Why My Former Employees Still Work for Me
From the January–February 1994 Issue
I own a manufacturing company in Brazil called Semco, about
which I can report the following curious fact: no one in the company
really knows how many people we employ. When we walk through our
manufacturing plants, we rarely even know who works for us. Some of the
people in the factory are full-time Semco employees; some work for us
part-time; some work for themselves and supply Semco with components or
services; some work for themselves under contract to outside companies
(even Semco’s competitors); and some of them work for each other. We
could decide to find out which is which and who is who, but for two good
reasons we never bother. First, the employment and contractual
relationships are so complex that describing them all would take too
much time and trouble. Second, we think it’s all useless information.
Semco has long been a laboratory for unusual employment and
management practices. What we’re now engaged in might be called a
radical experiment in unsupervised, in-house, company-supported
satellite production of goods and services for sale to Semco itself and
to other manufacturers by employees, part-time employees, ex-employees,
and people who have never had any connection with Semco whatsoever (but
who work on our premises and on our equipment). This is not at all the
same thing as outsourcing. This is a borderless system of short-term,
noncontractual task assignment often using Semco’s own fixed assets,
some of it in Semco plants and some dispersed at a dozen sites that
don’t belong to the company.
This satellite program, as we call it, sounds chaotic, can be frustrating, and is in some ways uncontrollable. It requires daily leaps of faith. It has serious implications for corporate culture. It has destroyed any semblance of corporate security. And, for the three years it has been in place, it seems to be working very well. Since 1990, 28% of Brazilian capital goods manufacturers have gone bankrupt. In 1990, 1991, and 1992, Brazilian gross industrial product fell by 14%, 11%, and 9%, respectively. Capital goods output has fallen back to what it was in 1977. But in this same period, Semco’s overall sales and profits have remained intact, and I attribute the difference first and foremost to our satellite production.
Ever since I took over the company 12 years ago, Semco has been unorthodox in a variety of ways. I believe in responsibility but not in pyramidal hierarchy. I think that strategic planning and vision are often barriers to success. I dispute the value of growth. I don’t think a company’s success can be measured in numbers, since numbers ignore what the end user really thinks of the product and what the people who produce it really think of the company. I question the supremacy of talent, too much of which is as bad as too little. I’m not sure I believe that control is either expedient or desirable.
I don’t govern Semco—I own the capital, not the company—but on taking
over from my father, I did try to reconstruct the company so that Semco
could govern itself on the basis of three values: employee
participation, profit sharing, and open information systems. We’ve
introduced idiosyncratic features like factory-floor flextime, self-set
salaries, a rotating CEO-ship, and, from top to bottom—from the owner to
the newest, greenest maintenance person—only three levels of hierarchy.
You might say that what we practice is an extreme form of common sense: “common” because there’s nothing we do that thousands of other people didn’t think of ages ago, “extreme” because we actually do it. Another way of looking at Semco is to say that we treat our employees like responsible adults. We never assume that they will take advantage of us or our rules (or our lack of rules); we always assume they will do their level best to achieve results beneficial to the company, the customer, their colleagues, and themselves. As I put it in an earlier article in HBR, participation gives people control of their work, profit sharing gives them a reason to do it better, information tells them what’s working and what isn’t.
With rare exceptions, this approach has been successful. We’ve had two or three strikes, but they were quickly settled, especially once the strikers saw that we would neither lock them out of the plant nor suspend their benefits during the work stoppage. (They were able to plan ongoing strike tactics while eating lunch in the company cafeteria.) We’ve had a few employees take wholesale advantage of our open stockrooms and trusting atmosphere, but we were lucky enough to find and prosecute them without putting in place a lot of insulting watchdog procedures for the nine out of ten who are honest. We’ve seen a few cases of greed when people set their own salaries too high. We’ve tried a few experiments that we later backed away from. We’ve had to accept occasional democratic decisions that management disliked, but we learned to swallow hard and live with them.
On the whole, as I say, our approach has worked. Loyalty is high, quality is excellent, and sales and profits are surprisingly good for a manufacturing company in one of the world’s most lunatic business environments. But in Brazil no state of the economy is permanent. Few last long enough to be called temporary. Surviving the ups and downs of the Brazilian economy is a little like riding a Brahma bull. It is even more like riding a Brahma bull in an earthquake. Some of the worst jolts come not from the bull but the landscape.
In 1990, the jolt that sent us into our present experiment came from the minister of finance, who, believing Brazil’s inflation was simply the result of too much money being used for too much speculation, seized 80% of the country’s cash and introduced an extended period of economic bedlam. Employers could not meet payrolls. Consumer spending vanished. Business spending shuddered to a halt. Bankruptcies soared. Industrial output plummeted.
At Semco, we had several months of zero sales. After all, what company was going to buy machinery with a ten-month delivery when it didn’t know if it could last out the week? Worse yet, back orders were canceled, or we found that our customers had gone out of business. Our marine division alone had $1.5 million of receivables that we couldn’t hope to collect and $4 million worth of products that shipbuilders could no longer pay for. We had to rent warehouse space just to store all the unsold goods.
We cut costs. We organized workers into teams and sent them out to sell replacement parts directly to ships and restaurants. We cut down on coffee breaks, locked up the copiers, canceled orders for new uniforms, turned off all the lights we could find, scrimped on telephone calls. None of it was enough, and anyway, I don’t really believe in cost-cutting. I like to think we don’t waste money even when we’ve got it. And who can say how many sales we lose when we play Scrooge with the travel money or penny-pinch the phone bills?
Finally, we called the workers together in groups of 100 and
discussed what we should do. They came up with lots of ideas, and we
tried them without success until we reached a point where no one had
anything else to propose and neither did we—except for two unhappy
alternatives: cut pay or cut workforce. We thought we could avoid
layoffs by cutting salaries 30%
across the board until business picked up again. But a lot of people
were already struggling with bills and rents and mortgages and wanted us
to start laying people off instead, so that those who stayed could at
least survive. We went on searching desperately for a third way out.
And then suddenly the shop-floor committee came to us and said, “Okay, we’ll take a 30% pay cut, but on three conditions.” The first was that we increase their profit sharing by 15%, from just under 24% to just under 39%, until they got back up to their former salary levels. The second was that management take a 40% pay cut. And the third was that a member of the union committee would co-sign every check we wrote; because the workers wanted to be absolutely certain that their sacrifice would be worthwhile, they wanted to oversee each and every expenditure.
Well, at that moment, we had no profits to share, so there
was nothing for us to lose and everything to gain. And by the second
month, we were actually covering expenses. In their drive to save, the
workers took on more and more of the former contract work. They did
security and cleaning, drove trucks, even cooked the food in the
cafeteria. No expense went unchallenged, and for four or five months, we
made a small profit in the worst economic times any of us had ever
seen.
But we kept on looking for a better solution.
In the first place, pure cost reduction has to be a temporary measure. What about training, research, new product development, and all the other seemingly peripheral activities that produce profits over the long haul? Those weren’t responsibilities we could abdicate.
And what about those checks? The dual-signature scheme was working for the moment, but management couldn’t permanently yield its power of the purse to a person chosen by the union without management input or approval.
Yet the explosion of energy, inventiveness, and flexibility we’d been witnessing was hugely attractive. And when we then added in several other factors—the need to cut our standing labor costs, the demands of Brazilian labor law, the dynamic example of our own peculiar Nucleus of Technological Innovation, which I’ll come back to in a moment—what began taking shape was a radically new principle of organization.
The three engineers suggested we call the new unit the Nucleus of Technological Innovation (NTI) and somewhat predictably proposed themselves as its first three members. We bought their odd idea, and then we worked out an odd form of compensation to go along with it. Their guaranteed salaries went sharply down, but they would now share in the proceeds of their inventions, innovations, and improvements. They would receive a percentage of any savings they introduced, royalties on new products they devised, a share of the profits on their inventions, and they would also be free to sell consulting services on the open market. They might have done even better as truly independent entrepreneurs, but as NTI members, Semco would cushion them against disaster and give them the support of an established and well-equipped manufacturing operation.
By the end of their first six months, NTI had 18 projects under way, and over the next few years they uncorked such an array of inventions, changes, and refinements (one of my favorites is a scale that weighs freight trains moving at full speed) that NTI’s members began to prosper mightily and Semco became unthinkable without their constant innovation and reform.
By 1990, we’d begun to feel that we’d like to NTI the entire company, liberate more creativity, tie compensation even more specifically to performance, loosen the ties that bound us all together, and scramble our overall structure. The 30% pay-cut-and-cost-reduction scheme had given us a breathing space of several months, but with the Brazilian economy in a bucket and no imminent prospect of recovery, we had to become permanently leaner and more flexible. At the same time, of course, we had a commitment to our workforce that was central to the way we did business. That commitment had been our principal reason for trying to avoid layoffs.
For most other companies, there was another reason as well: Brazilian labor law protects laid-off workers by granting them several different forms of special compensation. The largest of these comes from an individual fund for each worker to which the employer contributes 8% of wages every month. When people are fired (or retire), they collect all this accumulated money, plus interest, in the form of a lump sum. Less substantial—but, unlike the 8% fund, a great problem for many employers—is severance pay itself, which is paid on the spot out of current income and which can amount to two years’ salary in the case of workers with many years’ seniority. By the end of 1990, a lot of Brazilian companies drifted slowly into bankruptcy rather than lay people off and go bankrupt overnight. With our finances still more or less intact, this widespread problem proved in our case to be an opportunity.
Semco’s sales had gradually increased again, and we were making enough money to restore salaries to where they had been before the 30% cut. We took back our check-signing privileges. We were surviving in a crisis economy, but only just, and we began to face the fact that we had to cut our permanent staff and contract more of our work. We looked hard for a way of doing it without destroying the support system that Semco people lived on. It was here that NTI’s free-form structure suggested a solution.
Instead of giving contracts out to strangers, we decided we could just as well give contracts to our own employees. We would encourage them to leave the Semco payroll and start their own satellite enterprises, doing work, at least initially, for Semco. Like NTI, these satellites could stay under our larger umbrella by leasing our machines, even working in our plant. Like NTI, they could also do work for other companies, again on our machines and in our factories. Like NTI, their compensation would take a variety of forms—contract payment, royalties, commissions, profit sharing, piecework, whatever they could think up that we both could live with. And like NTI, they could have some beginning guarantees. In particular, we would offer all of them some contract work to cut their teeth on, and we would defer the lease payments on all equipment and space for two full years.
This satellite program would have obvious advantages for Semco. We
could reduce our payroll, cut inventory costs by spreading out raw
materials and spare parts among our new suppliers, and yet enjoy the
advantage of having subcontractors who knew our business and the
idiosyncrasies of our company and our customers. Moreover, we would pick
up the benefit of entrepreneurial motivation. Because of profit
sharing, our employees already worked evenings and weekends when
necessary, without any prompting from management. Being in business for
themselves ought to raise that sense of involvement higher still.
But what in heaven’s name were the advantages for our workers, who’d be giving up a secure nest at Semco for the risks of small business? And in the midst of economic bedlam? To begin with, of course, they all had the chance to make many times what they could earn at Semco—if the economy straightened out. Of course that was a big if. And should the recession persist, they might make less. But only assuming they continued to have a job at Semco, which was becoming an even bigger if with every day that passed. The fact was, they had distressingly few choices. And so did we.
We eased the transition in every way we could. We created a team of
executives to teach cost control, pricing, maintenance, inventory
management. To provide seed money, we gave people layoff payments on top
of severance pay and all the other legally required benefits. Many also
made use of their 8% nest-egg funds. No one had
to start a satellite. Some took their severance and left. Some managed
to stay on the payroll for months or for good. But despite the
difficulties, satellites sprang up quickly. White-collar workers were
the first. Our tax accountants, human resource staffers, and computer
programmers all went off on their own. Then blue-collar workers in food
service and refrigerations systems followed suit.
Today, about half the manufacturing we once did in-house has gone to satellites, and we think we can farm out another 10% to 20% in the coming years. Best of all, to this day only one satellite has failed. Some are expanding and looking for partners. Some satellite workers have been rehired by the company, and a few have moved repeatedly back and forth between satellite and employee status as needs—theirs and ours—shifted. Some satellites have broadened their scope so greatly that most of their time—often right on our premises, remember—is spent with customers and production partners who have no other connection with Semco whatsoever.
In 1990, Semco had about 500 employees. Today, we have about 200, plus at least that many in our satellites, with another 50 or 60 people who work for a satellite and also work for us part-time. We have employees with fixed salaries. We have employees with variable salaries made up of royalties or bonuses based on self-set objectives like cash flow, sales, profits, production units, or any one of a dozen other measures. We have employees with both fixed and variable salaries. All our employees share in our profits.
On the satellite side, compensation may take the form of a fixed fee, an hourly stipend, a percentage of increased sales, a finder’s fee, an honorarium, a retainer converting to an advance converting to a royalty, or even a simple win-or-lose commission. In one case, we had decided to kill a product-development project when one of our people picked it up from the table and said, “I’ll take it. If you’ll give me $1,000 a month, which will just pay my expenses, plus a 7% royalty for the first five years if I can make it work—I’ll take it.” So of course we gave it to him. The most we can lose is $1,000 a month, where we’d been spending $10,000 to $15,000 a month and getting nowhere. The most he can make is something like half a million, I’d guess, with the other 93% coming to Semco.
Once we posted a job for one engineer and got 1,430 résumés. We took them home in packs of a hundred, and then we interviewed for five months. In the end, we invited several dozen final candidates to a one-day seminar where we walked them through the entire company, opened our files, showed them everything we did, then asked them for proposals. We wound up hiring 41 engineers—one salaried employee and 40 satellite workers whom we paid on various forms of percentage-based commission.
In one of our plants, we’ve set aside a large room full of desks and
computers to give everyone within our company sphere and, for all we
know, a variety of guests and visitors from well beyond it, a place to
sit and plan and ask questions and solve problems. We call it the
Thinkodrome, and it’s a busy, quiet place. That Semco survives at all we
owe in large part to surrounding ourselves with people who look at
everything we do and ask why we can’t do it better or cheaper or faster
or in some entirely novel way.
Generally speaking, Semco’s production process works along similar lines, both for satellite operations and for the work we do in-house. All work, including some aspects of management, goes to people with proven track records who want the jobs and can compete for them successfully. Satellite as well as in-house business units rise and fall on their merits alone—at least in theory.
This commitment to free-market principles was put to the test about a year into the satellite program, when our marine division found itself with a good deal of idle capacity. Marine’s strategy was built on quality, not price—low volume but high margins. A shipbuilder seeking the very best performance and dependability in, say, a propeller system, tended to come to us. But with the economy in a straitjacket, orders had nearly disappeared.
On the other hand, our biscuit division, which designs and builds turnkey cookie factories for global giants like Nabisco and Nestlé, had two fairly big contracts in hand, one for about $2 million, another for $5.5 million, and was going to need a lot of skilled subcontracting. The portion of this work that marine could do would keep it occupied for four or five months, and marine’s top manager (called a counselor at Semco) went ahead and figured the contracts from the biscuit division into his budget. But biscuit’s purchasing people did not award the contracts to the marine division, which took too long to deliver, they said, and which charged too high a price for its exaggerated quality. They gave the contracts instead to satellite producers and outside contractors, including one of marine’s archcompetitors. The fight that triggered was a bitter one, and the attempts of the interdepartmental management meeting to act as go-between did not make things easier. We were of two minds ourselves.
On the one hand, we were going to pay marine employees to sit around on their hands while another division paid outsiders to do work the marine employees could have done. And on the other hand, how could we ask biscuit employees, who share in their division’s profits, to subsidize a business in trouble? Moreover, wouldn’t the subsidy just postpone marine’s inevitable reckoning with its own strategic predicament? In the end, we let biscuit have its way and endorsed the need to be as unforgiving toward our own business units as we would be toward outsiders. It was the right decision, of course. We finished the cookie factories on schedule, and the marine division—which decided to stick with its high-quality, high-margin strategy but to eliminate a number of products whose quality and cost were too high for the market—cut its staff by 70%, began farming out a lot of its work to satellites, and recovered its profitability.
The budget cycles are January to June and July to December, but the CEO cycles begin in March and September. In other words, we avoid what other companies and shareholders think they want—responsibility nailed down to a single man or woman. Our CEOs don’t wear themselves out trying to meet quarterly financial goals, and there’s no one person to blame if the company goes down the drain. When financial performance is one person’s problem, then everyone else can relax. In our system, no one can relax. You get to pass on the baton, but it comes back again two-and-a-half years later.
One consequence of this system is that we need to keep each other
well-informed, which we do at regular weekly divisional meetings and
biweekly interdivisional meetings. All these meetings are open and
optional, and those who attend make decisions that those who don’t may
simply have to live with.
This self-selecting element in decision making is another consequence of the deliberate fragmentation of responsibility. Like our predecessors the mammoth hunters, the people who get responsibility are the people who seek it out and meet it. In fact, the actual, ad hoc control structure we work with from day to day builds on this principle and on two others that, together, create a kind of invisible order from the apparent chaos that characterizes the Semco environment.
The first principle holds that information is the ultimate source of virtually all power. For this reason, we try to make all of it available to everyone. All meetings are open. Designs and specifications are shared. The company’s books are open for inspection by employees and for auditing by their unions. In short, we try to undercut and so eliminate the process of filtering and negotiating information that goes on in so many corporations. Meetings are chaired by the person who knows most about the subject under discussion rather than by the person who has the highest declared status or apparent income.
The second principle is that the responsibility for any task belongs to the person who claims it.
The third is that profit sharing for employees and success-oriented compensation for satellite enterprises will spread responsibility across the Semco map. With income and security at risk—and with information readily available—people try hard to stay aware of everyone else’s performance.
To give an idea of how all this works in practice, let’s take one of those turnkey cookie factories. The Big Cracker Company of Chicago wants a plant that will turn out a thousand tons a month of, say, butterscotch macaroons. To begin with, an independent agent will tell us of the project in return for a finder’s fee. We will probably put the initial customer interface in the hands of a satellite company—four men who used to be on our payroll and now work for themselves. They’ll go through the specifications with the customer, then they’ll share that information widely, announce a meeting (to which anyone can come and no one is summoned), and chair the discussion (which will cover several unexpected proposals from unanticipated participants like the guy from refrigeration with a special point of view about handling butter and coconut). Someone, a group of employees or satellites, will take on the job of costing the project, and with this cost estimate in hand, a Semco counselor and biscuit-division coordinators will then set a margin and deliver the quote to Big Cracker.
(A couple of times, we have even communicated this margin to the customer, because we thought it would be easier to justify, for example, a 12% net margin than to play the disingenuous game of claiming pencil-thin profits and no room for compromise. Our chief argument has been our profit-sharing program, since it seems so clear to us that people will work harder for more money and that a generous margin will therefore buy the customer much extra care and effort. But I’m afraid we’ve had only limited success with this approach.)
This margin-setting discussion often produces serious disagreements. In one case, we battled out the margin in a long, heated debate, and then the sales manager lowered it dramatically when he sat down with the customer. By Semco rules, that kind of last-minute capitulation is perfectly legitimate. Battle or no battle, he was with the customer, not we. Whoever holds the spear is completely in charge of bringing down the mammoth.
Let’s say Big Cracker accepts our bid and the order comes in, 600 pages long. Let’s also say we choose a coordinator for the project from outside the company, and let’s call him Bob. Bob will go through the contract and decide how he wants to divide it up. He may get help from engineering. He will certainly get help from all the meetings he holds to make decisions, which he will chair and where he will lobby for the people he wants to do each job. Next, the biscuit division’s purchasing department will negotiate contracts with the dozens of suppliers chosen. In 1991, we did about 70% of such a contract in-house. Today, that’s down to 35% or 40%.
When Bob has put together a completion schedule and time chart, everyone will go to work—each contractor, employee, and satellite responsible to no single authority but answerable to everyone. At most companies, when something goes wrong the real responsibility falls between the cracks. At Semco, the fact that Bob is not an employee makes everyone react much faster when there looks like trouble.
When we started, people warned me that all sorts of information about our company would get into the wrong hands, that we had to protect ourselves. I heard the same argument when we started distributing profit-and-loss statements to our employees. But it’s a waste of time to worry about leaks.
First of all, we no longer know whose the wrong hands are. The
competition used to be a company a mile away that made the same products
we did, but now the competition comes from companies we’ve never heard
of in Taiwan and Finland. Second, I’ve never seen a company overtake
another because it had seen its 10K or even the specifications for a
valve. Third, we want to be a moving target. We don’t care about
yesterday’s information or last year’s oil pump, which in any case the
competition can buy, take apart, and study to its heart’s content.
Finally, we don’t think people give out much information anyway. I know. I’ve tried on numerous occasions to get a copy of, let’s say, a pamphlet some company passed out to 1,000 employees, and nobody can lay their hands on one. The Chinese printed hundreds of millions of copies of Chairman Mao’s Little Red Book, and still they’re as rare as hen’s teeth.
People also warned me about the loss of central goal setting and control. I admit that the lack of control is often hard to live with. But let’s not compare Semco’s circumstances with some ideal world where managers actually get to decide what people will do and when and how they’ll do it. We have limited control over the day-to-day behavior of the people who make most of our components, but so do companies that do all their work in-house. At least none of our satellite people work nine to five and leave their problems at the plant when they go home at night—which means leaving them to management. We have motivation and responsibility working on our side. Our satellite workers are in business for themselves, so they’ll work all night to complete an order to specification and on time. And if the order is late or fails to meet our quality standards, then we’re free to give the next order to someone else. We can forget the witch-hunt and all the grief that goes into firing people or not promoting them.
As for planning and the control it presupposes, I think good planning is always situational. Thinking about the future is a useful, necessary exercise, but translating such conjecture into “Strategic Planning” is worse than useless. It’s an actual barrier to survival. Strategic planning leads us to make things happen that fly full in the face of reality and opportunity.
For example, Semco is today in the environmental consulting business, which I could not have imagined five years ago. Our gadfly NTI group was looking at one customer’s need for an environmentally active pump—a pump that would shred and process the material it moved—and saw that the company could reengineer its production line to do away with the pump altogether. Had we said, “We’re in the pump business, not the environmental business,” we might never have pursued that solution to the problem. As it was, we addressed the company’s overall need, jettisoned the pump, and when it was all over, we’d also acquired a small environmental consulting firm to flesh out our own limited expertise. More recently, we’ve also entered into a joint venture with one of the world’s leading environmental consulting groups. Today, the division represents about 14% to 15% of our total business and is growing at a rate of 30% to 40% per year.
The lesson this story teaches me is about the negative value of structure. Structure creates hierarchy, and hierarchy creates constraint. We have not utterly abandoned all control, but the old pyramidal hierarchy is simply unable to make leaps of insight, technology, and innovation. Within their own industries, pyramidal hierarchies can generate only incremental change.
Take dishwashers, one of Semco’s businesses. Dishwashers are expensive to operate and messy to use, but over the last 50 years, dishwashers have changed hardly at all. What the customer wants is a machine that washes dishes silently, cheaply, and without any mess at all, which probably means without water. I’ve recently seen indications that such a thing may be possible, but the idea could never have come from a pyramid of Semco dishwashing executives. It was one of our satellites that brought us the idea. In fact, about two-thirds of our new products come from satellite companies.
What goes for planning goes equally for culture, vision, and responsibility. We find that fragmentation is strength in all these areas. Semco has no corporate credo, for example, and no mission statement. An articulation of company values or vision is just a photograph of the company as it is, or wants to be, at one given moment. Snapshots of this kind seem to hold some companies together, but they are terribly static devices. No one can impose corporate consciousness from above. It moves and shifts with every day and every worker. Like planning, vision at its best is dynamic and dispersed.
At Semco, so is responsibility. We have little control, even less organization, and no conventional discipline at all. People come and go whenever they like; many set their own compensation; divisions and units perpetuate themselves however they can; satellite companies work on our machines in our factories for us and others in a great confusion of activity; the system tying it all together is painfully loose—and this is manufacturing, much of it assembly-line manufacturing.
When I describe Semco to other manufacturers, they laugh. “What do
you make,” they ask me, “beads?” And I say, “No, among other things, we
make rocket-fuel propellant mixers for satellites.” And they say,
“That’s not possible.” And I say, “Nevertheless…”
The point is simple but perhaps not obvious. Semco has abandoned a great many traditional business practices. Instead, we use minimal hierarchies, ad hoc structures, self-control, and the discipline of our own community marketplace of jobs and responsibilities to achieve high-quality, on-time performance. Does it make me feel that I have given up power and governance? You bet it does. But do I have more sleepless nights than the manufacturer who runs his business with an iron hand and whose employees leave their troubles in his lap every night? I think I probably sleep better. I know I sleep well.
We delivered our last cookie factory with all its 16,000 components right on time. One of our competitors, a company with tight controls and hierarchies, delivered a similar factory to the same client a year and two months late.
This satellite program, as we call it, sounds chaotic, can be frustrating, and is in some ways uncontrollable. It requires daily leaps of faith. It has serious implications for corporate culture. It has destroyed any semblance of corporate security. And, for the three years it has been in place, it seems to be working very well. Since 1990, 28% of Brazilian capital goods manufacturers have gone bankrupt. In 1990, 1991, and 1992, Brazilian gross industrial product fell by 14%, 11%, and 9%, respectively. Capital goods output has fallen back to what it was in 1977. But in this same period, Semco’s overall sales and profits have remained intact, and I attribute the difference first and foremost to our satellite production.
Ever since I took over the company 12 years ago, Semco has been unorthodox in a variety of ways. I believe in responsibility but not in pyramidal hierarchy. I think that strategic planning and vision are often barriers to success. I dispute the value of growth. I don’t think a company’s success can be measured in numbers, since numbers ignore what the end user really thinks of the product and what the people who produce it really think of the company. I question the supremacy of talent, too much of which is as bad as too little. I’m not sure I believe that control is either expedient or desirable.
I think that strategic planning and vision are often barriers to success.
You might say that what we practice is an extreme form of common sense: “common” because there’s nothing we do that thousands of other people didn’t think of ages ago, “extreme” because we actually do it. Another way of looking at Semco is to say that we treat our employees like responsible adults. We never assume that they will take advantage of us or our rules (or our lack of rules); we always assume they will do their level best to achieve results beneficial to the company, the customer, their colleagues, and themselves. As I put it in an earlier article in HBR, participation gives people control of their work, profit sharing gives them a reason to do it better, information tells them what’s working and what isn’t.
With rare exceptions, this approach has been successful. We’ve had two or three strikes, but they were quickly settled, especially once the strikers saw that we would neither lock them out of the plant nor suspend their benefits during the work stoppage. (They were able to plan ongoing strike tactics while eating lunch in the company cafeteria.) We’ve had a few employees take wholesale advantage of our open stockrooms and trusting atmosphere, but we were lucky enough to find and prosecute them without putting in place a lot of insulting watchdog procedures for the nine out of ten who are honest. We’ve seen a few cases of greed when people set their own salaries too high. We’ve tried a few experiments that we later backed away from. We’ve had to accept occasional democratic decisions that management disliked, but we learned to swallow hard and live with them.
On the whole, as I say, our approach has worked. Loyalty is high, quality is excellent, and sales and profits are surprisingly good for a manufacturing company in one of the world’s most lunatic business environments. But in Brazil no state of the economy is permanent. Few last long enough to be called temporary. Surviving the ups and downs of the Brazilian economy is a little like riding a Brahma bull. It is even more like riding a Brahma bull in an earthquake. Some of the worst jolts come not from the bull but the landscape.
In 1990, the jolt that sent us into our present experiment came from the minister of finance, who, believing Brazil’s inflation was simply the result of too much money being used for too much speculation, seized 80% of the country’s cash and introduced an extended period of economic bedlam. Employers could not meet payrolls. Consumer spending vanished. Business spending shuddered to a halt. Bankruptcies soared. Industrial output plummeted.
At Semco, we had several months of zero sales. After all, what company was going to buy machinery with a ten-month delivery when it didn’t know if it could last out the week? Worse yet, back orders were canceled, or we found that our customers had gone out of business. Our marine division alone had $1.5 million of receivables that we couldn’t hope to collect and $4 million worth of products that shipbuilders could no longer pay for. We had to rent warehouse space just to store all the unsold goods.
We cut costs. We organized workers into teams and sent them out to sell replacement parts directly to ships and restaurants. We cut down on coffee breaks, locked up the copiers, canceled orders for new uniforms, turned off all the lights we could find, scrimped on telephone calls. None of it was enough, and anyway, I don’t really believe in cost-cutting. I like to think we don’t waste money even when we’ve got it. And who can say how many sales we lose when we play Scrooge with the travel money or penny-pinch the phone bills?
Who can say how many sales we lose when we play Scrooge with travel money and phone calls?
And then suddenly the shop-floor committee came to us and said, “Okay, we’ll take a 30% pay cut, but on three conditions.” The first was that we increase their profit sharing by 15%, from just under 24% to just under 39%, until they got back up to their former salary levels. The second was that management take a 40% pay cut. And the third was that a member of the union committee would co-sign every check we wrote; because the workers wanted to be absolutely certain that their sacrifice would be worthwhile, they wanted to oversee each and every expenditure.
Lower pay and higher profit sharing helped us profit in the leanest times.
But we kept on looking for a better solution.
In the first place, pure cost reduction has to be a temporary measure. What about training, research, new product development, and all the other seemingly peripheral activities that produce profits over the long haul? Those weren’t responsibilities we could abdicate.
And what about those checks? The dual-signature scheme was working for the moment, but management couldn’t permanently yield its power of the purse to a person chosen by the union without management input or approval.
Yet the explosion of energy, inventiveness, and flexibility we’d been witnessing was hugely attractive. And when we then added in several other factors—the need to cut our standing labor costs, the demands of Brazilian labor law, the dynamic example of our own peculiar Nucleus of Technological Innovation, which I’ll come back to in a moment—what began taking shape was a radically new principle of organization.
The Thinkodrome at the Free-for-All Corporation
Years ago, in the mid-1980s, three Semco engineers proposed a new kind of work unit. They wanted to take a small group of people raised in Semco’s culture and familiar with its products and set them free. The new group would not have to worry about production problems, sales, inventory, equipment maintenance, delivery schedules, or personnel. Instead, they would invent new products, improve old ones, refine marketing strategies, uncover production inefficiencies, and dream up new lines of business. They would have no boss and no subordinates. They would pick their own focus, set their own agendas, and have complete freedom to change their minds. Twice a year they’d report to senior management, which would decide whether or not to keep them on for another six months.The three engineers suggested we call the new unit the Nucleus of Technological Innovation (NTI) and somewhat predictably proposed themselves as its first three members. We bought their odd idea, and then we worked out an odd form of compensation to go along with it. Their guaranteed salaries went sharply down, but they would now share in the proceeds of their inventions, innovations, and improvements. They would receive a percentage of any savings they introduced, royalties on new products they devised, a share of the profits on their inventions, and they would also be free to sell consulting services on the open market. They might have done even better as truly independent entrepreneurs, but as NTI members, Semco would cushion them against disaster and give them the support of an established and well-equipped manufacturing operation.
By the end of their first six months, NTI had 18 projects under way, and over the next few years they uncorked such an array of inventions, changes, and refinements (one of my favorites is a scale that weighs freight trains moving at full speed) that NTI’s members began to prosper mightily and Semco became unthinkable without their constant innovation and reform.
By 1990, we’d begun to feel that we’d like to NTI the entire company, liberate more creativity, tie compensation even more specifically to performance, loosen the ties that bound us all together, and scramble our overall structure. The 30% pay-cut-and-cost-reduction scheme had given us a breathing space of several months, but with the Brazilian economy in a bucket and no imminent prospect of recovery, we had to become permanently leaner and more flexible. At the same time, of course, we had a commitment to our workforce that was central to the way we did business. That commitment had been our principal reason for trying to avoid layoffs.
For most other companies, there was another reason as well: Brazilian labor law protects laid-off workers by granting them several different forms of special compensation. The largest of these comes from an individual fund for each worker to which the employer contributes 8% of wages every month. When people are fired (or retire), they collect all this accumulated money, plus interest, in the form of a lump sum. Less substantial—but, unlike the 8% fund, a great problem for many employers—is severance pay itself, which is paid on the spot out of current income and which can amount to two years’ salary in the case of workers with many years’ seniority. By the end of 1990, a lot of Brazilian companies drifted slowly into bankruptcy rather than lay people off and go bankrupt overnight. With our finances still more or less intact, this widespread problem proved in our case to be an opportunity.
Semco’s sales had gradually increased again, and we were making enough money to restore salaries to where they had been before the 30% cut. We took back our check-signing privileges. We were surviving in a crisis economy, but only just, and we began to face the fact that we had to cut our permanent staff and contract more of our work. We looked hard for a way of doing it without destroying the support system that Semco people lived on. It was here that NTI’s free-form structure suggested a solution.
Instead of giving contracts out to strangers, we decided we could just as well give contracts to our own employees. We would encourage them to leave the Semco payroll and start their own satellite enterprises, doing work, at least initially, for Semco. Like NTI, these satellites could stay under our larger umbrella by leasing our machines, even working in our plant. Like NTI, they could also do work for other companies, again on our machines and in our factories. Like NTI, their compensation would take a variety of forms—contract payment, royalties, commissions, profit sharing, piecework, whatever they could think up that we both could live with. And like NTI, they could have some beginning guarantees. In particular, we would offer all of them some contract work to cut their teeth on, and we would defer the lease payments on all equipment and space for two full years.
Instead of giving contracts out to strangers, we decided we could just as well give the contracts to our own employees.
But what in heaven’s name were the advantages for our workers, who’d be giving up a secure nest at Semco for the risks of small business? And in the midst of economic bedlam? To begin with, of course, they all had the chance to make many times what they could earn at Semco—if the economy straightened out. Of course that was a big if. And should the recession persist, they might make less. But only assuming they continued to have a job at Semco, which was becoming an even bigger if with every day that passed. The fact was, they had distressingly few choices. And so did we.
What’s the benefit for workers to run a satellite? To begin with, a chance at bigger earnings.
Today, about half the manufacturing we once did in-house has gone to satellites, and we think we can farm out another 10% to 20% in the coming years. Best of all, to this day only one satellite has failed. Some are expanding and looking for partners. Some satellite workers have been rehired by the company, and a few have moved repeatedly back and forth between satellite and employee status as needs—theirs and ours—shifted. Some satellites have broadened their scope so greatly that most of their time—often right on our premises, remember—is spent with customers and production partners who have no other connection with Semco whatsoever.
In 1990, Semco had about 500 employees. Today, we have about 200, plus at least that many in our satellites, with another 50 or 60 people who work for a satellite and also work for us part-time. We have employees with fixed salaries. We have employees with variable salaries made up of royalties or bonuses based on self-set objectives like cash flow, sales, profits, production units, or any one of a dozen other measures. We have employees with both fixed and variable salaries. All our employees share in our profits.
On the satellite side, compensation may take the form of a fixed fee, an hourly stipend, a percentage of increased sales, a finder’s fee, an honorarium, a retainer converting to an advance converting to a royalty, or even a simple win-or-lose commission. In one case, we had decided to kill a product-development project when one of our people picked it up from the table and said, “I’ll take it. If you’ll give me $1,000 a month, which will just pay my expenses, plus a 7% royalty for the first five years if I can make it work—I’ll take it.” So of course we gave it to him. The most we can lose is $1,000 a month, where we’d been spending $10,000 to $15,000 a month and getting nowhere. The most he can make is something like half a million, I’d guess, with the other 93% coming to Semco.
Once we posted a job for one engineer and got 1,430 résumés. We took them home in packs of a hundred, and then we interviewed for five months. In the end, we invited several dozen final candidates to a one-day seminar where we walked them through the entire company, opened our files, showed them everything we did, then asked them for proposals. We wound up hiring 41 engineers—one salaried employee and 40 satellite workers whom we paid on various forms of percentage-based commission.
We owe Semco’s survival to workers who ask: Why can’t we do it better?
Hunting the Free Market
Our ancestors laid out the ground rules of human teamwork several thousand generations ago, and they go like this: the woman with the keen eyesight is Chief Mammoth Finder, the guy with the strong arm and the long spear is Head Mammoth Killer, and the tribal elder with the special feel for herbs and spices gets to be Grand Mammoth Cook. For now. All these positions are temporary and to some extent self-selecting. If you want to be Chief Finder, go find some mammoths and the job is probably yours. But since everyone’s well-being depends on your success, your status is also highly situational. Fail to find, and the job will pass swiftly and naturally to someone else.Generally speaking, Semco’s production process works along similar lines, both for satellite operations and for the work we do in-house. All work, including some aspects of management, goes to people with proven track records who want the jobs and can compete for them successfully. Satellite as well as in-house business units rise and fall on their merits alone—at least in theory.
This commitment to free-market principles was put to the test about a year into the satellite program, when our marine division found itself with a good deal of idle capacity. Marine’s strategy was built on quality, not price—low volume but high margins. A shipbuilder seeking the very best performance and dependability in, say, a propeller system, tended to come to us. But with the economy in a straitjacket, orders had nearly disappeared.
On the other hand, our biscuit division, which designs and builds turnkey cookie factories for global giants like Nabisco and Nestlé, had two fairly big contracts in hand, one for about $2 million, another for $5.5 million, and was going to need a lot of skilled subcontracting. The portion of this work that marine could do would keep it occupied for four or five months, and marine’s top manager (called a counselor at Semco) went ahead and figured the contracts from the biscuit division into his budget. But biscuit’s purchasing people did not award the contracts to the marine division, which took too long to deliver, they said, and which charged too high a price for its exaggerated quality. They gave the contracts instead to satellite producers and outside contractors, including one of marine’s archcompetitors. The fight that triggered was a bitter one, and the attempts of the interdepartmental management meeting to act as go-between did not make things easier. We were of two minds ourselves.
On the one hand, we were going to pay marine employees to sit around on their hands while another division paid outsiders to do work the marine employees could have done. And on the other hand, how could we ask biscuit employees, who share in their division’s profits, to subsidize a business in trouble? Moreover, wouldn’t the subsidy just postpone marine’s inevitable reckoning with its own strategic predicament? In the end, we let biscuit have its way and endorsed the need to be as unforgiving toward our own business units as we would be toward outsiders. It was the right decision, of course. We finished the cookie factories on schedule, and the marine division—which decided to stick with its high-quality, high-margin strategy but to eliminate a number of products whose quality and cost were too high for the market—cut its staff by 70%, began farming out a lot of its work to satellites, and recovered its profitability.
Control…
At the center of Semco is a group of six so-called counselors, and all of us take six-month turns as acting CEO. We also do six-month as opposed to yearly budgeting, because an annual budget tempts managers to postpone unpleasant decisions to the third and fourth quarters.The budget cycles are January to June and July to December, but the CEO cycles begin in March and September. In other words, we avoid what other companies and shareholders think they want—responsibility nailed down to a single man or woman. Our CEOs don’t wear themselves out trying to meet quarterly financial goals, and there’s no one person to blame if the company goes down the drain. When financial performance is one person’s problem, then everyone else can relax. In our system, no one can relax. You get to pass on the baton, but it comes back again two-and-a-half years later.
Having six CEOs means financial results are shared—and no one can relax.
This self-selecting element in decision making is another consequence of the deliberate fragmentation of responsibility. Like our predecessors the mammoth hunters, the people who get responsibility are the people who seek it out and meet it. In fact, the actual, ad hoc control structure we work with from day to day builds on this principle and on two others that, together, create a kind of invisible order from the apparent chaos that characterizes the Semco environment.
The first principle holds that information is the ultimate source of virtually all power. For this reason, we try to make all of it available to everyone. All meetings are open. Designs and specifications are shared. The company’s books are open for inspection by employees and for auditing by their unions. In short, we try to undercut and so eliminate the process of filtering and negotiating information that goes on in so many corporations. Meetings are chaired by the person who knows most about the subject under discussion rather than by the person who has the highest declared status or apparent income.
The second principle is that the responsibility for any task belongs to the person who claims it.
The third is that profit sharing for employees and success-oriented compensation for satellite enterprises will spread responsibility across the Semco map. With income and security at risk—and with information readily available—people try hard to stay aware of everyone else’s performance.
To give an idea of how all this works in practice, let’s take one of those turnkey cookie factories. The Big Cracker Company of Chicago wants a plant that will turn out a thousand tons a month of, say, butterscotch macaroons. To begin with, an independent agent will tell us of the project in return for a finder’s fee. We will probably put the initial customer interface in the hands of a satellite company—four men who used to be on our payroll and now work for themselves. They’ll go through the specifications with the customer, then they’ll share that information widely, announce a meeting (to which anyone can come and no one is summoned), and chair the discussion (which will cover several unexpected proposals from unanticipated participants like the guy from refrigeration with a special point of view about handling butter and coconut). Someone, a group of employees or satellites, will take on the job of costing the project, and with this cost estimate in hand, a Semco counselor and biscuit-division coordinators will then set a margin and deliver the quote to Big Cracker.
(A couple of times, we have even communicated this margin to the customer, because we thought it would be easier to justify, for example, a 12% net margin than to play the disingenuous game of claiming pencil-thin profits and no room for compromise. Our chief argument has been our profit-sharing program, since it seems so clear to us that people will work harder for more money and that a generous margin will therefore buy the customer much extra care and effort. But I’m afraid we’ve had only limited success with this approach.)
This margin-setting discussion often produces serious disagreements. In one case, we battled out the margin in a long, heated debate, and then the sales manager lowered it dramatically when he sat down with the customer. By Semco rules, that kind of last-minute capitulation is perfectly legitimate. Battle or no battle, he was with the customer, not we. Whoever holds the spear is completely in charge of bringing down the mammoth.
Let’s say Big Cracker accepts our bid and the order comes in, 600 pages long. Let’s also say we choose a coordinator for the project from outside the company, and let’s call him Bob. Bob will go through the contract and decide how he wants to divide it up. He may get help from engineering. He will certainly get help from all the meetings he holds to make decisions, which he will chair and where he will lobby for the people he wants to do each job. Next, the biscuit division’s purchasing department will negotiate contracts with the dozens of suppliers chosen. In 1991, we did about 70% of such a contract in-house. Today, that’s down to 35% or 40%.
When Bob has put together a completion schedule and time chart, everyone will go to work—each contractor, employee, and satellite responsible to no single authority but answerable to everyone. At most companies, when something goes wrong the real responsibility falls between the cracks. At Semco, the fact that Bob is not an employee makes everyone react much faster when there looks like trouble.
…And Lack of Control
Semco needs to maintain in-house just a limited number of functions—top management, applications engineering, some R&D, and some high-tech, capital-intensive skills that we do exceptionally well. We don’t care how everything else gets done, whether by contractors or subcontractors, satellites or nonsatellites, former employees or total strangers or by the very people who do the same thing for our competition. None of that matters.When we started, people warned me that all sorts of information about our company would get into the wrong hands, that we had to protect ourselves. I heard the same argument when we started distributing profit-and-loss statements to our employees. But it’s a waste of time to worry about leaks.
It’s a waste of time to worry about leaks. After all, why worry about yesterday’s information?
Finally, we don’t think people give out much information anyway. I know. I’ve tried on numerous occasions to get a copy of, let’s say, a pamphlet some company passed out to 1,000 employees, and nobody can lay their hands on one. The Chinese printed hundreds of millions of copies of Chairman Mao’s Little Red Book, and still they’re as rare as hen’s teeth.
People also warned me about the loss of central goal setting and control. I admit that the lack of control is often hard to live with. But let’s not compare Semco’s circumstances with some ideal world where managers actually get to decide what people will do and when and how they’ll do it. We have limited control over the day-to-day behavior of the people who make most of our components, but so do companies that do all their work in-house. At least none of our satellite people work nine to five and leave their problems at the plant when they go home at night—which means leaving them to management. We have motivation and responsibility working on our side. Our satellite workers are in business for themselves, so they’ll work all night to complete an order to specification and on time. And if the order is late or fails to meet our quality standards, then we’re free to give the next order to someone else. We can forget the witch-hunt and all the grief that goes into firing people or not promoting them.
As for planning and the control it presupposes, I think good planning is always situational. Thinking about the future is a useful, necessary exercise, but translating such conjecture into “Strategic Planning” is worse than useless. It’s an actual barrier to survival. Strategic planning leads us to make things happen that fly full in the face of reality and opportunity.
For example, Semco is today in the environmental consulting business, which I could not have imagined five years ago. Our gadfly NTI group was looking at one customer’s need for an environmentally active pump—a pump that would shred and process the material it moved—and saw that the company could reengineer its production line to do away with the pump altogether. Had we said, “We’re in the pump business, not the environmental business,” we might never have pursued that solution to the problem. As it was, we addressed the company’s overall need, jettisoned the pump, and when it was all over, we’d also acquired a small environmental consulting firm to flesh out our own limited expertise. More recently, we’ve also entered into a joint venture with one of the world’s leading environmental consulting groups. Today, the division represents about 14% to 15% of our total business and is growing at a rate of 30% to 40% per year.
The lesson this story teaches me is about the negative value of structure. Structure creates hierarchy, and hierarchy creates constraint. We have not utterly abandoned all control, but the old pyramidal hierarchy is simply unable to make leaps of insight, technology, and innovation. Within their own industries, pyramidal hierarchies can generate only incremental change.
Take dishwashers, one of Semco’s businesses. Dishwashers are expensive to operate and messy to use, but over the last 50 years, dishwashers have changed hardly at all. What the customer wants is a machine that washes dishes silently, cheaply, and without any mess at all, which probably means without water. I’ve recently seen indications that such a thing may be possible, but the idea could never have come from a pyramid of Semco dishwashing executives. It was one of our satellites that brought us the idea. In fact, about two-thirds of our new products come from satellite companies.
What goes for planning goes equally for culture, vision, and responsibility. We find that fragmentation is strength in all these areas. Semco has no corporate credo, for example, and no mission statement. An articulation of company values or vision is just a photograph of the company as it is, or wants to be, at one given moment. Snapshots of this kind seem to hold some companies together, but they are terribly static devices. No one can impose corporate consciousness from above. It moves and shifts with every day and every worker. Like planning, vision at its best is dynamic and dispersed.
At Semco, so is responsibility. We have little control, even less organization, and no conventional discipline at all. People come and go whenever they like; many set their own compensation; divisions and units perpetuate themselves however they can; satellite companies work on our machines in our factories for us and others in a great confusion of activity; the system tying it all together is painfully loose—and this is manufacturing, much of it assembly-line manufacturing.
At Semco, we have little control, less organization, and virtually no discipline.
The point is simple but perhaps not obvious. Semco has abandoned a great many traditional business practices. Instead, we use minimal hierarchies, ad hoc structures, self-control, and the discipline of our own community marketplace of jobs and responsibilities to achieve high-quality, on-time performance. Does it make me feel that I have given up power and governance? You bet it does. But do I have more sleepless nights than the manufacturer who runs his business with an iron hand and whose employees leave their troubles in his lap every night? I think I probably sleep better. I know I sleep well.
We delivered our last cookie factory with all its 16,000 components right on time. One of our competitors, a company with tight controls and hierarchies, delivered a similar factory to the same client a year and two months late.
A version of this article appeared in the January–February 1994 issue of Harvard Business Review.
No comments:
Post a Comment