Some of the computer industry's sourcing problems are a nightmare, with volatile markets, ever-changing mix, and long pipelines.
You'd think after 30 years of making PCs that the industry would have inventory management down, but last year Acer took a bath over excess inventory, and this year the problem appears to be widespread.
Part of Acer's problem lies in the fact that PCs are being replaced by tablets, but that was predictable. Beyond that, understanding what went wrong gives pointers toward handling Internet-based commerce, which is driving toward fast satisfaction, minimal inventories, and low wastage.
The first failure in the chain lies in product mix. The PC market is model crazy. Every vendor stocks many case flavors, each of which can have several internal configurations. This is to satisfy small-step price-pointing and retailer differentiation.
These products are largely assembled in China, and shipped by container freight to the United States and European Union through a seven-week shipping pipe. Add in the manufacturing leadtimes for chips and plastics, disk drives, displays, and more, and we are looking at 20 weeks.
With four weeks at the retail outlet, and a manufacturing cycle of two weeks, the bottom line is those PC volumes were committed before May. In fact, because fourth-quarter production is the highest of the annual cycle, the total production time starts purchasing in March.
Long leadtimes carry considerable risk for low-margin products with ever-moving technologies. An inventory mistake can cause a long glut, as Acer found, rapidly moving into the red on pricing.
So are there strategies to avoid the pain? The objective is increasing flexibility while removing risk. Vendors in the logistics chain set their plans by getting forecasts from the retail end, calculating back and using them to order components.
ERP systems are good at taking forecasts and blowing them out to component orders, but they are incapable of handling the GIGO (garbage in, garbage out) problem. This is more pronounced when a market is volatile. Humans have to adjust the numbers to have a chance of being close, but sales-compensated staffs look to avoid inventory reductions.
Thus forecasts tend to be too high in falling markets. Correction involves several things. First, the forecasts need to be scrubbed hard. That requires an independent review team, separate from sales and marketing.
Second, most companies need to minimize product mix. Follow Apple's example: three case colors on the iPhone and that's it. Ratios of those colors are fixed over time. The small range means more units in each forecast bucket, which improves forecast accuracy.
Third, those leadtimes need to drop. This is a big problem, and the hardest to address, because of Shanghai manufacturing's low cost. But that's the longest pipeline.
In fact, the solution doesn't require entire production to move closer to retail. Split volume into several tiers. The largest tier is the irreducible volume, the worst-case sales scenario. These units are built in the long-lead-time, lowest-cost location (Shanghai in this case).
The next tier represents the likely upside based on retailer inputs adjusted for each buyer's history of forecasting. The forecasts at the end of this process are scrubbed retailer numbers, hedged based on macro-economic trends.
Assemble these units near the point of sale: the US/EU. This provides some last-minute mix flexibility, and much more volume flexibility. This can be committed as late as June, provided long lead items are in the pipe.
Long lead items don't need to be company-owned. I've found most vendors are happy to have in-house component stores that change ownership only when pulled for assembly.
Finally, there's the piece of sales we lopped off when we hedged. This part of the volume should be assembled locally, but the commit point should be as late as possible and expedited assembly used to build the units — a process that should be applied every week of the year.
Units will cost a little more, but everything gets sold, and inventory is minimized. I've successfully built high-volume PCs on this model. There's an upside. Online order fulfillment can be reduced to 48 hours from typically four weeks. Customers do like that!
Share your thoughts on sourcing and inventory below.