Forecasting Pipeline Stocks With Incomplete Data

 A well-known food manufacturer was struggling to replenish the stocks of the retailers that sold its products and needed some statistical sales or demand forecasting for its stock.

They were able to control pipeline stocks for the larger retailers since they supplied checkout sales data to help the manufacturer's production planning. However, they had no idea of pipeline stocks for the smaller retailers. For these, an in-house stock planning system was used. This required 12 man-hours per week of manual work to update stock forecasts for over 1,000 products.

Why the Company Came To Me & What I Did For Them

In the end, the company turned the problem over to me. They knew that I had the expertise to improve the quality of the forecasts and to make the system more efficient. Using Statistical Thinking, it did not take me long to design & develop a new system which delivered the following benefits.

1. Automatic estimates of current & future stocks for 95% of products instead of 50%.

2. A 60% reduction in the amount of error in the forecasts.

3. A fall in the amount of manual work from 12 man-hours per week to 1 man-hour.

I also discovered why the analysts were making errors in their forecasts. Previously, they assumed that smaller retailers had good stock control and planned their production accordingly. In fact, total pipeline stocks fluctuated between 3 and 5 weeks of demand which led to erratic shipment patterns.

What Happened Next?

With the improved accuracy of the stock planning system the company became more relaxed about short term fluctuations in their shipment levels. They now knew that poor stock control by their smaller clients was the most likely explanation for this.

However, they were not complacent and my work was the catalyst to begin a major capital investment program to improve the flexibility of their production lines. After all, if smaller retailers were going to be erratic in their stock control, the company needed to be able to change production levels quicker than they had been doing before.