The union was suspicious at first. Wiremold’s former director of labor relations had been an old-fashioned hard-liner and the union presumed that any management offer of job guarantees must contain fine print which somehow reversed its on-the-surface meaning. In the end, however, the union decided that Byrne would deliver on his promises.
Curiously, for reasons Art Byrne finds hard to understand, executives in many companies in the Hartford area were more skeptical about his job guarantee offer than his union. “People tell me all the time that I’m crazy to make an ironclad guarantee of jobs. They say, ‘What if something goes wrong and your sales fall off?’ But my view is that management has five lines of defense before showing people the door: (1) reduce overtime, (2) put the extra people on kaizens (to get a future payback), (3) in-source some components from marginal suppliers we plan to drop anyway (remembering that our equipment is now highly flexible), (4) cut the workweek across the board, and, most powerful, (5) develop new product lines to grow the business. Our employees are now all highly skilled in process improvements and only a concrete head would fire skilled people due to short-term business fluctuations.”
Re-creating the Product Development System to Channel the Value Stream
The product development system Art Byrne found in the fall of 1991 was clearly not going to grow the business. Engineering vice president Steve Maynard remembers that thirty products were under development and all were making slow progress. “We had long queues between the stages in development, we had departments within engineering with batch production, and we had expediters. There were absolutely no priorities except that some projects at some times had ‘the voice of the president’ behind them and received expeditious treatment.” The average project actually making it through the system took three years, but many stragglers were lost in action along the way.
Fortunately, Steve Maynard already knew what to do. He had learned, at a University of Hartford seminar in the fall of 1990, that Quality Function Deployment and dedicated development teams are an unbeatable combination. The seminar was affiliated with MIT’s Laboratory for Manufacturing and Productivity, and MIT professor Don Clausing, one of the disseminators of the “House of Quality” concept, 4 took Steve through the steps needed to introduce the “voice of the customer” in a highly structured, continuous-flow development process.
However, back at Wiremold, the senior management was so busy with the ongoing TQM effort that there was no time for another program. They told Steve Maynard, “Wait until next year.” Fortunately, by “next year” Art Byrne was on the scene. “When I first met Art, I said, ‘What do you think about QFD and dedicated development teams?’ He said, ‘Do both immediately. And by the way, your new target for product development time is now three to six months, not three years.’ We were off and running within a week.”
Steve Maynard’s first step in the fall of 1991 was to start formal in-house training in QFD, using a consultant for technical support. 5 All the senior executives attended this training, just as every manager, no matter how senior, no matter what their job, participated in shop-floor kaizen activities. Art Byrne’s theory was that every manager in an organization must understand the basic activities of that organization, notably product development, production operations, and sales/scheduling, and that the only way to learn was intense exposure to systematic principles.
Next, Maynard and the senior management team asked an obvious but previously neglected question: What businesses are we really in? They reviewed the thirty ongoing development programs and “deselected” those—most of them, in fact—which did not support a specific business: tele-power, power and data management, plastic products, etcetera. 6 This shrunk the number of projects dramatically, and those which remained were prioritized. These projects were then placed in a product plan, showing their target dates for introduction.
For each program judged worthy of continuing, Maynard designated a three-person team consisting of a marketer, a designer/product engineer, and a production/tool engineer. The team was sent to talk directly with prospective customers in the building design and construction community to come up with a broad definition of the product through an initial QFD process. They asked the “value question” described in Chapter 1 and came back saying, for example, “What we really need is a Tele-Power[H23008] Pole which can accommodate any height of ceiling, which can be ordered in a very wide range of colors, and which is unobtrusive.”
Steve Maynard remembers the amazement among many Wiremold old-timers when these teams were formed. “They asked me, ‘Why have we got a tool-design guy out in the field talking to customers? Doesn’t the need for specialization and the division of labor require that tool designers stick to designing tools?’ The security many of them felt in the old departmentalized, everything-in-its-place way of organizing work was truly striking.”
Once the broad definition of the surviving products was determined, a truly multifunctional team was formed to develop a detailed product specification in engineering language. The team was co-located in a dedicated space in the Engineering Department and included the team leader from the appropriate product family (Tele-Power[H23008] Poles in our example) plus the production planner, the production/tool engineer (a member of the original three-person, product-definition team), and a buyer. The team was told to achieve a target cost determined by estimating the market price and subtracting an acceptable margin.
When the precise specification of the product was accepted, detailed part and tool design was conducted by the team, again working to target cost. Toward the end of the process, the whole team moved its desks to the factory floor to go through process-at-a-glance and standard work exercises with the production team handling the product. (Remember that thinking about manufacturability has been present from the beginning. The production/tool engineer was on the original definition team.)
By mid-1992, Wiremold was ready with its first product under the new regime. It had taken only six months and tool costs were only 60 percent of what had been originally budgeted, based on past experience. Even as Wiremold’s managers in the physical production and order-taking process were learning how to see, Wiremold’s marketers, product designers, and engineers were learning how to hear the voice of the customer and how to make designs flow quickly and directly through the development process. 7
Fixing the Order-Taking Process
The third key activity of any business is order-taking, scheduling, and delivery, and Art Byrne made no distinction between this “business process” and the firm’s physical production. It was subjected to the exact same kaikaku and kaizen process at the same level of frequency as every production activity.
As in most batch production organizations, Wiremold’s order entry and shipping was disconnected from physical production. A master schedule in the MRP system, based on market forecasts, was supposed to ensure that adequate stocks of finished goods were always on hand in a massive centralized warehouse, so that when an order was received it could be processed and then shipped from inventory.
The orders themselves were also processed in a batch mode through a central Customer Service Department. This department entered orders throughout the day into a computerized order-processing system. The orders were processed overnight in a batch and, if inventory was on hand, pick lists of what to ship were printed out the next morning in the Shipping Department. Over the next two or three days the Shipping Department at the warehouse would gather the goods and send them to Wiremold’s distributors.
However, items on a customer order often were not available, despite large inventories, so very few orders were shipped complete. Instead “backordered” items were shipped over an extended period as they became available. Because of the MRP system and the large batches in each production run, it was not unusual for a single customer order to ship over many weeks or months. Also, because most orders had delayed items, a large Customer Service Department was needed to keep track of orders and to respond t
o customer questions about delayed items.
The end result of all this handing off of orders and the massive warehouse was that it took almost a week to process and ship an order when everything was in stock. Yet most orders called for items which were delayed for extended periods and the system had many potential sources of errors. The Customer Service Department found it very difficult to keep up with its dual role of making customers happy about delayed or incorrect shipments and spurring the rest of Wiremold to get the job done right.
After a series of kaizen teams went through the entire series of activities—from order-taking to shipping—it was possible to shorten the order-receipt-to-ship time from more than a week to less than a day. To achieve this, orders were sent to shipping four times during the day (rather than in one big batch overnight) and the central warehouse was closed, freeing up 70,000 square feet of space. Upon receipt of the orders, shipping circulated carts past the small finished stock racks at the end of each product team’s production process.
As the shipper withdrew parts from the rack and pushed empty parts containers down a return chute, this became the signal—the only signal—for the product team to make more of a given part. (The MRP system which formerly kept track of the movements of individual parts within the Wiremold production system was gradually given the much smaller task of long-term capacity planning and ordering parts from suppliers not yet on pull systems.)
This new approach, which required many fewer people and resulted in fewer errors, could only be introduced over a period of about two years as Wiremold began to convert from batches to product teams with single-piece flow. Parts which had been produced in one-month batches were soon being produced every day, a feat which required that many machines be changed over twenty to thirty times per day rather than the former three to four times per week .
Although Wiremold’s competitors in the electrical industry are now being forced to match its quick-shipping capability, they seem to be doing it the way so many American firms are achieving “just-in-time,” by maintaining even larger inventories of finished units or by switching to a “Max-Flex” system such as we saw at Lantech, in which mountains of component parts are prepared in advance so that final assembly can be conducted in direct response to customer orders. Both approaches are inferior to a truly lean pull system from start to finish.
Linking Compensation to Profits
Wiremold had always paid base wages at slightly above the average level for the Hartford area. It then tried to reward its workers for good results through a profit-sharing plan funded with 15 percent of pretax profits, paid quarterly in the form of a check, and by contributing shares of company stock as the employer contribution to a savings plan. The problem was that, in the years just prior to Art Byrne’s arrival, there had hardly been any profits and stock values had slumped. In addition, the old batch production system made it hard for employees to see any connection between their own efforts and the success of the firm.
Art Byrne resolved to keep the existing profit-sharing arrangement but to steadily increase profits (“by working smarter than our competitors”) and to show everyone the financials so that the reasons for profitability would be clear. During the first years of “lean management,” profits at Wiremold have increased from 1.2 percent of wages in 1990 to 7.8 percent in 1995, and Byrne is still strongly committed to increasing profit sharing to 20 percent of every employee’s pay.
Improving Suppliers
After many internal improvements were made, it became increasingly apparent that many of Wiremold’s problems were external. Purchased goods and raw materials accounted for a significant percentage of Wiremold’s total costs, yet no effort had ever been made to improve supplier performance. Instead, Wiremold’s traditional purchasing operation had concentrated on controlling supplier profit margins by ordering every part and type of material competitively from multiple sources.
Kaizen teams moved quickly to dramatically reduce the number of suppliers, from more than 320 in 1991 to 73 by the end of 1995. This was essential if Wiremold was going to be able to take time with each supplier to improve its performance. But then it was necessary to start with the most critical suppliers and teach them to see.
In April 1992, a Wiremold kaizen team paid a first visit to Ryerson, a giant steel fabricator much larger than Wiremold, with fabrication facilities spread all across North America. Ryerson supplies Wiremold with large rolls of steel which Wiremold stamps or bends to make the cases for many of its products. Ryerson had adopted state-of-the-art techniques to the extent that it had just begun to deliver to Wiremold every day, “just-in-time.” However, at the back of Ryerson’s plant, the Wiremold JIT team found just what they had expected: a neat row of steel coils, each a day’s supply for Wiremold, fifty days in a row produced by Ryerson in one enormous batch. Just-in-time had been nothing more than an inventory shuffling exercise because Ryerson didn’t know how to produce in small lots.
The Wiremold team therefore went to work on Ryerson’s massive steel cutting machines, which took two shifts to change over from one cutting pattern to the next. This, of course, was the cause of the massive lot of coils laid out in the shipping area. In a short time it was possible to bring changeover times down from two shifts to about thirty minutes, and Ryerson began to meet Wiremold’s needs each day for delivery during the day.
Even better, from both Ryerson and Wiremold’s standpoint, Ryerson was soon able to produce for all of its other customers on a true “Just-in-Time” basis, driving down costs across the board. Wiremold, of course, expected something from Ryerson in return for its trouble, and negotiated a range of special services—like absorption of materials cost increases for extended periods and extra-short runs of steel for certain low-volume applications. As a result of Wiremold’s proactive stance toward a key supplier, Wiremold, Ryerson, and all of Ryerson’s other customers were much better off, a win-win-win achievement for lean thinking.
Devising a Growth Strategy
Art Byrne notes that “our production system and its needs are fundamental to our strategy.” Because the application of lean thinking to batch-and-queue organizations liberates tremendous amounts of resources—people (including engineers and managers), space, tools, time (to get to market much faster), and cash —it is both possible and necessary to grow rapidly. It is possible to grow rapidly because the means are self-generated; it is necessary to grow rapidly to provide work to support the job guarantees which are the social basis of the system. In consequence, Wiremold has grown rapidly along three tracks .
One important means of growth for a lean organization is to rethink what can be done in continuous flow. We believe that many organizations try to do too much—in particular, to control suppliers of “key” technologies. But many organizations, like Wiremold before Art Byrne arrived, also do too little in the way of physical production because they imagine that scale economies require the purchase of many items from firms using enormous, high-volume machines in centralized plants to supply these items in massive batches to many customers.
Cord sets are a nice example. Wiremold products use enormous numbers of cord sets—the wire and plug ends used to connect surge protectors and other power conditioning devices to a power source. In the past, these were produced in large batches by cord set manufacturers supplying many firms like Wiremold across a range of industries. The problem was that Wire-mold’s production was constantly being jeopardized by the lack of the proper cord sets as sales trends changed. Wiremold would have brown when only white cords were wanted or have twelve-foot cords when the customer wanted fifteen-foot cords. Resolving these shortages often took two to four weeks due to the batch production methods of cord set suppliers.
When Byrne arrived at Wiremold he asked, “Why can’t we produce cord sets, indeed at the same rate and in continuous flow with our end product?” And, as is usually the case, when Wiremold’s tool engineers looked at the economics of cord set production, they found that the cost and
time savings from using small, simple machines merged into the production sequence for the finished product not only overcame the problem of having the right cord set on hand as demand shifted, but also reduced the cost per cord set. So, Wiremold has begun to supply its cord set needs in-house. After all, Wiremold has plenty of excess space, plenty of extra people, and cash readily available to buy or make the necessary, simple machines.
Any would-be lean producer needs to look at this issue more generally, asking in each case, “What physical activities can we incorporate directly into a single-piece-flow production process?” Doing this also reduces the number of suppliers dramatically, making improvement of the remaining suppliers much easier.
Wiremold’s second growth strategy has been to buy up small firms with allied product lines (and who use batch-and-queue methods) in order to increase the scope of Wiremold’s product offerings. The first wave of inventory reductions at Wiremold (during the first two years of comprehensive kaizen activities) produced $11 million in cash. This money was used to buy five firms with complementary product lines generating $24 million in sales volume.
In essence, Wiremold was able to convert $11 million of muda (in the form of inventory), which would have cost about $1.1 million in carrying costs (assuming 10 percent as the cost of money and storage), into $24 million in new sales volume, which at a 10 percent operating margin generates $2.4 million in income. The $3.5 million income swing is highly significant for a company the size of Wiremold (with about $250 million in annual sales). Equally important, because the product lines of the five firms were complements to existing lines, Wiremold’s sales force suddenly had a much more complete range to offer customers, which helps increase the overall growth rate.
Lean Thinking Page 17