Pat Lancaster ran a highly automated company. That was his problem
Pat Lancaster makes no claim to being a low-tech prophet, and his record bears him out. For 20 years, as the founder and chairman of Lantech Inc., a manufacturer of industrial packaging equipment, he roared down the same fast lane as most American industrialists, acquiring ever-bigger mainframes and ever-speedier machines while hiring battalions of programmers, process engineers, and consultants to keep things up and running. He was so gung ho for the cutting edge that at one point he had seven different computer systems operating in his Louisville factory. In short, Lancaster was no Luddite.
But that was then. These days the Lantech plant is running on technology that could have been installed 40 years ago. Instead of a shop humming with numerically controlled lathes and automated assembly machines, workers are wielding drill presses and hand tools. In place of computers generating specialized work orders, conspicuous "whiteboards" diagram the assembly process step by step. And instead of a mainframe crunching algorithms to maintain process control, Lancaster's managers use simple visual aids like cue cards (to signal when to order new supplies) and strips of tape (to indicate the direction of the production flow).
It's enough to make an automation guru blanch, but there's no arguing with the results. Lantech's productivity is up almost 100% since the company scuttled its old high-tech system in 1992. Completed orders now routinely arrive at the shipping dock in 12 hours instead of five weeks. Production defects are down by half. And most impressive of all, a company that seven years ago was reeling from serious losses in market share and a perilous crisis in corporate confidence has found new life in low tech, and its prospects are rosy again.
Lancaster is quick to caution against overgeneralizing from his company's turnaround. Not all the gains can be tied directly to the decision to go low tech, and obviously some companies would fail in a day if they tossed out their machines. But his experience suggests that despite what we're often told, not investing enough may not be a reason we're having so much difficulty increasing productivity in this country. In fact, we might be investing too much.
Anyhow, Lancaster won't deny that Lantech's revived fortunes stem from a profound shift in managerial philosophy. For what Lantech has done is not simply to abandon automation. It has supplanted the dominant manufacturing paradigm of American industry -- division of labor -- with an organizational imperative predicated on precisely the opposite principle: generalization of labor.
In years past, a typical Lantech factory worker would toil away at a lockstep task prescribed by computer printouts, isolated from fellow line workers respectively minding their confined stations. Today, that worker is part of an integrated team that oversees the entire production process from start to finish and is involved in every step along the way. Instead of the traditional assembly line, Lantech workers are now organized into collaborative "microlines," relying on visual aids and manual procedures wherever possible to keep things shipshape and up to speed.
In essence, Lancaster has wrested his production system from the clutches of computer circuitry and put it back in the hands of his workers. Industry analysts have taken to calling this kind of holistic approach "whole-cycle management" or "lean production"; Lancaster calls it "one-piece flow." On Lantech's shop floor, it all comes down to what is literally an eye-opening proposition: Everybody sees everything.
Lantech's conversion is all the more remarkable when you consider how entrenched the division-of-labor mindset had been at the company. For most of its history, Lantech led the industry in the manufacture of "unitizers," machines that bind items together so they can be handled as a single package. At one time, standard approaches to unitizing included metal-strapping and shrink-wrapping (wrapping items with plastic and then heating the wrap to form a single membrane). In the early 1970s, Lancaster found that stretching plastic wrap around the items, so that they are held by the natural elasticity of the wrap rebounding against the strain, works much better and is also cheaper. The market agreed, and for many years Lantech grew at rates in the high double digits.
The job of the Lantech plant is to assemble the wrappers, massive devices that often occupy more than 1,000 cubic feet. They're also complex. The basic components include roller conveyors, ramps, turntables, wrapping arms, frames, motors, belt drives, stretch-wrap threaders, safety shields, and control systems. Assembly is made more complicated by the fact that a significant portion of the orders call for custom designs. Even a "standard" machine can be ordered in thousands of distinct configurations.
Combine all this technical complexity with the pressure of filling orders for hundreds of wrappers a year, and you have an operation that would seem to be a textbook case for assembly-line automation. And that's exactly the gospel that once was preached at Lantech. For its first 20 years, the company modeled its production system after the prototype Henry Ford had pioneered in his Highland Park Model T factories. Lancaster and his managers defined a set of core processes (sawing, machining, fabrication, painting, electrical assembly, and final assembly) and organized internal operating divisions around those processes. Under that system, you raise productivity by increasing the speed of the core processes, usually by intensifying the division of labor and accelerating automation.
Given the enthusiasms of the times, the standard recipe for higher productivity also included buying a succession of ever more powerful computers to streamline operations. Lantech's workforce was then trained in the care and feeding of the modern mainframe, which boiled down to a few "Dos" and "Don'ts": Don't enter or withdraw any item into or out of inventory, or add value to a part, or move a part from one operating division to another, without explicit orders from the computer. Do tell the computer each time an order is carried out. As T.H. White wrote of ant society, "Everything not mandatory was forbidden." Work orders for human and machine came only from the mainframe. "We didn't drive the computer; the computer drove us," recalls Terry Ferrill, manufacturing engineer. Even so, he can't remember hearing any particular objections. Everyone knew that having computers run the show was just part of modern industry. Obviously you couldn't operate a manufacturing business without them.
In 1989, Lantech lost a key design patent, and with it went the company's edge in the wrapping-machine market. New competitors entered the field, bringing with them extremely aggressive pricing tactics. Lancaster, who up to that point had presided over a thriving company, began to see business fall off sharply. Internal committees were formed to find ways of increasing productivity, but no effective strategy emerged. The company drifted, its people arguing among themselves. "It was our stage of doubt and denial," Lancaster says.
It was during this rocky period that Lancaster began reading about the principles of whole-cycle management, which in manufacturing circles is more commonly referred to as the "Toyota production system." Originally devised and implemented at Japan's "Toyota City," a key doctrine of the system is continuous improvement on a factorywide basis. Higher productivity is achieved not by reducing costs per part, but by lowering costs for the production cycle as a whole, from design to distribution to sales. Yes, you can lower costs by stepping up the pace of certain stages of production. But quicker rates must be defined in the context of the entire production cycle, not as accelerations of a few suboperations. In other words, the job of each employee in a whole-cycle assembly plant is no longer to simply put peg A in hole B as quickly as possible. It is to find a succession of better ways to get the entire cycle to meet the overarching objectives of the business.
Lancaster remembers being struck by how well the idea resonated with his own experience. Time and again, he had seen quick thinking on the factory floor come to the rescue of ill-considered top-down directives. The main thrust of continuous improvement, he recognized, was to systematically cultivate that kind of resourcefulness, not to wait for glimmers of it to surface by accident. "It's always just blown me away how much creative capacity is put into neutral, or even reverse, by limiting people to one specific contribution," he says.
In a plant like Lancaster's, where much of the work is assembly labor, incorporating the tenets of continuous improvement would mean relying on the line worker to spot most problems and to propose appropriate solutions. And therein lay a daunting task. The managerial imperative behind the classic Ford assembly line is to narrow the focus of the workforce as much as possible. The result: Most employees have no real understanding of the full production process.
At Lantech -- and at the majority of U.S. manufacturing enterprises -- the "whole cycle" was a mystery to line workers; and it had become increasingly inscrutable, obscured by reams of computational data. From a continuous-improvement perspective, the theoretical and technological forces driving the American industrial workplace have been evolving in the wrong direction for nearly a century.
"It's my experience that as your business grows, it becomes easier to be talked out of your common sense," Lancaster says. "You turn to automation at a certain stage of revenue growth, and the abstract logic of standard cost analysis takes over. You're looking to reduce overhead by calculating the lowest cost per part, which naturally leads you to match the economic order quantities of the parts required. One setup dictates 100 parts per order; another dictates 1,000. You're building inventory as a buffer between different output capacities. In that kind of accounting environment, you can be seduced into thinking that the black box knows all. Over time, people's common sense gets disconnected, and they inevitably dummy up. The thing that clicked for me about this whole process was that it builds directly on common sense, reinstituting a system that can be understood by the people who are actually running it from day to day."
For Lancaster, all the tenets of the Toyota production system hit home. With the loss of the patent, the technological advantage that had allowed Lantech to carve out up to 50% of the wrapping-machine market had disappeared. For the company to retain its competitive edge, it would have to shift its focus to improving its basic customer services: filling and delivering orders more quickly, customizing machines more readily, producing wrappers with fewer defects and cost overruns. "We had to change the way we think," Lancaster notes. "We had to find a way to turn our workers into stakeholders."
Even so, Lancaster might not have contemplated such a sweeping shake-up had he not happened to interview Ron Hicks for a VP position in the middle of Lantech's doldrums. Hicks had worked for Danaher Corp., one of the first American manufacturers to change over to continuous-improvement production on a major scale, and he knew the problems that had to be faced.
Talking with Hicks, Lancaster began to see that converting Lantech to a system of whole-cycle management would require nothing short of a revolution in the company's institutional culture, but Hicks convinced him that it could be done. In the end, Lantech's business slump was so severe, and its attempts to handle the problems so unproductive, that Lancaster decided to hold his nose and jump. "Being on your back looking up is a good place to be," he says now, though he might not have said so then.
Hicks joined Lantech as vice-president of operations in 1992. The enormity of the task at hand soon became apparent. First, there was the problem of overcoming powerful and deeply entrenched resistance from the workforce. "The old process had created a status hierarchy," Hicks says. "People working on what they thought were the high-status pieces were concerned that now everyone would know what they knew. Others were worried that they were going to be held responsible for everything that happened anywhere in the factory."
With Lancaster's backing, Hicks set about assembling transition teams that evaluated and redesigned every aspect of Lantech's production process. Aided by the TBM Consulting Group of Durham, N.C., a company that helps businesses through the initial transition to lean production, he reorganized production from a "hurry-up-and-wait" pattern, in which inventory was moved in batches from one sector of the factory to another, into a system that consists of several microlines, or production cells. Each of those cells would be responsible for all the processes -- sawing, welding, electrical wiring -- once spread out through the whole factory. The reorganization, in turn, would increase the attention given each step, both because more people would be involved and because the smaller line configuration would make it easier for each person to see what the others were doing.
With this shift from a division-of-labor model to a generalized-labor approach came changes that reverberated throughout the factory. Lantech's mass-production assembly line gave way to what Lancaster calls a "one-piece flow line," which means that any given component moving through production corresponds to a specific customer order.
To streamline that system, Lantech redesigned its assembly carts. Control points were moved out of the computers and made visible. Strips of tape laid out on the factory floor directed workers and carts to the next stage of production. Whiteboards went up all around the plant, charting the production cycle for all to see. Inventory, or "kanban," cards were placed on the bottoms of storage bins; when a card became visible, it was picked up and an order faxed to the supplier.
Four years into Lantech's metamorphosis, there's little question that the new procedures have made the plant's operation not just a paragon of simplicity but a showcase of efficiency. Because each of the plant's 14 assembly carts now represents one particular order and one specific wrapper configuration, there is no longer any extraneous inventory. Under the old system, Lantech produced parts in lots and batches. At any given moment, there might have been sufficient quantities of a particular part for 1,000 wrapping machines. With the new system, Lantech makes one set of parts at a time, fabricating them as dictated by actual orders rather than by spreadsheet projections.
And Lancaster's hunches about cultivating a more involved and knowledgeable workforce have paid off in productive suggestions. Take, for example, the great pallet breakthrough. While Lantech's wrappers are being assembled, they rest on pallets of various dimensions. In the past, the company bought its pallets from an outside vendor. Because the pallets were not designed specifically for Lantech's equipment, occasionally an odd shape or size would interfere with production. However, fixing the problem -- even recognizing that there was a problem -- was no one's responsibility.
The new regime was only a few months old when a shop worker suggested that Lantech build the pallets in-house to its own specs. The company started to do so immediately, with excellent results. As it turned out, the idea made perfect sense, especially with all the extra room available on the factory floor now that Lantech was no longer stockpiling inventory. A breathtakingly simple solution, Lancaster says, but one that probably wouldn't have emerged under the old managerial hierarchy.
Safe to say, too, that manual labor would never have experienced a renaissance at Lantech if the status quo had held. But as the workforce began to hit its stride on the microlines, it soon became clear that Lantech's automated machines were not compatible with the new assembly process. For one thing, outfitting each production cell with numerically controlled machines and rewriting all the materials-handling algorithms in the mainframe were financially out of the question. More problematic still, the high-tech system operated on the assumption that the fewer people with access to data the better. In the new climate of whole-cycle management, Lantech's big machines were dinosaurs.
Today Lantech's mainframe room, once the nerve center of the factory, lies dark and shuttered, the computers unplugged, the mainframe sold. Nobody misses them, and it's no wonder. The low-tech procedures work better, and you don't need an advanced degree in information science to stay on top of the day's tasks.
From where Lancaster sits now, the old system seems downright bizarre. "We were lying to the computer all the time," he recalls. "If we had trouble getting deliveries as fast as a customer wanted, we'd tell the computer the order was two weeks older than it was. The computer would reschedule that job, but then all the rest of the orders would be held up."
"The numbers the computer was working on were never right," Hicks adds. "It might say there were 10 items in inventory, and we could see there were 8 or 12. So we'd just enter the right number over the wrong one. That guaranteed the error would show up again. But we had no idea where those wrong numbers came from or why they were wrong. You could almost never track down where the error occurred." Sums up Lancaster: "We were just automating chaos, buying expensive machines to do wasteful things at higher and higher rates of speed."
The broader lessons of Lancaster's cautionary tale are supported by some important recent studies of industrial productivity. One of the most extensive was a massive survey of the world's automobile industry carried out by MIT's International Motor Vehicle Program over a five-year period. According to its 1990 report, "The Machine That Changed the World," researchers found almost no correlation between efficiency and automation. The most efficient plant (measured by the effort required to perform a standard set of assembly tasks) was one of the least automated, while the most highly automated required 70% more effort to perform the same tasks. Jim Womack, senior author of the report, says that automation imposes three categories of costs that can ultimately offset its benefits: the capital investment for the machines themselves, the high-priced technical and service workers required to look after the machines, and the hidden costs of stockpiling inventory.
The last point is all the more critical for often being overlooked. High-tech processing machines are usually fast, and it only seems sensible, given a fast machine, to make a lot of parts at once. However, it turns out there are lots of indirect costs in having mountains of inventory sitting around a plant, most dramatically when the market evolves out from under all that inventory, leaving the enterprise with an expensive write-off.
One virtue of a system like Lantech's, Womack says, is that the nonautomated machines -- the "hand tools" -- used in the microlines are so flexible that they obviate the need to maintain war chests of inventory. Womack feels that the savings from making parts on demand are so important that he uses the name "lean production" for the whole continuousimprovement system. (A book by Womack on "lean enterprises" is due out this fall from Simon & Schuster. One chapter is about Lantech.) Certainly the absence of towering piles of inventory makes a dramatic impression on anyone visiting the Lantech plant. There's so much space and light, the factory floor feels like a convention center.
Ironically, Lantech's mainstay customers turn out to be manufacturers that are heavily invested in automated mass production. They're the ones most in need of the company's unitizing machines to wrap up piles of parts for transporting from one operating division to the next. If they all were to follow Lantech's lead and switch to lean production, Lancaster would lose business.
Yet according to Anand Sharma, CEO of the TBM Consulting Group, there's not much chance of that happening anytime soon. Sharma estimates that only 50 or 60 American businesses have fully backed out of their commitment to high-tech, division-of-labor management. The obstacles thrown up by institutional cultures are too great, especially when a business is making money anyway.
And even if lean production and continuous improvement do revolutionize U.S. industry over time, Lancaster expects to stay well ahead of the curve. After all, he has his R&D staffers running on continuous improvement too, and their productivity has increased as fast as everyone else's.* * *
Fred Hapgood (firstname.lastname@example.org) is a freelance writer based in Boston.