The state of automotive make-to-order
Poor demand picture and legacy systems delay progress toward custom configurations
By Gary Ruderman, senior editor -- Manufacturing Business Technology, 8/1/2004 12:00:00 AM
Make-to-order is a manufacturing process wherein raw material or component procurement doesn't begin—or at the least, the product isn't finished—until the actual customer order is received.
Today's producers of the kind of high-value goods typically made "to-order" still have to balance the need for efficient inventory, purchasing, and scheduling policies with the imperative to—as the cliché goes—give the customer what it wants, how it wants it, and when it wants it. But with technology advances of the last five years, some say producing goods to-order without penalizing the customer with long delivery times is increasingly possible.
At least in theory, every step in the make-to-order process could be done concurrently, to the point that the only thing holding back the ultimate make-to-order scenario—a customer-specified automobile produced in three days—is pure physics. But the reality is quite different.
In fact, the automotive industry is a poster child for all the ills that inhibit make-to-order capabilities. According to experts, two things especially thwart make-to-order in automotive: lack of a clear demand signal, and a plethora of legacy IT systems.
What's also clear is that the potential benefits of make-to-order in automotive are enormous, and not just for today's picky consumers. Stock-outs and excess inventory cost automakers tremendous amounts of money.
A demanding picture
A white paper from ERP vendor QAD and the University of Michigan notes the following: "Although the Big Three automotive manufacturers have not universally adopted [it], the industry is moving toward a leaner business model ... This strategy is most effective when customer demand is the signal that pulls product through the supply chain."
Today, though, instead of responding to a clear demand signal, automotive dealers either hunt for a car in the distribution network, or give incentives to customers to buy what's on the lot.
According to a report from ARC Advisory Group, Dedham, Mass., things are changing only slowly. "OEMs are creating interdealer transfer systems that identify a car at a nearby dealer that perhaps matches the customer's requirements. An alternate approach is to hold all finished stock at a central distribution center, and deliver to the dealer from this center. More sophisticated systems allow dealers to view the OEM's production schedule and select an unallocated car. However, the scope for changing the specification is limited."
Advances in supply chain planning and scheduling—as opposed to these half-measures—will drive any quantum leap ahead. Just as important as brute-force computing, however, is use of PCs as communication devices to allow more immediate sharing of actual demand data up and down the supply chain.
According to ARC, automotive build-to-order entails "the ability to automatically process a call-off, pass appropriate pull signals to suppliers, and coordinate material movements through the inbound warehouse, manufacturing work centers, quality control, and the outbound warehouse." For true make-to-order, suppliers will deliver complete modules and systems not only just-in-time, but to a unique configuration. This is called "in-line sequencing."
Case example
Navi Radjou, VP of enterprise applications at Forrester Research, Cambridge, Mass., points to Nissanas a leader in automotive make-to-order and one that makes use of commercially available software. Nissan's make-to-order scheme, called Integrated Customer Ordering Network (Project ICON), was born in Japan but is coming to the U.S. and will be used in Europe next year.
In brief, ICON feeds order attributes—derived from Trilogy Software'scustomer relationship management (CRM) software—directly to the shop floor. Production flows are updated almost immediately to reflect the emerging preferences of customers. At the same time, Manugistics' supply chain management software supports components ordering down several layers of suppliers. Radjou refers to this as a "sense and response" scenario. And for Nissan, he says, it will pay off handsomely.
By 2005, Nissan expects this make-to-order scheme to cut a 40-day order-to-delivery cycle to 14 days initially, and then to seven days. With advances in robotics and shop-floor automation, Nissan will be capable of mixed-model production. At the same time, it will reduce its product platforms from 24 to 10. Finished goods inventory also will be reduced.
This reduction in platforms is important, as the sheer number of features the OEMs offer inhibits the move to make-to-order. According to ARC, in 1999, feature options ranged from 820 build specifications for the Nissan Primara, to 3,930 billion for the Mercedes Benz E-class.
"Nissan is delivering what the customer wants—and what the customer is willing to pay for without incentives—but is not willing to wait months for," says Radjou.
The IT side
While it's widely recognized IT is the key to making make-to-order possible, automakers struggle to a) move off of entrenched legacy systems; b) choose the most appropriate solutions for a make-to-order environment; and—most recently—c) find the investment capital for new systems.
Dr. Matthias Holweg of the Massachusetts Institute of Technology and its International Motor Vehicle Program says that while the automakers are working to integrate best-of-breed applications into ERP platforms, the bill of material and scheduling systems at the Big Three are run on "a complex web of outdated computer systems."
Holweg says most systems are 20 years old and coded in COBOL and Fortran languages, adding, "These are scheduling applications with a lot of manual intervention."
Yet according to Boston-based AMR Research, automotive companies spent only about 1 percent of total company revenue on IT in 2003, and this number is projected to actually decrease in 2004.
ERP is the leading area of software investment for automakers, at 29 percent of application budget. But not addressed is the question of what actual use OEMs make of ERP, nor of the fact that ERP's batch methodology is not considered ideal for lean or make-to-order. (Granted the ERP vendors are working feverishly to make their systems more appropriate for lean production environments.)
AMR says portal technology is the largest area of growth within automotive OEMs—presumably used to attain a single view across fragmented systems and, at the same time, forge electronic connections to suppliers.
OEMs, according to AMR, plan to increase 2004 budgets for supply chain management software by 2 percent, "to improve manufacturing operations through collaboration and visibility initiatives," and will increase by 2 percent their investments in applications like procurement and sourcing.
GM and round the world
General Motors (GM) has its own version of "sense and respond."
Brad Ross, executive in charge of GM's six-year-old order-to-delivery (OTD) program, says the company wants to change the way customers and suppliers interact with the automaker—shortening the time it takes to go from order to scheduling, including with suppliers. It will then deliver a quasi-custom car to the consumer—or to the dealer for further customization—in the shortest time possible.
Ross's team already has made changes that improve visibility and scheduling of custom orders. By moving customer-specific orders ahead of dealer stock orders in the production queue, for example, OTD cut to one week from four weeks the time it takes for such an order to reach production. OTD disproves the widely held notions that making a car faster would cost more, or that changing a production line to respond to customer demand would mean more out-of-stocks.
But change is slow, he says, in a "strongly siloed company"—especially one impaired by complex relationships with independent dealers. Unlike Dell's customers, for example, who are quite used to buying a PC configured online in one Web-enabled action, automotive customers have been programmed to accept what is on a dealer's lot. In the past, that probably was the best strategy, given that any special order had to be integrated with weekly batch orders.
In reality, says Ross, it only takes three days to build a car that is customized to a limited degree. That means one could be ordered, built, and delivered in less than two weeks—with delivery causing the longest delay.
What's still missing, Ross concedes, is good plant-floor and supply chain analytics. That's part of the OTD ideal, but absent from current production mentalities. "When the new model of OTD does mature, it will blur the line between the CRM business process and the manufacturing process," maintains Ross.
Automakers will move to make-to-order because they must. The McKinsey Institute, San Francisco, estimates that producing cars to meet orders that don't materialize costs manufacturers $80 billion a year, and Nissan has estimated that build-to-order manufacturing could save up to $3,600 per vehicle.
