If you manufacture or distribute products, it's safe to assume that you have a specified safety stock level— or at least we certainly hope you do. If you're on the fence about whether or not to hold safety stock, we recommend reading up on why it's critical to overcoming variability in your supply chain.
At this point, you may be saying to yourself, "but carrying excess inventory is expensive!" and you would be right. However, this is exactly why some of the best-run and most profitable businesses are so focused on holding the optimal amount of safety stock.
For those of you who do hold safety stock: how do you calculate it? Do you do it statistically, or is it a "rule of thumb"?
Do you use a number given to you by someone else in your organization? If so, do you know how they calculated it, and which formula they used? If you generated the number yourself, did you rely on a textbook formula that you may not fully understand? You're not alone.
The goal of this article isn't to get into the intricacies or mathematics of calculating safety stock, but rather to highlight that this is a very complex endeavor, not to be taken lightly. A business that fails to understand the logic underpinning their safety stock calculation is at risk of being exposed to unnecessary cost and risk.
With that in mind, let's look a little more closely at how companies determine their safety stock levels.
The Dangers of "Rule of Thumb" Safety Stock Planning
The primary purpose of safety stock is to buffer the unexpected and unanticipated events that impact our supply chains. Yes, we're talking about forecast error. As you know, (and as we've discussed in previous posts) demand forecast accuracy is, arguably, the single biggest factor that impacts the ability to meet customer service targets.
Why, then, are so many businesses still using a "rule of thumb" method for calculating safety stock?
If your business is one of them, what is the basis for your "rule of thumb" metric?
The ideal way to set safety stock levels is to calculate them statistically based on previous forecasts. Safety stock should be driven by your forecast error, which isn't a one-size-fits-all proposition. It's a number that should be measured at the item and location level—information that many companies don't have access to, which then forces them to use a general of rule of thumb model to compensate.
This isn't to say that this kind of approach should be overlooked, as there are some that truly do provide value to a business. But let's be frank: a "rule of thumb" approach is certainly not the best or most reliable method for calculating safety stock in today's challenging business environment—especially when it can have such a dramatic impact on your business.
If your business is still using rule of thumb metrics for safety stock, you're likely carrying too much inventory and possibly underperforming on your customer service targets.
Dealing with Lumpy Real-World Demand Patterns
The companies that are a few steps ahead of the game are the ones who calculate safety stock based on a carefully measured and tracked forecast error, with the crux of this argument being that they have a comprehensive understanding of the formula that's being used for said calculations. Without this understanding, businesses could still be exposing themselves to unanticipated and undesirable results.
Without reliving college statistics class too much, let's talk about Normal Distribution and how it relates to safety stock calculations.
You see, most safety stock calculations are driven by the assumption that the pattern of demand follows a Normal Distribution— that familiar bell-shaped graph from Statistics 101. As the bell curve gets narrower, dispersion is less, and less safety stock is required for a given service level target. As the curve gets wider and flatter, dispersion is greater, and more safety stock is required.
If a business's demand follows a Normal Distribution and the data points are independent and identically distributed (IID), then calculating safety stock is quite easy. We can use well-documented formulas for calculating safety stock based on the standard deviation of the probability distribution. This is what APICS recommends, and what most package software programs use for calculating their statistical safety stock levels.
But what if your demand pattern doesn't follow a Normal Distribution? What if your data is not independent and/or identically distributed? We hate to be the bearers of bad news, but in that case, the formula that most businesses use to calculate safety stock is quite simply incorrect.
In the real world, many businesses will find that a lot of their items probably exhibit sporadic or "lumpy" demand patterns that are anything but "normally" distributed. For these items, you can't blindly use the same formula for calculating safety stock levels. If you do, the basic assumption of your model will not be true—and results will be inaccurate.
How to Plan Around Not-So-Normal Distribution
To compensate for the reality that your business is complex and that the demand for many of your items doesn't fit neatly into a Normal Distribution, you need software that recognizes that more than one approach is needed for calculating safety stock. At Demand Solutions, we offer a variety of safety stock formulae, including proprietary methods that have proven effective in helping companies hit their customer service goals while minimizing inventory investments.
You've been tasked with the tremendous responsibility of setting intelligent customer service targets and then executing safety stock strategies that enable you to reach them. Whatever you do, don't succumb to the naïve assumption that every item's demand pattern matches a mathematical Normal Distribution. Instead, use a sophisticated engine to calculate safety stock—an engine that allows you to account for the imperfections of the real world.
Keep an eye out for our next post in this series on safety stock, where we'll discuss two common approaches to safety stock planning and weigh the merits of each.