Remember when build-to-order was limited to big, expensive products like airplanes and highly customized plant machinery? Those long ago days before Joe Pine wrote his groundbreaking book on mass customization, before Michael Dell turned the computer industry on its head with his wildly successful build-to-order model, and before the internet brought product configurability to everyone's desktop? In less than a decade, those days have been swept away and a new world is being ushered in.

It's a world where companies of all stripes and sizes are trying to adopt the build-to-order or assemble-to-order concept. Granted, these enterprises have little choice in the matter, since choice - and lots of it - is today's most pressing customer demand. A made-to-stock product that comes close to meeting a customer's needs is no longer sufficient. Buyers want to choose options that result in a tailor-made product that fits their needs, desires and budgets precisely - and they want it delivered without a weeks-long wait.

"I think that is a theme running across most industries now," says Bob Vinton, CEO of RockySoft Corp., Fort Collins, Colo. "Everybody is getting into a mode where they expect to have pretty much immediate configurability to their particular interests."

As a result, many companies are looking for ways to implement mass customization: the production of individually customized goods at mass production speed and efficiency. Mass customization is not as easy for manufacturers as mass production, but it does promise to solve many problems inherent in a make-to-forecast model - problems that are exacerbated by today's proliferation of products, short product life cycles and extreme variability in demand.

Take the pesky and perennial problem of unsold inventory. Forecasts are always wrong, to a greater or lesser degree, so companies always have too much of some products in one place and too little in another. In the latter case, potential sales are lost and, in the former, unsold inventory has to be drastically discounted to get it out the door. And with the multiple sales channels now available to consumers, accurately forecasting demand will only get harder. "Today's seasonal demand variation will pale in comparison with the demand swings that firms will face in 2003," according to a July report from Forrester Research, Boston. Unpredictable buying patterns of web-wielding consumers, it says, "will destroy firms' ability to accurately forecast production."

Moreover, when products have long lead times and are built according to inflexible schedules, companies are unable to respond effectively to marketplace changes. And competitive advantage is lost if new products have to be delayed while obsolete inventory is cleaned out of the pipeline - a particularly serious problem for short-lived high-tech goods.

Mass customization or build-to-order (BTO) has the potential to cure these headaches, if companies are able to implement it. As the personal computer industry has shown, however, BTO execution is easier for young enterprises that never were mired in traditional "push" forms of manufacturing. Michael Dell, the poster child for this model, was never a mass producer. In the beginning this was because "he could not afford to build computers and put them on the shelf in the hope that a customer would come along and buy," according to Dave Gardner, president of Gardner & Associates Consulting, Marlborough, Mass. But Dell saw the value in building a computer only after receiving a firm order and this practice continues to help his company thrive, while older, build-to-stock competitors like Compaq and IBM have scrambled to catch up.

Make no mistake, however - moving from a make-to-stock to a build-to-order model means changing every aspect of your business. It is not an incremental step. "This really has to be driven by someone very senior in the organization, typically the CEO, to get that critical mass going across all the traditional boundaries within a company," says David Bovet, vice president of Mercer Management Consulting and co-author of the new book Value Nets. He also cautions that any action should be preceded by in-depth customer research. "You have to really understand what it is your customers want and value before jumping into the details of manufacturing processes, assembly, what to outsource, what not to, how to actually build to order," he says. "Our advice is to get a lot more detailed insights than is available from traditional market research before you blow up your factory."

One thing that makes it difficult for traditional companies to shift to a BTO model, Gardner says, is the prevalence of ERP and legacy systems that are optimized for mass production and are set up to create separate bills of material for each configuration. "While this process is efficient when there is a limited number of configurations, it has dire impacts if each order configuration must be documented," he says. "This problem is compounded when you realize this effort will likely have no benefit for any future orders."

Overcoming this problem with a product like personal computers, which have only about 30 different parts, is one thing. The issue becomes more complicated with very complex products that have thousands of parts and a much larger supplier base.

"Dell is successful because the product is fairly simple," says Gardner. "Ford and General Motors are going to have much bigger challenges when it comes to automobiles," which have about 3,000 parts.

That is not stopping these automotive giants from pursuing the goal of a five-day, make-to-order car. GM's purchasing czar Harold Kutner's summary of what will have to change for GM to reach that goal, as reported in a recent Wall Street Journal, also summarizes the demands of BTO execution: "Everything [must change]," he says. "Vehicle design would have to change to make cars easily buildable. Factories and suppliers would have to be wired together via the internet. Shipping times would have to be whacked in half." Kutner's ultimate aim is to be able to get cars into production within a day of being ordered online and out the factory door within three days.

Shorter Lead Times
As this timeframe makes clear, speed is a key component of today's BTO strategies.
"The challenge for our build-to order customers is how to reduce lead times," says Molly Hunter, vice president of products marketing at Epicor, Irvine, Calif. "That normally is the No. 1 business issue they face." Epicor provides software solutions for pure build-to-order companies in the metal fabrication industry.

Reducing order-fulfillment cycle times is difficult even in a make- to-stock environment, says Bob Reary, product evangelist at PeopleSoft, Pleasanton, Calif. "In build-to-order it is even harder, so most companies today tend to quote standard lead times, say six weeks, on any make-to-order. The problem is that if they make it long enough to cover standard deviations they may be non-competitive, but it they cut it down to perhaps four weeks, they will be late 30 percent of the time. Either way, they still end up with problems."

The culprit is a lack of information, says Hunter. The set BTO lead-time that companies typically quote is based on their historical performance, she says. "They may lose a job because they do not quote a delivery date that is fast enough, when in fact, given what is currently happening on the shop floor, they could do it sooner. But they don't have enough information to commit to that."

Technology is helping companies attack both these problems: reducing long lead times and providing the supply-chain visibility needed to make reliable delivery promises for individually configured orders.

To begin with, companies need to look at their front-end planning for timesaving opportunities. "If you look at a total lead time of maybe six to eight weeks, the internal planning time is maybe two to three weeks of that, so there is a big opportunity there," says Reary.

Hunter agrees. "A lot of the lead time is associated with just the planning process it takes to get the order onto the shop floor to begin the production process," she says. "So what we do is focus on tools to help our customers reduce that front-end planning."

One key tool is the configurator. "Ours is a rules-based configurator as opposed to a features-and-options configurator," she says. "It is designed not only to allow you to step through the options associated with this product, but to apply logic behind the scenes."

Reary also emphasizes this point. "A rules-based configurator lets you be much more flexible. It is not tied to a rigid bill of materials, which you have even when working with modular builds and set features and options." With a rules-based system, for example, customers are not asked to provide redundant information. If you are buying a cabinet and decide that the color is black, you will not be asked to re-enter the color on details like knobs. The rules will fire that information based on the earlier decision.

He draws a simple analogy with a fast-food restaurant. "If you buy something like chicken fingers, you will automatically be asked to choose a sauce. If you buy a hamburger, that question will not be asked." This type of logic can dramatically reduce the time required for product configuration.

Rules also allow companies to program a configurator to prevent mistakes, such as combining incompatible components. Johan Arts, director of marketing at QAD, Carpinteria, Calif., says his company's Chameleon Configurator can reduce the time needed to generate complex configurations from hours or days to minutes. And this does not count the many more hours saved by not having to deal with incorrect orders.
Among PeopleSoft customers, says Reary, typical results after installing a rules-based configurator are these: a reduction of 30 percent in order-to-fulfillment cycle time; order accuracy improvements of 5 percent to 30 percent or more; and a 30 percent to 40 percent reduction in the number of bills of material, though this reduction can be as much as 80 percent.

Thinking about the role of configurators continues to evolve. Reary says PeopleSoft is working on adding industry-specific intelligence. "The vision we see going forward is further simplifying these things by personalizing the configurator for different industries like healthcare equipment. We see this idea of a rules-based configurator, plus a workflow engine coupled with personalization by industry as really being dynamite."

Mercer Management thinks many businesses will move toward web-based "choiceboards" that include not only product features but options for service surrounding the product as well. Mercer emphasizes the strategic value of choiceboards. They can help a company develop an ongoing one-to-one relationship with customers, for example, and can improve profits by making it easy for customers to add to or upgrade their purchases. "No longer forced to spend money on features they don't want, customers are predisposed to purchasing more of what they do want," says a Mercer presentation. "The margins on the upgrade may be much higher, decommoditizing the purchase and producing additional profit."

The consultancy warns, however, that "without the correct supply network, choiceboards will inevitably fail."

Supplier Networks
The supply network certainly plays a key role in reducing cycle times and is critical to providing the visibility needed to make reliable delivery promises. Very tight supplier relationships are essential to successful BTO operations.

A leading technology in this area leverages the internet to model the supply-chain in real-time, identify potential constraints, and come back with a reliable delivery date.

One vendor advancing this technology is webPlan, Kanata, Ontario. Its SupplyIT product completely integrates key suppliers and customers into the supply chain so that, using only a browser, they can do detailed "what if" planning that incorporates the manufacturer's actual production schedule.

"When a customer places an order, the manufacturer needs to understand its impact on inventory, production, labor and so forth. That requires consultation with its suppliers - and those suppliers consulting their suppliers and so on down the chain," says a webPlan white paper. Under a typical system today, even using electronic communication, it often takes days or even weeks to fully analyze the impact of an order when a multi-tiered supply chain is involved. "What's more the further you are down the supply chain, the more of a 'bullwhip' effect you endure - which costs time and money," it says.

By bringing both suppliers and customers into a hosted web setting and giving all stakeholders access to the same information and to analytical software tools, webPlan dramatically reduces these delays. "Now you can get the answers you need, and get back to your customer with a quote, in a matter of hours," the company promises.

For example, SMTC Inc of Markham, Ont., is a provider of electronic manufacturing services and a major supplier to Dell Computer. SMTC has implemented the entire webPlan suite of products. If Dell unexpectedly increases the size of an order, SupplyIT automatically alerts STMC's suppliers to the change and allows them to respond with best-case scenarios for meeting the increased demand, all at web speed.

A key element in the webPlan solution is that it does not rely on interactions with traditional systems, says Darryl Praill, vice president of marketing. "We extract all the manufacturing data, processes and rules associated with MRP [materials resource planning] and it sits on our platform entirely in memory so there is one repository of information. We have series of servers with many gigabytes of RAM. We're talking serious computational horsepower."

RockySoft, which targets the electronics industry, says that with its e-SupplierLink and real-time capable-to-promise engine "manufacturers can simulate their supply chain in seconds to identify material and capacity constraints, eliminating weeks from complex planning processes."

"What we do is connect the manufacturers' data and systems together with the data that is in the supplier's planning system, so when a manufacturer receives a change in demand they can test it by just entering the quantity, time frame and part number," says Vinton. "Our real-time capable-to-promise engine then will execute a simulation of that demand to the existing plan that resides both with the manufacturer and supplier to determine if they are going to run into any constraints to fulfilling that request."

e-SupplierLink can handle very large simulations of more than 100,000 part numbers in less than 5 seconds, he says. "What the system has done during those five seconds is explode through the manufacturer's bill of materials and capacity plan, as well as the suppliers' bills of material and capacity plans, on-hand balances, on-hand orders, those sorts of things. Then it comes back and tells us for that specific scenario what can actually be delivered and when. We can get a real-time response so that in the space of a phone call to a customer or salesperson we can put the answer back and have an extremely high degree of confidence that it is an executable answer as opposed to one that is guesswork."

Where constraints are found, these systems allow manufacturers and suppliers to collaborate on a solution, such as adding overtime or contracting out work. "This is where the internet really adds value," says Reary. "If you are filling orders from capacity instead of inventory, you have to have flexible suppliers and be tightly integrated with them."

"Our view on collaboration is to first find out where we might have a problem and then do a very focused collaboration," says Vinton. "What we want to do is connect the data together and let the manufacturer who has the demand in front of him answer the question immediately, am I going to have any problems or not, without having to interact with any suppliers - all this happens automatically. Then if an answer comes back and says we have a problem with a supplier, then we want to start a very focused collaboration on what we can do to solve the issue."

Epicor's integrated advanced planning and scheduling product also allows users to determine the capacity needed to fill orders and then to do "what if" scenarios to see the impact, for example, of adding shifts or overtime. "From the first time that I quote a job to a customer, I can say - based on everything we have running on the shop floor right now - how quickly I can get this order to him and be reasonably assured that we can meet that commitment," says Hunter.

Chris Jones, senior vice president of marketing at SynQuest, Atlanta, stresses the importance of being able to model the supply chain from end to end, including logistics. SynQuest's Dynamic Sourcing Engine allows trading partners to consider not only capacity and manufacturing lead times but sourcing location and transportation lead times as well. Looking at all aspects of the supply chain, Jones says, allows customers to "make decisions based on what we call 'lowest delivered cost.' That means they are able to intelligently understand such factors as transportation cost and lead times, the fact that one plant might be more expensive than another to produce a product, how close the customer is to certain suppliers and so on. Our product will dynamically look at all of these and source that individual order for the least amount of money possible."

Scheduling
Encompassed in the process of promising an order is scheduling it to meet that promise. "Probably one of the biggest requirements of supply-chain technologies in BTO is to be able to receive an order and schedule or allocate it (against capacity already reserved for that customer) and then provide a real-time availability quote," says Chris Givens, vice president of product development at Adexa, Los Angeles.

"This is fundamentally something that traditional ERP systems have not been able to do because of their batch orientation and limitations of architecture that requires hours, in some cases, to run a solution," he says. "We can run our solution both in batch, which is very fast because of memory resident technology, and then also run incremental plans for a single order."

From a data modeling point of view, he says, this is an area where BTO requirements differ fundamentally from make-to-stock. Make-to-stock tends to look more at process flows and work-in-progress because orders are being fulfilled from finished-goods inventory. Build-to-order looks at capacity and raw materials, so "it is important to be able, for every single demand and forecast, to identify all of the work orders and purchase orders and inventories that have been allocated to build that item."

Synchronized scheduling becomes particularly difficult when parts are being built in different locations, an increasingly common occurrence with the growth of contract manufacturing. "Let's say you were buying office furniture with individual specifications and you have promised delivery Sept. 1," says Jones. "Of course, you want it all on that date, so what SynQuest's solution does is to look across the supply chain and say, 'OK, to meet Sept. 1, chairs need to be produced on Aug. 15, desks need to get done on Aug. 12, and you need to use this form of transportation to go from where the desk was made to the DC."

Arts of QAD predicts that BTO manufacturers soon will begin to more actively bring suppliers onto the manufacturing floor by asking them to take responsibility for kanban replenishment systems that often are used in BTO operations. "We see kanban being extended beyond the four walls of the enterprise," he says. "We think suppliers will be asked to replenish on kanban systems by using the internet as the vehicle for communicating signals. This is an area where we are doing a lot of development with our new products and we think it will make for a more efficient, more agile process."

Major changes also have to be made in production operations to move from a build-to-stock to build-to-order model. Companies must be able to work with shorter production runs, says Arts. This means reducing setup times so that lines can change over more quickly. "It's really agile manufacturing," he says.

Even before manufacturing, products have to be designed for mass customization or build-to-order. This typically means consolidating components or raw materials so that many different parts can be made from a single source part. A company using sheet metal, for example, may move from using many different sized blanks to one standard size, which could then be cut by programmable laser cutters for each order. Similarly, plastic parts could have multiple features molded in to fit a wide range of applications.

As solutions continue to evolve and companies continue to align their inner company e-business processes, the trend toward BTO will accelerate, says Givens.

"It's the Holy Grail of the supply chain because people want to be able to configure and meet the specific demands of all their different customers and they want to be able to do it without inventory risk. So yes, absolutely, we'll see more of it."