Mass Customization Seeking Mass Appeal

While the full potential of mass customization — producing individual garments precisely tailored to each customer while maintaining the cost efficiencies of mass production — hasn’t been reached yet, the industry is moving closer to the goal.

Mass customization offers not only the elusive “perfect fit” to customers but also the possibility of lower inventories and fewer product returns. In addition, the information retailers need to gather to participate in mass customization can, when combined with other customer data, become a powerful marketing tool.

Computer-aided design and computer-aided manufacturing will play integral roles if mass customization develops further, according to industry observers. Indeed, CAD’s ability to translate individual customer orders into patterns and finished garments would be crucial to a wide-scale mass-customization program.

Advanced scanning technology that can record virtually every contour of the human body and translate it into data that can be used by a CAD system is available, but its high cost and relatively large footprint have so far limited its installation in retail locations.

Some retailers, notably Brooks Bros. here and Paris-based Galeries Lafayette, have instituted limited mass-customization programs. Brooks Bros.’ program is an extension of its existing custom apparel business. The retailer uses technology that combines detailed customer profiling with an electronic order-entry feature for one of its premier apparel vendors, Martin Greenfield.

“The manufacturer gets the order as an e-mail, but since it’s a hand-tailored shop, a lot of the garments are still cut by hand,” said Scott Pearson, director of special order at Brooks Bros. “The next step would be to tie [the electronic order] into the electronic cutting equipment.”

Brooks Bros. has been testing the electronic-ordering system, from ImageWare Technologies, Atlanta, for six months in its flagship Madison Avenue store and five other New York locations, and has cut one week off its custom clothing delivery time, which averages six weeks.

In addition, “the system cuts down on potential errors and deviations,” said Pearson. A traditional paper order “leaves certain things up to interpretation, but in the [electronic] order-writing process, the system requires a specific way of asking for something like an alteration.”

Brooks Bros. has also been testing full-body-scanning technology developed by Textile/Clothing Technology Corp., or [TC]2, an apparel-industry technology consortium based in Cary, N.C., for customized shirt manufacturing. The retailer is “happy with the results we’ve had with shirts,” said Pearson, and eventually plans to install the technology in at least one of its stores, but is still working on the details of how its customers will interact with the technology.

“Long-term, this technology could have an impact on how products are measured and manufactured,” said Pearson.

“Short-term, it will give us a digital image of the customer, with myriad measurements we can take. Some of the questions we have are what measurements do we want to take, and how does [the technology] fit into our system?

“In addition, the scanner takes a lot of square footage,” said Pearson. “In a flagship store, square footage is at a premium.”

Scanner manufacturers are attempting to make the technology both more compact and less expensive. The SYMCAD scanners from Telmat Industrie, Laneuveville, France, are reportedly being used by some French and British retailers, including Marks & Spencer. Each scanning unit costs approximately $43,000 according to the manufacturer.

David Bruner, research and development manager for [TC]2, acknowledged that both the technology’s footprint and its cost are still difficult issues for many retailers. Its scanner, with software developed as part of a joint venture with the U.S. Department of Commerce, costs $100,000. “That’s down from $200,000 last year, but we’ve long targeted a cost of under $50,000,” said Bruner. “The hardware [comprises] the main part of the cost, and that’s been on a rapidly declining price curve,” he added.

[TC]2 has also focused on CAD’s role in making mass customization a reality. “We’ve worked on completely automating the link with the store site, so that the order for a shirt feeds into a made-to-measure CAD package,” said Bruner. [TC]2 has run tests where a custom shirt is produced only two to three hours after a scanned image is taken, he said.

In addition, when the customer’s exact measurements are added to a database that contains other data, such as preferences, hobbies and purchase profiles, it becomes a significant asset. Retailers can target their marketing more efficiently, for example: “When you’re getting more information from a wider group of customers, you can take care of your better customers better,” said Pearson.

Mass customization and more flexible manufacturing processes also offer manufacturers a number of opportunities. “Inventory costs and markdowns are a significant fraction of the cost of a garment, often 20% to 30%,” said Bruner. “But a mass-customization product is essentially sold before it’s made.”

Mass customization is the latest high-tech oxymoron

Mass customization is an oxymoronic concept that may be the next major trend in the marketing of PCs. The basic idea behind mass customization is that a vendor would ship about five basic systems, leaving the ultimate configuration of those systems up to the reseller. In this way, customers could select the features they wanted and both the manufacturer and the reseller could increase their price points. To some extent, this mode of operation has been a strong marketing point of the mail order operations of such vendors as Zeos International Inc, Dell, and Gateway 2000 Inc. To really impact the mass market, however, system design would have to customized at the level of the CPU. In this model, vendors would ship a number of basic designs, and the CPU modules would be ‘plugged in’ at the retailer level pursuant to customer specifications.

So you’re in the PC business and volume is the game, but everyone wants multiple product lines that each have five price points. Nice for marketing but a true rubric for manufacturing, purchasing, and finance. Wouldn’t it be great if there were some way to “tune” a generic system unit to fill many positions? If you say “Yes,” you get what is meant by mass customization.

This concept of mass customization is the latest in a long line of great PC industry oxymorons (artificial intelligence, multiuser PC, etc.). The concept here is that it is imperative for PC manufacturers to develop a way for a very small number (five or less) of basic system offerings to be customized/configured at final point of sale or by the end customer into many different product position points.

This isn’t a new idea though. It is really one of the fundamental advantages that mail-order PC suppliers have always enjoyed. While Dell, Gateway, and Zeos don’t quite build everything to order, their inventory of finished product is tiny compared to reseller-oriented vendors.

Don’t believe that similar designs deliver mass customization, either. It’s easy to offer the same basic system board and enclosure with multiple CPU types. Many vendors already do, but it’s not even near a model for mass customization. Once the system is built, it’s cast in concrete for all intents and purposes.

In order, then, to develop a useful approach to mass customization, the product designs have to be changed to really make the “customization” of the specific system a lot less like manufacturing, and a lot more like putting a tape in a VCR. No cabinets need be removed, for example.

So what to do?

The single biggest customization point is the CPU. “Positioning” in the PC business is driven by the CPU. Focus there because it determines which systems are “in.” Industries such as apparel (with 20 times our gross margin) can cut price to move unwanted product. PCs don’t work that way. Ask Compaq about the Aero or IBM about some Ambra products.

Clearly, significant mass customization means making it internally manageable. Trying to take on the mass customization of the CPU, storage, and video all at once is foolhardy. If you look at the product reality today, it’s clear that most storage subsystems are similar, if not identical, within the business or consumer product lines. Video, too. CPU customization is the lever.

My expectation is that the reseller will stock at most three to five platforms without a CPU: the consumer pizza box, the consumer tower, the business pizza box, the business tower, and the “Cadillac” box. For example, Circuit City would probably have only the consumer lines, while Entex would have the business and Cadillac ones. These boxes are shipped without CPU modules, which are installed at delivery or by the customer.

The second key to success in mass customization is the physical connection of the CPU module. It would plug into the system via a PCI or other bus, in about the same way you plug a Gameboy cartridge into the console. This mass customization is a hit-or-miss situation. Do it right and inventory costs will drop and manufacturing forecasting will no longer be an obtuse black art. The key is to take the volatility of the CPU choice out of the equation. Engineers must make it simple and effective, and for the customers that will be all the attraction necessary. After all, when your company has to replace its tens of thousands of bad Pentiums, imagine what the cost difference on this kind of “mass customized” system would be!

ResumeRouting and mass customization in the call center

Timely presentation of customer data to call center agents has become essential. However, this in itself does not accommodate the concept of mass customization. It is also necessary to deploy a mechanism to route each call to the agent with the most appropriate skills to meet each caller’s needs. The automatic call distribution (ACD) system determines where each call is routed. Earlier ACD systems merely routed calls to the agent that had been idle the longest, in an attempt to even out the workload. This is efficient only if all calls are basically alike, and all representatives equally qualified. However, by first determining each caller’s unique needs and routing each call appropriately, the call center can deliver a higher level of customer service. The ACD may take into account the caller’s type of account, what the caller has purchased in the past, or special service requirements or contract terms.

Many factors determine the extent to which a call center can customize response to callers. The kind of information in the database and the timely presentation of that information to representatives is essential and has become a focus of many CTI solutions providers.

Though inarguably powerful, these factors do little to advance the concept of mass customization in the call center unless they also use a mechanism to ensure calls are connected to a representative with the proper skills, desire and attributes to meet the customer’s needs.

In the days before automatic call distribution (ACD), companies often had a receptionist who knew every customer and every employee. With a few questions, he or she could identify the best person to handle the call. Customers were satisfied with this personalized service, and employees were efficient, effective and confident in their ability to help the caller.

In today’s call center environment, determining who picks up the phone when a customer calls is the function of ACD. ACD is the software algorithm that controls call routing within a call center. Traditionally, ACD routes calls based on which agent has been idle longest, with the intent to even out the workload among representatives. The Implicit assumptions in this environment are that all calls are for the same basic purpose and all representatives are equally qualified to handle any call. The goal is to minimize idle time of representatives and answer a greater volume of calls. The business results are impressive, particularly compared to the costs and difficulty of answering and routing calls individually in the pre-ACD era.

Therein lies today’s call center challenge: maintain the business economics of ACD while providing personalized service. In order to provide a customized service response to each customer, the call center needs to assess the unique needs and opportunities associated with each caller and connect that caller with a resource that delivers the complete set of skins and information. Neither the caller nor the call center representative can be viewed generically.

To provide a customized response, the call center might consider the customer’s account type (consumer/commercial, major/minor, new/existing), products or services they have purchased before, special service requirements or contract terms, language, how long it has been since they last called, who helped them on their last call, account standing, what media they prefer to use for communicating with the business, and the caller’s level of knowledge (for example, you would provide a different level of technical support to an end user than you would to the IT manager).

This information can be used to select a representative with the best corresponding set of skills and skill levels, not only in products and services. but also in working with different types of customers, transactions, media and business situations.

ACD manufacturers have tried to respond to this challenge with a number of software enhancements to existing routing products. These include overflows groups, queuing to groups of similarly skilled representatives. routing to specific representatives. and allowing representatives to be logical members of more than one group. These extensions to the classical “next-available-agent” algorithms are known as “skills-based routing.” The applications were designed to address the growing range of customer requirements by creating collections of interwoven skills-based groups, referred to as supergroups, splits or vectors.

This approach can work where two or three skills are considered, but has serious draw-backs as a vehicle for achieving solid customized service. Current skills-based routing products (until ResumeRouting) limit the number of criteria considered in the routing decision, placing limits both on the number of skills associated with a representative and the number of skills identified as critical for each call.

Furthermore, interwoven groups are cumbersome to design, administer, maintain and troubleshoot. Consider the following simple calling tree for a computer help desk. Callers are answered by an Interactive Voice Response (IVR) unit and asked to respond to four multiple-choice questions:

Assuming that branches for LANs, mainframes and peripherals (not shown) are the same size as this PC branch, the call center manager, using conventional group-oriented skills routing, must design and administer a minimum of 192 skill groups to allow comprehensive matching of the callers’ requirements to qualified representatives. However, current products generally limit the number of skills associated with a representative to from four to eight, making true customization in this relatively common scenario impossible with conventional skills-based routing technology.

To accomplish this, call-routing software must better distinguish among call center representatives and better define callers’ requirements on a dynamic, call-by-call basis, choosing the right representative for each caller. This is neither technically possible, nor logistically desirable, with group-oriented skills-based routing.

Virtual Groups, Call by Call

There is a new software product available that overcomes the limitations of group-oriented skills routing by creating a fully qualified virtual group on a call-by-call basis. The product is a server-based application called ResumeRouting. This powerful new technology creates a database of representative skins, skill levels and preferences – detailed resumes of individuals, rather than generic profiles. Based on input from sources such as Automatic Number Identification (ANI), Interactive Voice Response (IVR) responses, Dialed Number Identification Service (DNIS), and the customer database, it is capable of defining a Skills Expression for each call.

The Skills Expression lists up to 100 skills and attributes needed to serve the caller efficiently and effectively. Representatives are then selected based on the Skills Expression and the caller is queued to this virtual group. This scenario is analogous to automatically redesigning a skills-based call center for each call to precisely meet the needs of every caller – the critical mechanism in achieving mass customization.

The Skills Expression

A Skills Expression is the set of skills and skill levels needed to effectively complete each call. For example, Bob is an IT manager calling for product support. He answers a series of questions, indicating that he needs information on accessing a mail-server over an Ethernet LAN from his Windows 95 workstation. ANI identified Bob and the fact that he is technically skilled. The skills expression includes a high degree of technical skill on e-mail, Ethernet and Windows 95. If Bob is an important client who normally receives special treatment, “Bob” is also part of the Skills Expression, gleaned from the ANI. Bob’s call is routed to appropriately skilled representatives dedicated to serving his company and aware of its policies and practices.

Using this new technology, call center managers can set priorities and thresholds for skills requirements. These priorities let them know when skills requirements are relaxed, when to broaden the pool of representatives and the order in which calls need to be answered.

The Resume As An Enabler

For call center managers to meet the requirements of the caller, agents need to be identified by their resumes. In the traditional sense, a resume describes an individual’s capabilities and level of experience. Here, the representative’s resume lists a virtually limitless number of skills, a skill level (ranked 0 to 9) and a level of preference for using that skill (also ranked 0 to 9). As representatives receive training and gain experience in additional areas, their resumes can be adjusted, using simple point-and-click commands.

This new technology allows an expansion of “skills” beyond the boundaries imposed by group-oriented skills-based routing to Include such things as product knowledge, familiarity with specific customer accounts, languages, state-by-state licensing requirements, and transaction types (sales, service, product support, collections, etc.).

For some call centers, representatives with comparable product skills may have different transaction skills. Representative X may have a high success rate in sales calls, but average results in customer support. Representative Y has average results on sales, but exceptional talent for customer support calls. Taking these differences in transaction skills into consideration can result in greater call center productivity and caller satisfaction.

The ability to address a broader view of the representative is critical in ensuring that customers are handled by the best resource to address their particular needs.

ResumeRouting and Media Blending

Today a company’s best customers probably communicate by phone. Tomorrow they will surely communicate via the Internet, as is often the case for high-tech call centers. or by fax or video. The range of skills and other criteria necessary for the most competitive “caller” contact thus expands beyond products, transactions and languages to include media competence – such as skill in telephone communication, text-based communication, video equipment and presentation, and the ability to navigate the Internet – as well as having a workstation with the appropriate equipment and applications. Again, this cannot be accomplished without the ability to consider detailed criteria for each caller and each representative. If it could be done in a conventional fixed group call center, not only would it require incredible design and administrative effort. it would result in division of resources into such small groups that economies of scale normally associated with a call center would be eroded.

However good the technology is today, tomorrow it will be better. And as technology becomes even more capable of delivering seamless “mass customization,” the norms of today will become the exception, rather than the rule. One thing that becomes immediately clear is that the call routing engine in call centers must be more powerful, flexible and dynamic. ResumeRouting uniquely enables true implementation of mass customization and the multimedia call center to provide the next increments of customer satisfaction and productivity improvement for tomorrow’s call center.

  1. Computer Press:

1 for PC 2 for LAN server 3 for mainframe 4 for peripherals

  1. Operating System Press:

1 for DOS 2 for Windows 3 for OS/2 4 for Macintosh

  1. Applications Press:

1 for word processing 2 for spreadsheets 3 for desktop graphics 4 for Lotus Notes 5 for other applications

  1. Network Press:

1 for local area network 2 for wide area network 3 for public network


You’re an IT manager at a growing major medical association that has 24 geographically dispersed locations, over 2000 doctors, and is handling 700,000 inbound and outbound calls and 70,000 appointments a month.

One day the boss tells you that last year’s growth figures were way up, next year looks even better, and you’ve got to find a way to further enhance the association’s communications and networking capabilities in order to keep up!

The IT manager at MacGregor Medical Association faced those challenges and was still able to continually improve patient care, enhance productivity and minimize expenses by using CTI.

MacGregor did this by using Siemen’s CorNet to link over 30 PBXs combining both voice and data into a single network with feature transparency at every location.

The new system ties patient records with calls, via Automatic Number Identification and Siemen’s CallBridge link. Agents can see patient records at the same time they receive calls, enabling them to give more efficient, personalized service.

No matter what location a patient calls in to, they receive uniform and seamless treatment. For instance, if a patient is placed on hold for more than two minutes when calling into a remote site, the system’s Automatic Call Distribution (ACD) function transfers them to the message center at the main hub in Houston. This reduces caller frustration and the number of abandoned calls.

The technology also provides MacGregor a centralized voice-mail system and appointment desk. The centralized appointment desk has greatly improved responsiveness to patients – 98% of inbound calls are answered within 30 seconds.


MaxServ is a rapidly growing provider of information and consulting services and telemarketing for the home services industry. Using a virtual call center solution, the company increased its level of customer service and gave its call center operation the flexibility to adapt to quickly changing business environments.

With call centers in Austin, Tex.; Scottsdale and Tucson, Ariz.; and Mobile, Ala., MaxServ has grown exponentially over the last five years. During that time, revenues have risen from $2 million to over $50 million and call center staffing from 40 agents to about 2,000.

According to Tom Gentry, MaxServ corporate information officer, all four sites are part of the company’s virtual call center, which handles as many as 80,000 calls per day. The sites are all linked together by Siemens’ CorNet product and MaxServ’s wide-area network (WAN).

Via Siemens CallBridge CTI link, MaxServ’s WAN-based application polls the call centers at each site for activity levels and makes call routing decisions within the computer application, which then tells the public network where to route the call. The switch does not receive calls before routing decisions are made, optimizing network utilization and simplifying call processing.

In the conventional automatic call distribution (ACD) scenario, call routing is based primarily on telephony criteria like the number and order of calls in queue and the length of time calls have been in queue. MaxServ has added value to that model by implementing skill mapping, based on customer requirements and agent skills. CTI lets the computer application define the routing scenarios and triggers and can use virtually any information in the information systems environment. This added information allows MaxServ to further automate call routing via software applications, substantially reducing the time call center managers spend defining routing scenarios and balancing resources.

MaxServ has also created a call center management application that will eventually encompass real-time information from other computer applications. This will enable call center managers to view not only agent and group telephone activities, but factors such as daily revenue can open service tickets all within one application.

Let customers have it their way

IT tools are being used to enable mass production of customized products. growing number of manufacturers are pursuing a value-added competitive strategy of providing more products and services tailored to individual customers. The goal is differentiating companies and their offerings, and it is being done with production costs comparable to mass-production methods. Making the transition from standardized products and production to flexible production and customized products is difficult for many companies. They are relying on IT to provide the flexible, cost-effective manufacturing systems that can fulfill the requirements of the new product paradigm. These tools build Manufacturing Resource Planning-II and Enterprise Resource Planning technologies. The mass customization strategies of Motorola’s Pager div, Andersen Windows, Levi Strauss, John Deere Harvester Works and ITT Hartford are briefly discussed.

Let’s say you’re in the market for a new pair of shoes. You could go down to the Shoe Warehouse and search through every brand of footwear ever made. Or you could go to a shoe store that lets you pick the design and color as well as the size. And, oh by the way, the price is the same.

Goodbye, Shoe Warehouse, right?

That’s the idea behind what many are calling a new business paradigm taking root in a wide range of industries. Rather than continuing to mass produce for the increasingly elusive “average customer,” these manufacturers are using state-of-the-art information technology to build and deliver products and services designed to fit the precise specifications of each individual customer. And they do it while retaining the cost efficiencies offered by mass-production techniques.

“I personally believe mass customization is an important new business paradigm in our industry,” says Hugh Martin, vice president and director of agency personal lines insurance at ITT Hartford in Hartford, Conn. “Competitors who think this is only an optional thing are going to be at a disadvantage.”

Many others feel the same way:

* ANDERSEN WINDOWS began allowing individual customers to design their own windows four years ago. Customers design their windows on graphical PCs in hardware stores and other retail outlets, and their specifications go directly to Andersen’s plant in Bayport, Minn., where the windows are manufactured. The “Window of Knowledge” system, now in use at 387 Andersen dealers, has substantially increased the company’s business in custom-designed windows and has forced competitors to copy the system.

* MOTOROLA’S PAGER DIVISION in Boynton Beach, Fla., manufactures hand-held pagers to individual customer specifications. Third-party retailers use Apple Macintoshes to help customers design the pager features they want Within 20 minutes, an individual customer’s order in lot sizes as small as one is started on its way down Motorola’s highly automated production line. Within an hour, it’s completed and ready for shipment

* LEVI STRAUSS has begun installing PC-based systems in retail stores that allow customers to design jeans to their own specifications rather than picking them off the rack. The customer-generated specifications are transmitted to Levi’s plants, where the custom jeans are produced on the same lines that manufacture standard Levis. The option, called Personal Pair, could become a key feature of 200 new Levi-owned stores due to open by the year 2000.

“In just about every industry, price and product quality have been such a focus in the last few years that you need those just to play,” says Pam McNamara, director of supply-chain management consulting at Arthur D. Little in Cambridge, Mass. “What differentiates you now is value-added service – whether you’re able to do mass customization in terms of packaging, production specifications, and delivery,” she says.

A recent A.D. Little study shows that such factors are key to retaining customers. And, says the study, depending on the industry, a 5% increase in customer loyalty, can generate a 25 to 85% improvement in corporate profits.

Manufacturers that produce individual orders also find it very easy to quantify their savings on finished-goods inventories. The John Deere Harvester Works manufacturing plant in Moline, Ill., is an example. Some of the wide variety of crop planters it produces sell for over $100,000, and until a couple of years ago, the plant was a typical mass-production operation, with long manufacturing cycles. Being unable to respond quickly to individual orders forced the plant to project demand and keep a finished-goods inventory of about 300 planters. But since completely revamping its manufacturing processes to respond to individual orders, Harvester’s Moline plant keeps only 20 finished machines in inventory.

Although the basic idea behind mass customization is fairly simple, for most companies the transition is not. Mass production techniques have pushed companies into standardized, one-size-fits-all products; long product lifecycles; and highly automated, inflexible manufacturing that values efficiency and low cost over flexibility. That focus historically drove IS to create manufacturing systems such as MRPII that were first and foremost targeted toward supplying the raw materials needed to feed the mass production manufacturing machine.

Companies pursuing mass customization strategies are trying to turn that thinking around. Rather than pushing standard products out the door, they’re trying to build flexible manufacturing systems that are tightly integrated with order-fulfillment systems.

“We want to use the same economies of scale and take advantage of the same efficiencies and reconfigure those to deliver highly customized services,” says ITT Hartford’s

Without exception, companies pursuing mass customization are relying on their IS organizations to help pull off this transition. They need new types of systems that go beyond a narrow material-management view to take a broad view of overcoming the constraints on meeting individual orders. They’re asking IS to apply that view across the entire supply chain-from taking a customer’s individual order at one end to providing the logistics needed to deliver the finished product.

Says Bill Fulkerson, staff analyst at John Deere: “You have to engineer your systems so that they are able to minimize constraints and maximize flexibility. The best way to do that is with software.”

In many cases, IS managers like Fulkerson, are building these flexible systems to augment and integrate with, rather than replace, MRPII or its relational successor, Enterprise Resource Planning (ERP). New software tools make it possible to build on top of MRPII or ERP, which then become a central transaction engine (see table, “New Tools for Mass Customization”).


While new IT tools can help IS managers in their mass customization strategies, there are still several gaps. Perhaps most glaring is the lack of consistent integration of new tools or real-time scheduling with mainstream ERP packages. While some vendors such as Tucson, Ariz.-based Avalon Software have been aggressively integrating their products with the new tools, and others including Baan USA in Reston, Va., have been building customization capabilities into their core ERP products, still others, such as SAP and Oracle, have been slow to do so.

Lack of standard integration has caused many companies pursuing mass customization to limit their application of new IT tools to a small piece of the supply chain. Some, like Andersen Windows, have focused on the front end-collecting customized orders – and haven’t yet begun to revamp the rest of the process. At Andersen, once customized specifications are transmitted through the “Window of Knowledge,” they go to a separate custom line where they’re assembled mostly by hand.

Perhaps the bigger issue is that many IS organizations will need to change the way they operate if they expect to take on a starring role in enabling mass customization. For one thing, some IS departments are finding they must change their historical focus on supporting single functions and systems to become part of cross-functional teams.

Just as important, IS will need to fundamentally change the way it goes about identifying system requirements. Since by definition, customization means that customers are in control, requirements and system capabilities will be a moving target rather than a fixed goal.

Customizing Products

Each planter made at the John Deere Harvester Works factory in Moline, Ill., is highly customized. Customers ran chose from scores of options, including row count and liquid or dry fertilizer, that add up to literally thousands of configurations. The challenge is making the factory flexible enough to accommodate every customer’s requests and to do it quickly; the planting season is short.

Until recently, that wasn’t easy. The wide range of product configuration options made scheduling factory floor resources complex. Production schedules, based on quarterly forecasts and usually manually created, were fixed weeks in advance.

In 1992, when Deere/Harvester set out to change that with a re-engineering project called “Vision 21,” IS decided the factory needed new production-scheduling software tools that would provide shorter lead times and greater flexibility. Deere/Harvesters Advanced Technology Group first searched for ready-made software that would accommodate highly complex products. They didn’t find it But they did find software developed by Bolt Beranek and Newman of Cambridge, Mass., that was being used by the military to schedule certification of aircraft components. Deere/Harvester became the first customer for a commercialized version of the technology, which Game to be called OptiFlex, and is sold by a BBN spin-off, Optimax Systems, also in Cambridge, Mass.

Now, Deere/Harvesters production-scheduling system is flexible enough to allow the plant to reschedule production each day in response to customer orders and other constraints. With more responsive production, the company can get by with less finished-goods inventory, according to operations manager Joe Mausser.

But, flexible scheduling wasn’t the only change. Deere/Harvester also instituted a more modular manufacturing style and began awarding incentive bonuses to workers based on total plant output. The changes, says Mausser, helped the Moline plant to record sales and profits last year.