Skip to Content

Best Practices for Getting an Accurate Demand Forecast Post-Pandemic

This year has certainly been unpredictable, and as such industrial distributors have probably seen a dip in demand this year, particularly during the quarantine months. You may also now be experiencing a spike as things begin to normalize in areas of the country.   

The problem is that your inventory settings for demand forecast were set during a more normalized demand period, and now may not be allowing your forecast to respond quickly enough to the rapid changes. This could result in over or underbuying inventory.

How Your Forecast Is Derived – Gordon Graham Style

If you are using a forecasting buy method, that means you are using the demand forecast to help determine how much inventory to buy. Your item demand is tracked throughout the month, and at the beginning of each month, your historical demand is used to calculate a new forecast for the upcoming month.

Your forecasting method has settings that let you decide how far back in the the demand history you go to calculate a new forecast value. If you are familiar with the Gordon Graham method, some of the following terms will sound familiar.

The exponential smoothing method compares the demand for the month just completed (aka current demand) and compares it to your current forecast. A smoothing constant is used to put as much (or as little) weight on the current demand as you choose. If more weight is on the current demand, then your new forecast will be a number that reflects that recent demand. If more weight is on historical demand, then your new forecast will be a number that reflects historical demand.

More Review and Maintenance Needed During Economic Downturns

If your demand is pretty consistent over the long haul, then your forecast stays steady as well. However, when you have economic downturns or times of uncertainty, as we have this year, your demand can take a drastic downturn. This year, you may have seen 3-6 months of substantially lower demand than usual. A weighted average method of calculating a forecast may not respond to the changes quickly enough, depending on what your smoothing constant is set to. Adjusting the smoothing constant to focus more on the current demand helps your forecast to adjust more quickly, resulting in forecasts that more accurately reflect recent trends.

The same rule applies even if you are using a straight average of demand to calculate a new forecast. If you are looking too far back in history to get your average, for example 6-12 months, then you might be including demand before the pandemic in your forecast, which could result in an overstated forecast number. It might be beneficial to shorten that average, even if temporarily, so that your forecast can respond quicker.

Applying the Concepts

Let’s apply these concepts to a real-world example.

In  the chart above, you can see a healthy demand trend in the first quarter of the year, and then demand drops off in the second quarter. The third quarter is showing an uptick in demand, but still not near pre-pandemic numbers.

You don’t want January through March demand to be included in your current forecast because it will be overstating your average demand.

Do you want your forecast to cause you to buy 95 or 400? Based on the current demand trend, you want to be closer to that 100 mark, not 400.

If you use a setting in your software system that places focus on the numbers in the July/August/September range, for example, you’re going to be buying much more accurately. Additionally, you’ll be paying less in inventory taxes, and protecting yourself from a potential dead stock scenario later on. The bonus is that as the demand trends upward more, your forecast will continue to reflect that.

Other Calculation Methods

Jon Schreibfeder, President & Owner of Effective Inventory Management, has also recommended additional steps to take so you can temporarily modify your forecasts to compensate for the pandemic.  Along with the smoothing algorithm, he also suggests:

  • Create an inquiry to calculate the percentage difference between the forecast created for each item prior to the start of the pandemic, and then for actual sales/usage for the month. If your forecast for the month is 100 pieces, but you actually sold 75, the percentage difference would be “-25%”.
  • Reduce your forecast in your ERP system by the percentage determined above known as a “collaborative percentage, a prediction of how future demand will vary from what you sold in the past. This can be done directly in your ERP system if the system allows that measurement, or you will need to manually adjust.
  • Schreibfeder also says to adjust the resulting forecasts for other collaborative information from management, sales personnel, trade associations and customers.

About Us

If you’re a typical distributor, you have a complex mix of items in your warehouse, with a wide range of customer demand, lead times, cost, and other factors. You need a sophisticated software system to manage it all - that’s where TrulinX, by Tribute, Inc., comes into play! Tribute Inc. is a provider of industrial distribution management software with over 36 years of experience in the fluid power, motion control, industrial hose, fluid handling, pump, sealing, instrumentation & process control, and automation marketplace.

With TrulinX, you can group your items or product lines and manage them differently. Each group can have its own buying parameters, demand calculation, lead time calculation, and safety stock calculation for optimal inventory management. TrulinX offers several industry-standard calculation methods for buy quantity, forecasted lead time, and forecasted demand.

For more information, or to see a demo, contact us!

Connect with Tribute on Social Media

Talk with Tribute

Subscribe to our Newsletter

Sign Up Now

Stay up to date on the latest industry trends & learn more about industrial distribution & Tribute’s ERP software solutions. Check out our blog today!