June 11th, 2009 at 10:41am
Automakers have been studying how to devise a build-to-order system for years. The idea is to only build cars which customers have ordered, rather than fill up dealers’ lots with months worth of cars and then have customers buy what’s on the lots. So far almost everyone has focused on how to build a car faster, or process the paper work more quickly. But in the following article Steve Beeler proposes an intriguing way to do it. I’d love to hear what others have to say with his proposal.
Nothing much good can be said about the vast stocks of unsold vehicles on dealer lots: billions of dollars of scarce capital are tied up, lot rot deteriorates quality, customers can’t find exactly what they want but delivery times for special order vehicles are interminable, and margins suffer as incentives are required to move the metal. The first manufacturer to solve the build-to-order puzzle will have competitive advantages in customer satisfaction and premium pricing.
An auto company’s order-to-cash process spans all of its functional chimneys. A system-level, customer focused view of order-to-cash will encounter a number of dependencies: platform strategy; product plans; features and options complexity; manufacturing sourcing, flexibility, and capacity; and so on. It will also encounter two long standing Detroit paradigms: minimum cost equals maximum profits and booking profits at gate release.
The current wholesale “push” vehicle distribution system has roots back to the end of World War II. In the years following the war, there was so much pent up demand for vehicles that anything with four wheels was snapped up. In an environment of excess demand, it made good sense to mass produce vehicles in inflexible plants at the lowest cost mix and sequence. Booking profits at gate release gave accounting profits a little boost and seemed harmless when captive credit companies had unlimited markets for their paper and the vehicle was sure to move off the dealer lot…and quickly.
Times have changed and the wholesale “push” is no longer serving Detroit well. The industry is plagued by excess capacity, no pricing power, and more than four months of finished vehicle inventories. With credit markets frozen, financing finished vehicles is problematic. No longer is cash flow positive at gate release. No longer are finished vehicle inventories assets.
Is a build-to-order “pull” distribution model a viable alternative to the current build-to-forecast “push” system? Yes, with a twist. The feasibility of a retail customer driven pull process has been demonstrated through a discrete event simulation. The proof-of-concept build-to-order model (see schematic below) centralizes the finished vehicle inventory at the assembly plant and segments the market into “stock” vehicles and “special order” vehicles.
In more normal times (e.g., an economy with functioning credit markets), aggregate annual forecasts for vehicle families are reasonably accurate. But with thousands of dealers, the “right” vehicle’s customer is almost certainly at the “wrong” dealer. The twist is to not ship a vehicle from its assembly plant until a retail customer asks for it.
A typical vehicle family has tens, if not hundreds, of thousands of buildable combinations. However, a relative small number (10% to 20%) account for a relatively large number (80% to 90%) of the vehicles built and sold. This provides an opportunity to segment the market into “stock” vehicles and “special order” vehicles.
Click Image for Larger View
Paradoxically, stock vehicles can be built to forecast. On any given day, five or six red sedans with black interiors and V-6 engines will be purchased at a dealer somewhere. Tomorrow, another five or six of this same build combination will be purchased somewhere else. And then again the day after that. Why not have these vehicles in the assembly plant yard waiting to be shipped? The proof-of-concept simulation predicted 90% order fulfillment from only a three-day supply of finished vehicles at the assembly plant. Ten percent of the production schedule would be set aside for special order vehicles…with premium pricing.
The customer is happy, but the supply base is being killed by daily schedule instability, right? Not at all. In the proof of concept simulation, the take rate for V-6 engines was set at 60% of a total production of 1200 vehicles per day. Daily usage varied from 900 to 540 engines and was in statistical control. This level of daily usage variation is easily absorbed by typical assembly plant inventories. The engine plant can ship 720 V-6 engines everyday. The assembly plant inventory will grow and shrink but it will be stable and predictable.
But what if the market changes? Feedback will be almost instantaneous. If there was a gas price spike and I-4 engine demand grew relative to V-6′s, inventory levels in the assembly plant yard would reflect the shift within days. In the time it takes the supply base to adjust, pricing actions would be taken to bring supply and demand back into equilibrium.
Will some dealers still choose to carry large stocks of vehicles? Perhaps. Adam Smith’s invisible hand will determine the optimum level of dealer inventories. A dealer in a high-traffic suburban location might find it beneficial to carry more inventory than a rural dealer with a lower sales volume. In all cases, vehicles will have been “pulled” not “pushed” into dealer lots and both the OEM and the dealer will obtain market clearing (and better) pricing.
In summary, here are the competitive advantages of a “pull” vehicle distribution process:
- (1) improved customer satisfaction and better pricing (the right vehicle in the right place at the right time at the right price)
- (2) huge cost savings in finished vehicle inventory carrying costs (90+ days @ $10 per day is real money)
- (3) nearly instantaneous market feedback
- (4) a barrier to entry against Indian and Chinese imports (build-to-order cannot be duplicated from an Asian assembly base)
Detroit’s build-to-forecast “push” distribution model is unsustainable. As described above, build-to-order “pull” is an operationally feasible alternative.
Steve Beeler is a director at Production Modeling Corporation (PMC), an operations engineering and management consulting firm based in Dearborn, Michigan. His practice targets middle market manufacturing and service companies. Steve joined PMC after a 20-year career at Ford Motor Company where he guided assembly plants through ISO 9001 registration and led cross-functional / multinational enterprise simulation projects. He is a professional engineer and holds a BSME from Massachusetts Institute of Technology and an MBA from Indiana University.