Forecasting Seasonality for Efficient Distribution

For distribution companies, learning to distinguish between spikes and seasonality is one of the hardest things to do. Even Gordon Graham, considered by many to be the guru of forecasting, never quite nailed down a surefire way to sort the sporadic from the seasonal. His solution? After applying a simple formula to flag the potentially seasonal items, sift through the product history manually to sort out the seasonal trends from spikes in demand. Right about now you’re probably estimating how many product records that would be for your company, and how much time it would take to review them all. Obviously, that’s not a practical solution.

The good news is that a lot has changed since 1996, when Graham retired. He couldn’t have imagined the solutions on the horizon, such as Microsoft Dynamics NAV, GP, and AX. In fact, his early work set the stage for the forecasting tools available today.

True seasonality is the Holy Grail of forecasting — once you’ve nailed it down, you’ve got the best predictor of future orders you can get this side of a time machine. The larger the percentage of your sales that stem from seasonal and intermittent large purchases, the more valuable it is to be able to separate one from the other. When a trend is accurately identified as seasonal, you’ll know when you can allow the inventory levels to drop low (or to zero), and when to build levels back up in anticipation of the next wave of demand.

Graham’s formula was a simple one; he said that an item is seasonal if 80% of its sales over a year’s time take place in just six months. This isn’t a bad formula, but the data needs to be run through a few more tests to separate out spikes from other factors.

For example, a distributor may see an uptick in baseball mitts in July. Common sense suggests that yes, baseball mitts are a seasonal item, but forecasting software plugs in a series of formulas to test that assumption. In this case, the forecast agrees that the mitts are seasonal in general, but that the July spike was unusually high. A quick phone call reveals that the local team sponsored a “free mitt” day last year, but they have no plans to repeat the event.

Some cases are more straightforward, like the demand for cut evergreen trees in December or American flags leading up to Independence Day. But identifying seasonal trends can also be tricky, as illustrated in that example. Just because an item is seasonal doesn’t always mean that there aren’t other factors influencing demand as well. That’s why a multitude of formulas is employed, parsing your sales history in a number of different ways to make sure you don’t order for a demand which won’t be there, or get caught short when it turns out that the demand was seasonal, after all.  While there will always be a few cases that require a human judgment call, those cases should be a mere handful, not number in the thousands.

A lot of guesswork is taken out by thoughtful customization of your forecasting solution. Many formulas have been developed with specific industries in mind, and choosing the right ones will result in more efficient inventory management. Having a powerful forecasting solution will also improve the value of business intelligence and analytics that are essential for maintaining your competitive edge.

Interested in learning more about how to incorporate substantive, relevant forecasting into your business management solution? Contact OmniVue to see if you’re eligible for VueFinderTM, our unique approach to help distribution companies uncover and solve fundamental business challenges.

Author Jeff Pyden

Jeff Pyden is President and CEO of OmniVue, which he founded in 2003. For more than 20 years, Jeff has been following his passion for improving business execution and efficiency through technology.