Why Weekly Forecasting? (Part Two)

“What you sold or used in the past is usually a good indication of what you will sell or use in the future”. This is one of the basic “truths” of forecasting future demand of stocked items. In our last article, “Why Weekly Forecasting (Part One),” we examined why, for some products or industries, maintaining usage history by week rather than by month will result in more accurate forecasts. These tended to be items whose usage followed a cyclical pattern throughout the month. For example look at a bar graph of the usage, by week, of this item over a three-month period:

Notice that 50% – 53% of total usage each month occurs in the first week of the month. If we are to base our stocking decisions on selling an average of 5.0 – 5.6 pieces per day, we would probably not have enough stock for the first week in each month where usage is 10.7 to 12.9 pieces per day. As a result, the demand forecast for this item should be based on weekly usage. There are other situations where maintaining usage by week is necessary for accurate forecasts of future demand. In this article we will look at two of them: new stock items and products whose usage is dependent on a specific event.

 

New Stock Items

It is common for new stock items to have a spike in sales or usage volume soon after they are introduced. This temporary high volume may be due to:

• Promotions for the new item or salespeople featuring the new item in sales calls.
• Customers wanting to try the new product.
• Customers establishing a normal stock quantity of the product in their inventory.

Whatever the cause, this spike in sales is often followed by a dramatic decrease in usage:

Though the duration of the temporary increase will vary, it can usually be measured in weeks rather than months. Look at the usage of a specific new item that was added to inventory in January:

The item will be overstocked if we base the stocking decision for February on January’s usage of 295 pieces. Because it is hard to predict when the “peak” of the new item usage will occur (as well as the subsequent “trough”), buyers should review the usage of new items every week (and make any necessary adjustments to replenishment parameters) until a consistent pattern of usage is observed.

 

Items Associated with a Specific Event

Lights for Christmas trees, fireworks for the Fourth of July, and pumpkins carved for Halloween are examples of items associated with a specific event – but all specific-event items are not necessarily “holiday goods.” Several of our clients sell industrial supplies. Many of their manufacturing customers schedule plant shutdowns and maintenance around July 4th and the week between Christmas and New Year’s Day. Again, for many items, we see a usage pattern in which the quantity sold in one or two specific weeks in July and December is very different from the usage in other weeks of the month:

As in our other examples, if we were to stock based on monthly usage (i.e., four or five pieces per day), we would not be adequately stocked for the scheduled plant shutdown weeks. Accurate forecasting for these seasonal events again requires examining weekly usage – that is, the quantity sold or used in the same week last year, adjusted for increasing or decreasing trends in business.

An accurate demand forecast allows use to meet or exceed customer expectations of product availability with the least amount of inventory. While few demand forecasts are 100% accurate, we must continue to strive to reduce the forecast error (i.e., the difference between the forecast and actual usage) to better predict future demand of products. After all, no major league baseball player has ever achieved a batting average of 1,000 – but this fact does not stop them from trying to improve and play better baseball. Shouldn’t you also continually do your utmost to improve the profitability and productivity of your investment in inventory? One of the ways to do this is to apply forecasts based on weekly usage whenever it is appropriate.