Jack Harwell

A lot of attention is paid to the value of demand planning and forecast accuracy as it relates to benefits to the enterprise. These benefits include better production, inventory, distribution and buying plans. Customer deliveries are improved, while excess inventory is reduced and operations on the production floor can perform like clockwork. The “right stuff” is purchased, built, and distributed to the right location on time and in full. However, the effect of demand planning extends upstream to the supply network as well. And, with some effort, the payback can be multiplied to not only help the supplier, but also to further contribute to the success of your enterprise. It’s like a smile, the more you give to others, the more you will receive.

Suppliers rely on the forecast of their customers to make their production and procurement plans. In my experience, supplier support for their customer is proportionate to the quality of forecasts they are provided. When a supplier has multiple customers with varying degrees of demand planning capabilities, the customers that share the most accurate predictions of their demand will receive the most support for their future needs.

For example, when the bottom dropped out of the telecom infrastructure market in 2001, I was director of raw material procurement at a fiber optic equipment manufacturer. We had implemented SAP prior to 2000 (Y2K), and had made some great strides in demand planning and forecast sharing with our suppliers. As a result of our efforts, our suppliers were receiving weekly forecasts for all of the items we purchased from them. Because we had an effective S&OP process and shared our materials forecasts, these suppliers enjoyed a fairly accurate picture of our needs. As the decline in telecom equipment demand unfolded, our suppliers were receiving messages every week that told them to back off on their production and procurement plans.

Demand for telecom equipment dropped so far and so fast, no one escaped the challenges of keeping their inventory levels in check. However, I believe we were in better shape than our competitors. In fact, one of our key semiconductor suppliers confided in me that we received favorable treatment in our rescheduling requests because we had become so reliable in planning our demand.

There are other ways to gain from supplier support as a result of sharing good forecasts. For example, an effective just-in-time program requires a level of supplier commitment that goes beyond a transaction-based relationship. A reasonably accurate forecast is a necessary foundation for a mutually beneficial JIT program. In most JIT arrangements, the supplier takes on the majority of the risk. Without the visibility into true demand provided by forecast sharing, this risk can become quite large.

What are some of your experiences with JIT programs? How did your forecasting process contribute to their success? Are there other upstream benefits of forecasting that you can share?

Jack Harwell
Former Vice President / Corporate Officer
Radio Shack

Jack Harwell- Radio Shack

Forecasting & Planning Professionals strive to reduce error in their work.  For all this effort, what is the reward? When asked what the cost of forecast error is, most discussions turn to the consequential impact of inaccurate forecasts.  Consequential impacts include lost sales and profits, expedited freight, excess and obsolete inventory, and unhappy customers. There are also defensive measures that companies take, which are meant to reduce the impact of forecast error.  One commonly used measure that immediately comes to mind is the use of safety stock.  Carrying additional inventory to compensate for unexpected customer demand places a heavy burden on companies, but is considered essential to serving the customer and avoiding lost sales. There are other defensive measures that aren’t typically identified as a cost of forecast error, but have a significant impact to the bottom line.  Consider the activities and infrastructure that we have in place to manage the consequences of forecast error.  These defensive measures easily come to light if you imagine an alternative reality:  The Perfect Forecast.  How much that we do on a daily basis would not be needed if we could predict customer demand with certainty? One of these activities is rescheduling vendor deliveries and production.  The tasks that come along with this such as adjusting requirements, communicating with suppliers or production planners, calculating schedules and confirming new commit dates, in addition to the need for  updating systems takes a tremendous amount of work.   Then there are the systems that support rescheduling which also need attention.  Master scheduling, MRP, capacity planning, and shop floor control all have a significant amount of source code dedicated to the rescheduling of orders. If forecasts were perfect, there wouldn’t be a need to reschedule.  Production schedules, once set, would not require any changes.  Vendor deliveries would be allowed to come in as originally planned.  Imagine if there were no rescheduling activities required.   How much of your employees’ time would become available, what routine activities would no longer be required, and how much of your ERP systems would be idled? For example, at one manufacturing company a very robust Purchase Order (PO) rescheduling process has been developed over the years.  Initially, the process was manual, requiring approximately 12 hours per person each week.  The process was as follows:

1. Run a report that identifies which POs need rescheduling
2. Buyers/Planners manually review reports and mark adjustments
3. Reports are faxed to suppliers for response
4. Suppliers write responses on report and fax back
5. Buyers/Planners manually review responses and change POs

With 8 Buyers/Planners on staff, the annual time spent on this process was about 3800 man-hours or almost 2 FTEs (full time equivalents). After a considerable amount of system development, this process has been automated.  In the new upgraded system, the report could be reviewed online before being released by the buyers/planners.  The vendor would review the reports online or received the data via Electronic Data Interchange or EDI.  Suppliers would respond online or via EDI.  These responses would then be routed back into the ERP system. This new process reduced the staff’s work load by about half.  Even though this was a significant improvement, the process still takes about 2000 man-hours per year, or one FTE.  This manufacturer succeeded in reducing the effort required to deal with imperfect forecasts, but did not eliminate it altogether. As you know, there is no such thing as the perfect forecast.  Forecast error is a necessary part of doing business.  However, there are steps we can take to reduce forecast error and its associated costs.  These are:

• Measuring forecast error
• Continuously improving forecasting models
• Linking unit plans to financial plans
• Collaborating on forecasts both internally and externally.

I look forward to discussing the costs of forecast error and exploring the notion of a perfect forecast at the IBF Conference in Dallas Texas May 4^th – 6^th. Your comments and experiences in dealing with forecast error are welcome!

Jack Harwell

VP Global Sourcing and Supply Chain Operations

Radio Shack

Hear Jack’s Keynote Address at