Mastering Forecast Accuracy: Conquer Artificial Volatility for Supply Chain Excellence

This is an excerpt from my new book, Lean Forecasting Demystified. If you find this valuable, please click on this link to place a preorder on Amazon.

In 2004, director James Cameron said “Hope is not a strategy” when addressing the NASA Administrator's Symposium. He was referring to his exploration of the Titanic wreck to which he has made thirty-three dives. Suffice to say “hope” did not play a large part in his planning process. In fact, his engineering team spent seven years building the submersible capable of performing this feat.

I mention this anecdote, because I have worked in many organizations that create arbitrary targets for forecast accuracy. Management sets these targets and insists that forecast accuracy gets better. They do not seem to realize that there is a strong inverse relationship between the volatility of a demand pattern and our ability to forecast accurately. As the diagram below shows, more demand volatility results in lower forecast accuracy.

Demand Volatility

Demand volatility is typically measured using the Coefficient of Variation (CV). It indicates whether a demand pattern is stable, staying close to its average value, or is highly unpredictable. CV is simply the standard deviation of demand divided by average demand. If only we could reduce volatility, we could improve forecast accuracy. You do not necessarily need to live with the current volatility level.?

However, I have good news – not all volatility is necessary. It is important to understand that there are two types of volatility: inherent and artificial.

?Inherent volatility is the natural variation of demand by consumers, and we typically have no control over this. However, artificial volatility is when shipments throughout the supply chain are more erratic than the inherent variation. For example, when a manufacturer ships its products to a distributor, they typically have a much higher CV than the inherent volatility. Here are some explanations:

• Shipping cost reduction – larger loads are less expensive per unit. For bulky items, this could be a major consideration.

Potential solution

If your organization has several customers in the same region, then change the sales terms so that you become responsible for freight costs, and consolidate shipments to the customers in this region.?

• Volume Discounts – it may be that your sales team is incentivizing the customer by giving them volume discounts.

Potential solution

Accept the large orders, but deliver in multiple shipments to avoid the large spike in demand created by one large shipment.

• Minimum Order Quantities – your manufacturing operations have large set up costs to switch from one product to another. Manufacturing managers will want large batch sizes to reduce costs. In this case, an organization creates demand variation internally, rather than externally created by the customer.

Potential solution

Adopt lean practices on the manufacturing floor to reduce changeover time between production runs to smaller batch sizes to become economical.

• Promotions – many organizations shoot themselves in the foot, by creating additional volatility with incentive programs driven by management. Their demand-generation activities create a spike in demand.?

Potential solution

Educate management on the additional cost created by these demand spikes. Then they can weigh this information with the perceived revenue increases resulting from these promotional campaigns. I say “perceived” because promotions could be merely changing the timing of demand, with no net increase.

The message is not to merely setting arbitrary targets and hoping that forecast accuracy will somehow magically improve. Instead, finding ways to reduce artificial volatility will move the needle!

If you found this article valuable, please click on this link to place a preorder for Lean Forecasting Demystified on Amazon.

To learn more about how to transform your demand forecasting performance, please contact me at [email protected]