In a preceding article, I talked about the complexities that many methods and applications bring to calculating what stocking quantities should be for each SKU-Location, or SKUL. I also pointed out that, in a great many cases, all that complexity ultimately gets trumped by the ultimate in simplicity.
For example, a $2.5 million ERP system may have calculated that stocking levels for a particular product line in the firm’s warehouse in Seattle ought to be X. But, I can pretty much assure you that, if the executive vice-president for sales and marketing just had a key customer disappointed (or, worse, lost) because a handful of those products were out-of-stock in Seattle last week, all the complexity of calculations in that costly ERP system will be overridden by the simplistic ranting and raving of the VP of sales, and stocking levels will increase for those particular products in the Seattle facility.
We also talked about how, through complexity, stock quantities and the underlying calculations are divided and subdivided. Stock quantities are divided between working stock and safety stock. Calculations and averages are taken of several separate components, including actual demand, forecast demand, lead times, and more.
What it Really Boils Down to…
What it all really boils down to, however, is this: we need enough stock on-hand to cover what we expect to consume (through sales or other consumption) between replenishments. We also need to account for variables like unusually high demand or delays in replenishment.
Now, it is true: we can take the complex route to figuring out what that number should be, but why do we need complex algorithms and costly applications to help us precisely calculate a number that, in the end, will (almost always) be wrong anyway. The number might be wrong because it’s too high; or it might be wrong because it’s too low. It might be wrong by just a little bit; or it might be wrong by the proverbial “country mile.” But, statistics will show that the number is wrong far, far more times than it is precisely correct.
So, again I ask: why do we pay the big bucks to buy computer systems that run algorithms we don’t understand to calculate very precisely a number that is going to be wrong most of the time?
What we really need is this: to be approximately right, not precisely wrong!
Here is a method that employs the concept of inherent simplicity to help you calculate a beginning stocking level that is approximately right. Use this simple formula:
Beginning Buffer Size =
[Avg. Daily Demand] * [Replenishment Cycle Days] * [Paranoia Factor]
What’s the “Paranoia Factor”?
The “Paranoia Factor” (PF) is a number that helps you cover all of those “other factors” that cannot be quantified or programmed into an algorithm. It leverages your firm’s “tribal knowledge” and the intuition of all the bright folks you have hired—from executives to sales to marketing to inventory management and beyond.
In short, it is the factor that helps you get to approximately right without that $2.5 million ERP system (mentioned above) being trumped by someone’s rants.
What are some of the likely contributors to the “Paranoia Factors” you might apply on a SKUL by SKUL basis? Here’s a suggestive list:
-
How important is the SKU to profits?
-
How important is the SKU to the sales of other products (product affinities)?
-
How important are the customers who rely on this product?
-
How reliable is the vendor for this product?
-
How reliable are the transportation and logistics channels for this product?
-
Do we have an alternate source for this product if the primary vendor drops the ball?
-
Do we sometimes go a long time without any orders for this item, but then get one or two big orders in a short period of time?
A Good Starting Place
Here is a good starting place for PFs: for retail outlets, start with a PF of 2.0. For distributors and wholesalers, where demand is somewhat aggregated already, start with a PF of 1.5.
Also, if a SKUL shows erratic behavior (as suggested by number seven in the list above), and other factors suggest that it is worth your supporting the larger inventory, consider substituting the average order size over the last year or the statistical mode order size over the last twelve months times the PF, rather than average daily demand.
Remember! This helps you calculate a beginning buffer size (stock quantities for each stocking location). You still need a method to adjust buffer sizes over time as things change.
If you have comments or questions, please leave them here, or feel free to contact us directly.
