15 March 2012

Increased supply chain confidence through simplicity

Traditional approaches to inventory management and replenishment divide inventory stocks into two portions:

  1. Working stock – the inventories designed to cover daily demand
  2. Safety stock – the inventory quantities designed to cover variation in supply or demand or both

ToC Distr Trad IM View

Years of statistical analytics and software development have been focused on improving the ways in which lead-time, demand and safety stock values are calculated. So much, in fact, that most of the people who use supply chain management, inventory management, or replenishment software frequently do not even understand what the software is doing, how it is doing it, or why it works or does not work.

Some years ago I was consulting a firm and, in the course of the business, reviewing how they went about their inventory management and replenishment. They had software that did inventory management and that included replenishment calculations.

So, we were sitting together and he was describing to me what he was doing on his computer. He said, “Here’s the ordering screen. It shows historical demand here [pointing], and the recommended order quantity here [again, pointing]. And, I don’t know exactly what this number is for [pointing], but if I think the system is suggesting that I buy too much or two little, I can adjust this number until the suggested order quantity lines up with what I think it ought to be.”

Well, of course, what the system was doing was exponential-smoothing of demand and the value he was adjusting was the value of alpha in the formula.

What I refrained from asking him (only by biting my tongue) was, “If you are going to simply adjust the system’s findings to your intuition, why use the system at all?”

The moral is: Systems that are not understood—and most complex systems are not understood—are also not trusted. Especially if they frequently—or even, regularly—produce what are perceived to be unreliable results.

The artificial divide

The artificial subdividing of stock quantities into “working stock” versus “safety stock,” and adding complexities around the factors used to calculate the one value versus the other provides no added value. In fact, the complexity actually leads to less reliability because the users frequently do not know how to set the input parameters effectively. Not to mention the fact that the parameters that are effective today may not—in fact, likely will not—be effective tomorrow or next week.

The fact of the matter is, in most cases, the only awareness of the division between “working stock” and “safety stock” quantities is found in the software itself and those that may be intimately acquainted with the software and its configuration. The people on the warehouse floor typically do not know when they have made an incursion into “safety stock.” They don’t know that the first 41 units they picked for order number 8789089 were from “working stock,” and the last nine units were taken from “safety stock.” And, they should not care.

Even the managers frequently have no visual signal that an incursion has been made into “safety stock.”

Inherent simplicity

ToC Distr DBM IM View

Employing Theory of Constraints (ToC) Dynamic Buffer Management (DBM) makes life easier to understand for those responsible for inventory management and replenishment (read: supply chain managers). The buffer size (for any given item in any given stocking location) is a single number. (Let’s say, 1,000 units.)

The formula for setting the initial buffer size is simple and easily understood. Typically that formula is something like this:

Initial Buffer Qty = [Average Daily Demand] * [ToC Replenishment Days] * [2] * [Paranoia Factor]

The only factor that really needs any kind of explanation is the “Paranoia Factor.” This is merely a multiplier selected by intuition and based on senses of the criticality of an item. An item might be critical because it is used in the production of 800 other items; or because the majority of your customers all buy this item; or because one hugely important customer relies upon you for this item; or dozens of other reasons.

Once the initial buffer size has been calculated and set, the buffer is divided (mathematically) into three “zones.” The top third is called the green zone, the middle third is called the yellow zone, and the bottom third is called the red zone.

Going forward, the DBM system simply monitors for conditions at each replenishment cycle and adjusts the buffer size according to rules. The rules are typically:

  1. Too Much Green – The item has been found in the green zone on three consecutive replenishment cycles; therefore, reduce the buffer size by one-third.
  2. Too Much Red – The item has been found in the red zone on two consecutive replenishment cycles; therefore, increase the buffer size by one-third.

It’s that simple. No complex formulas for calculating and managing variability in demand or supply.

On top of that, supply chain managers can have simple visual signals as to the status of their buffers. A simple view of the inventory data (by location) can readily provide red light, yellow light, and green light indicators for the buffer status in any stocking location for any item. No math and easy to equate to action:

  • Green light – no action required
  • Yellow light – take note, perhaps investigate critical factors like larger-than-normal orders or orders pending for critical customers
  • Red light – consider expediting measures, if necessary

NOTE: There are more options available with DBM, such as identifying and managing SDCs (sudden demand change items—like seasonality), managing Virtual Buffers (between stocking locations, such as warehouse-to-warehouse replenishment, or broader supply chain visibility and collaboration). It is not the intent of this article to exhaust the applicability of DBM.

RKL eSolutions, LLC is in the process building a cloud-based solution to help you manage your inventory in just such a way—using Dynamic Buffer Management and the Theory of Constraints. Contact me or fill out the contact form here if you would like more information.

12 March 2012

The biggest supply chain management mistake over the last 30 years?

I would have to say that the biggest mistake made in SCM over the last 30 (or more) years is the industry’s reliance upon forecasting.

  1. Forecasts are virtually always wrong. They may be wrong by a little bit, or they may be wrong by a lot. But they are--for all practical purposes--always wrong. The forecast may be wrong and you have too much inventory--which your firm may call "good' ("Great job! We didn't have an out-of-stock.") or it may call it "bad" ("Hey! Wake up! We are holding too much inventory!"). The forecast may also be wrong and you have too little inventory, which (again) management may call either "good" ("Great job! We sold out of that!") or "bad" ("Hey! Wake up! We lost sales on that because we ran out of stock!").
  2. Forecasts only lead to one of two conditions: over-stocks and out-of-stocks.
  3. Forecasts offer no assurances of being responsive to the market.

Personally, I believe that if the industry had spent as much time, effort and money on increasing replenishment frequency (reducing lead-time), improving supply chain visibility (end-to-end), making inventory management more agile (providing rapid response to changes in end-user demand) and better understanding and management of sudden demand changes (seasonality and similar events) there'd be a more sales, lower prices, reduced obsolescence and happier supply chain managers everywhere today.

Replenishment frequency

Both Lean and Theory of Constraints management have certainly taught us that replenishment cycles should be as short as possible. One-for-one replenishment is ideal. But short of that, daily is better than weekly; weekly is better than every two weeks; and so forth. When the costs of obsolescence, lost sales, lost customers (due to lost sales), marketing costs required to recover for lost customers, and the many other costs associated with out-of-stocks (on the most popular times) and over-stocks (on the "dogs") if find it hard to believe that most organizations would not perform better with more agile suppliers and logistics even if the so-called "cost of goods" might be marginally higher. Correct valuation of Throughput certainly should teach us that lesson in many, many cases.

End-to-end supply chain visibility

One of the things wrong with today's supply chain is that the manufacturers actually believe that they have made a "sale" when then they sell the product to the distributor. In turn, the distributors believe that they have made a sale when they unload some product on a wholesaler--and so forth on down the supply chain.

The truth is, until the end-user has made a purchase, all the other "sales" have simply put inventory into the supply chain. Inventory that will become obsolete or eat demand for newly-introduced products when liquidated at "discounted" prices. Either way, it's bad for profits in the supply chain.

Imagine how much better it would be if the manufacture (in Malaysia, or wherever) knew within 24 hours precisely how many finished goods were being purchased by end-users every single day. They would know how to pace their production and manage their inventory buffers--as would everyone else in the supply chain!

Inventory management agility

Instead of setting inventory policy once a year, or even several times a year, systems should dynamically adjust for changes in demand (via supply chain visibility) constantly. And, instead of complexity and hard-to-understand formulas, inventory managers should be able to respond to simple visual signals indicating the condition of inventory in their direct control--as well as signals coming from across the supply chain.

Managing sudden demand changes

Supply chain systems should be able to rapidly analyze historical data and identify SDC (sudden demand change) items by simple rules. The systems should then help the supply chain managers understand how to manage build-ups and build-downs for SDC items based on the supply chain production capacities for each item or group of items.

Personally, I think time, energy and money spent in these areas--some of which is now happening--would do a "world" of good (pun intended).

What do you think?