(By Mike Raftery)
Ah…. the soft warm blanket of safety stock. Who doesn’t love an extra pallet or two of inventory to help you sleep well at night? In the theory of every supply chain textbook ever written, safety stock is purely the amount of inventory required to prevent stockouts. This has a loose interpretation of meaning ‘lots of inventory is good’. Or even worse, a tool for modeling a predefined prescriptive production plan. But safety stock is a statistical calculation, pure hard math to hedge against one thing and one thing only: variability. In this, part 1 of a semi-regular series, I’ll cover the dangers of straying from this theory of inventory management to compensate for variability.
Variability is the Enemy
One of best supply chain professors learned his craft from the Army Corps of Engineers (Hoo Rah!). Throughout every day of every class, he would pound the desk and shout “Variability is the enemy!” I thought it was funny, an attention grabber for the class. But even today, some number of years later, I still hear those words in a drill sergeant tenor whenever I’m in a meeting about safety stock. Variability is the enemy, and safety stock is the countermeasure. Given how this was drilled into my supply chain brain in these formative years, I find it appalling how many companies I see with a bastardized version of safety stock to serve multiple purposes. Make no mistake, safety stock is meant to protect against variability and nothing else.
I’ve seen safety stock used to drive production, model seasonal demand and represent capacity. It may just be my opinion but any use of safety stock for these purposes is just flat wrong. When I see these symptoms in the design of any supply chain system, it tells me that whoever designed this has really lost control of modeling basic supply chain concepts. Once you start manipulating the safety stock target to meet any needs other than hedging against variability, you have surrendered all pretenses of a real inventory strategy. At this point, you’re really saying that you’re smarter than math and might as well just key orders into any planning system manually.
Variability. Is. The. Enemy.
The Enemy of My Enemy is My Friend.
Statistics… ugh. Outside of you know, nerds, everyone hates numbers, specifically statistics. But this is where numbers, shockingly, can actually be useful. If you accept the concept that safety stock’s sole purpose is to protect your supply chain from variability, then it’s not a big leap to think that “there has to be some math that can help quantify this.” Good news, there is. Stats, specifically your friendly neighborhood Z value can do wonders to protect against variability.
Despite many examples of tribal knowledge, there is a scientific way to quantify the amount of inventory needed to protect a node in the supply chain against stockouts. You can quantify these things with historical data. And use the infamous Z value to compensate for highly variable nodes, or minimize investment in inventory for rock solid nodes in the supply chain.
A Z value goes to the ol’ bell curve. How many sigmas of certainty you can afford is a function of the Z value, along with its friend, the standard deviation of course. At the highest level, the optimal safety stock is the following equation:
Safety stock = Z value * (std deviation) * average demand
Of course, this is way too simplistic to be of any value in real life, and I’ll get a bit deeper into the calculations that drive a usable safety stock value in future articles. But it does help to conceptualize the relationship between the real variability (std deviation), volume of demand (avg demand) and required service level (Z value). The interplay between these factors derive the actual inventory required to prevent stockouts. There are nuances around what constitutes real demand, periods of evaluation and how to calculate real variability. But at its core, anything above and beyond this calculation is a luxury, just extra inventory to meet another purpose or make people feel comfortable.
Fight the Enemy Where They Live.
You can’t lower safety stock in a fair fight, 1:1; any movement to this needle is a proxy war. Remember, variability is the enemy. Safety stock is the objective. So lower safety stock by reducing variability wherever you find it. Long lead times can be tolerated if they are predicted. Length becomes less important as long as it is predictable. This is kind of a new concept at first. It requires looking at a supply chain in relative terms of predictability, rather than in absolute terms of cases, pallets and distance.
So where to invest to fight that enemy of variability? Invest in asking the right questions and segmenting the drivers of your inventory issues. Not all nodes of the supply chain are equal. Everyone knows the “good plant” and the “black sheep” of the supply chain. An increased focus on improving the reliability of the black sheep might reduce risk to all the nodes in your network more effectively than polishing the trophies of the golden boy.
Hope is not a strategy. You can’t just want inventory to be lower, wish risk to go away, or hope you catch a break. Reality is a harsh mistress and the only way to lower inventory with confidence is to lower the inputs to safety stock. Reducing variability in transportation, production and demand are the only levers you have to truly reduce inventory without adding risk unnecessary. Anything else is just crossing your fingers.
Reasonable safety stock levels that result in zero outages is not the goal, nor is it practical. It will result in an inventory strategy that meets the service level expectations of your supply chain while minimizing those pallets of dollars sitting on the racks on your warehouse. This is the black and white of it. You have customer service expectations, a finite number of dollars and real world performance metrics. With this triad, something has to give. Most customers I know are not super tolerant of constant disappointment. And people hate spending more money than they have to (exceptions, please contact me immediately). That leaves us to the real world metrics that drive an inventory strategy.
Such a critical metric as safety stock would seem to have zero tolerance to screw up. This, however, is not the case. I have seen too many people treat safety stock targets as light suggestions, or an input to flatly ignore that I’m starting to believe there’s a real conspiracy from pallet manufacturers to increase inventory across the globe. Safety stock is not a production plan, it’s not a capacity plan, and it’s not a seasonal prebuild. It’s the cold hard result of an equation you learned in stats class. Anything else is a waste of money.
Stay tuned for part 2 of this series.