An accurate forecast of future demand is a crucial element in properly stocking products. Many computer systems include time-series forecasting models. Some of these tools do a good job of estimating future demand of products, but all have one major problem; they can only work with the information presented to them. This is usually past usage history. How often have you heard someone say, “What you sold in the past is a good indication of what you will sell in the future”?

But, as we all have learned from experience, the future does not always (or even often) mirror the past. Product popularity and market conditions change over time. Predictions of these changes from customers, salespeople and other sources must be included in a comprehensive forecasting program. This is usually done by: 1) applying a percentage change to the results of a time-series forecast (e.g., “we think that sales of this product line in January will increase by 10%) or 2) predicting a quantity that will be purchased or used (e.g., we will use 100 pieces of the A100 widget per month).

But often these predictions are not accurate:


  • Customers often buy less than they say they will. Maybe some customers think that if they estimate they will use 100 of an item next month, then you will have the 20 pieces that they actually need.


  • Salespeople overestimate how much they will sell in the future. Are they trying to impress management with their predictions? And will they be directly affected if your warehouse is overstocked in the future?


  • Management may be overly optimistic about their company’s future growth. In over 30 years of consulting, I have only had five situations (out of over 2,500) where management predicted lower sales in the future.


Like any forecast, information obtained from customers, salespeople, management and other sources (typically called “collaborative” sources) will never be 100% accurate. But we can work to improve these predictions using a three step process: Record, Evaluate and Report.


  • Record – Record each “collaborative” prediction from each source of information.


  • Evaluate – Compare the actual sales or usage recorded in the month to the estimate from the source of supply.


  • Report – Report back to the source the actual sales or usage as well as accuracy of their prediction. Hopefully this will help them provide better forecast information in the future.


Not only will the Record – Evaluate – Report process help improve the accuracy of collaborative information, it will also help you identify what sources of information are more reliable than others. Next month we will explore more ideas for obtaining accurate collaborative information.


Post a comment

Your email address will not be published. Required fields are marked *