easy goals, people may meet them but probably won't exceed them by very much. If you set extremely hard goals, people lose faith in them.

Also, in this interdependent world of ours, you have to realize that when you set a stretch goal, it has implications for somebody else. Say I set a stretch goal for sales. You have been selling ten widgets a day, and now I want you to double it to 20. But you can't sell those extra widgets if manufacturing doesn't make them. So once we set that stretch goal, how many widgets should manufacturing build? If you don't make your numbers, you end up with that dreaded thing called inventory. It's hard not to get grouchy when manufacturing built stuff that didn't sell. That's a tough thing not to punish.

What's the solution? In this example, manufacturing and sales are going to have to communicate much more closely and adjust output as the year goes on.

Should everyone be given the same stretch goals?

No, not at all. The danger is that you can end up hurting your best people. A golden rule of every work system is: Don't hurt the high performers. The folks in your best-performing business units may already be stretching themselves to the limit.

What repercussions are there?

One issue is self-punishment. If you're truly setting stretch goals, by definition you can't have a high degree of success. For individuals who are high achievers, it's not their style to miss goals. So you end up making people who are winners feel as if they're losers.

So how do you keep people from burning out?

Rather than making them work harder, you have to show them how to think in new ways. We're looking for answers--I'm not sure we've found them yet. At GE we held an innovation fair to look for new ideas. One image: You give a team an orange and say that each person must handle the orange--you can throw it to each other, do anything you want--but the orange has to end up in the hands of the person who started with it. The group throws the orange back and forth, and you time it. The first time we did it, it took nine seconds. When asked to improve, they stood a little closer, threw it a little faster, and got it down to seven seconds. Then we said, "Many groups do this in less than a second, and it's possible to do it in less than half a second." In the third trial, the team did it in less than a second, by simply stacking up their hands. The guy with the orange dropped it, it went swoosh through everybody's hands, and he caught it at the bottom--that was it. It was a neat example of the power of a stretch goal.

Another thing that came up in the innovation fair was something called brainstorming. When someone brings up an idea, you do a kind of carpet bombing. Everybody has to say what they like about the idea. That's the support. Then they must bring up their concerns. But you cannot bring up a concern unless you also bring up your suggestions to deal with the concern. This technique takes the pressure off people and may help them think more creatively.

How do you know stretch goals are working? You measure progress in three ways. One, you compare what you're doing now with your own past performance; two, you look for meaningful progress toward the stretch goal; and three, you benchmark competitors and see if you're doing at least as well as they are.

And if it does work, what does the employee get out of it?

Money, for one thing. In a typical gain-sharing arrangement, you might split the incremental gain or savings sixty-forty or fifty-fifty. Say, if the people at a factory save $300,000, they get $150,000 and the company $150,000. Then there are the nonfinancial rewards. These may include increased job security and personal satisfaction.

What are the most important things for managers to remember about stretch goals?

I think you absolutely have to honor the don't-punish-failure concept; stretch targets become a disaster without that. Also, you have to provide the right tools. Finally, you have to understand that stretch goals cannot be targets that you absolutely have to reach. Because if you absolutely need the results, let's face it, you're going to have to punish failure to reach it. On the other hand, there should never be excuses for failure to reach real goals like quarterly earnings or sales targets. Stretch goals are supplemental to those basic goals. All these points may not be rocket science, but you would be surprised how often managers violate them.