OEM product forecasting: through a glass darkly.

Here's a true story for you: In early 2001--just as all idleness was about to break loose in electronics production--an electronics manufacturing services (EMS) provider received some orders from an OEM customer. The orders looked suspect to the EMS company--not because of the type of product to be built, but how much of the product the OEM optimistically believed it would need in the coming months.

After talking with its suppliers about a possible impending downshift in the industry and seeing similar orders slack off from other customers, the EMS provider now faced a dilemma: Should it fulfill the order for the OEM customer, knowing that demand probably was not there, or should it limit the amount of product to be produced, saving itself and its customer from possible, but not yet proven, inventory and cash flow problems?

Ultimately, the EMS provider refused to fulfill the order in its entirety. When questioned by the incredulous OEM as to why, the EMS said: "We've seen some orders slack off from other customers in this sector, and we talk to our suppliers in the industry." All indications led to the EMS company's assessment that demand for this product was waning--how fast and how deep would only be proven later.

And, ultimately, because of its foresight, the EMS provider saved itself and its customer some unwanted inventory in the pipeline.

The moral of this story, as related by Technology Forecasters, Inc.'s (TFI) Pain Gordon at her company's recent Quarterly Forum for Electronics Manufacturing Outsourcing and Supply Chain, was not to showcase an uppity EMS provider defying its OEM customer. Instead, the story showed that EMS providers--perhaps now more than ever--really do have a tremendous amount of input concerning the nuances of electronics production. The point, according to Gordon, is that OEMs should welcome this experience and let their EMS partners actually be partners, which includes contributing sound information to product forecasting.

Forecasting really is an abysmal task, whatever your industry. Faced with conflicting reports from the research analysts covering your market niche--and, er, they've been wrong before--a nebulous future in which demand is a slippery target, past histories that can only show you where you've been and sales forecasts often tainted with a particular spin, no wonder product managers at OEMs are often wrong when trying to forecast product demand.

Why is forecasting so hard for OEMs? Gordon listed several market-driven reasons TFI discovered in the research for its recent report, Best Practices in Product Forecasting, with the difficult art of prediction staining them all, Accurately foreseeing customer behavior, the new product introduction ramp, end-of-life decline, competitive behaviors and price erosion and, with wounds still fresh, economic upturns and downturns is just plan hard. In fact, Gordon stated that a few OEMs TFI interviewed actually related that they feel "very unsuccessful" in predicting product demand.

These market-driven factors are hard to control, but Gordon and her team also found some internally driven factors contributing to poor product forecasting that are more controllable. Often, the differing forecasting methodologies across the OEM organization confuse the process; hand in hand with that is the lack of internal emphasis and understanding concerning forecasting. Other factors are the extensive data scrubbing that occurs between data sources used for forecasting and a lack of automation and integration with legacy systems.

The bottom line is that forecasting is a difficult process, but that's no excuse. Several factors can be controlled by OEMs to improve the forecasts they hand to their EMS providers. And one item is improving communication with their EMS partners and other suppliers by involving them proactively in the forecasting process.

Other tips Gordon recommended were to monitor consumption to forecasts for products on a weekly basis, instead of the typical monthly timeframe, and make adjustments as necessary. In addition, some best practices reported by the OEMs TFI interviewed included disciplined, structured measurement and monitoring and cross functional alignment within the organization-with everyone held accountable for the forecast. Finally, develop a clear sales and operations planning process, where statistical-based forecasting is used as the starting point. Shockingly, over a fourth of OEMs TFI interviewed stated that they do not use statistically based forecasting to plan the production needed.

No doubt about it, OEMs are under tremendous pressure from their customer base to get products out on time. As Gordon's story proves, input from EMS providers--who are in the manufacturing trenches daily talking with materials suppliers and other customers--can save their OEMs wasted inventory and money. And isn't that the point of a business partner?

Lisa Hamburg, Editor-in-Chief e-mail: [e-mail]