In recent times, I’ve been
getting quite a few questions about Supply Chains. More specifically, the questions have
centered around how the Theory of Constraints addresses inventory levels and
the age old problem of stock-outs. So
today I’m going to start a discussion on TOC’s Distribution/Replenishment
solution, sometimes referred to as the Dynamic Replenishment Model. Before I discuss the solution, let’s take a
hard look at the problem to better understand why things are such a mess for many
companies.
How many times have you gone
into a grocery or department store wanting to purchase something specific? You search and search until you finally ask
someone where your item of interest is located.
This person leads you to where they should be, but when you get there,
you find that there are “stocked out” of you item. It happens often doesn’t it? If it’s a piece of clothing, then maybe they
have it, but not in your size. The
question is, why don’t stores keep the right level of stock to satisfy customer’s
needs? I mean they always seem to have a
lot of stock available, but many times not what you’re looking to
purchase. You wonder to yourself, why
can’t they seem to forecast correctly?
Think of all the sales they must lose.
Let’s try to better understand why these stock-outs occur so frequently.
In the first place,
distributors and manufacturers have very sophisticated forecasting software
designed to predict how many items (i.e. stock keeping units (SKUs)) they will
sell. But even with these sophisticated and
very expensive software programs (e.g. Enterprise Resources Planning (ERP)),
they still aren’t able to forecast how many units to send to the stores they
serve. They all seem to suffer from low
inventory turns, high inventory investment and stock-outs which of course
result in lost sales at some locations and high inventory at others. And if the inventory is too high and the
products aren't sold, then they also suffer from high inventory obsolescence and
customer dissatisfaction. These are all
real problems for stores.
Most supply chains today use
push systems and therein lies the
first of many problems in your typical supply chain. According to APICS, a push system in
distribution is a “system for replenishing field warehouse inventories where
replenishment decision making is centralized, decisions are usually made at the
manufacturing site or central supply facility.”
So with this definition in mind, the distributor has to decide how much
to keep upstream and how much to send downstream and the natural tendency is to
keep most of the stock as close to the consumer as possible. The rationale here is that by keeping it
close to the point of purchase, then there is a much larger chance that it will
be sold. But is this the correct
thinking?
One thing we know to be true
is that for in order for a push system to work effectively, you must have a
good forecasting model to be able to answer what
product to distribute, when to
distribute it and where to send
it. If you can’t accurately answer these
three questions, then your forecast will be incorrect and unfortunately the
forecasting models in place today don’t do so.
Amir Schragenheim, in Chapter 11 of the TOC Handbook, tells us there are
four statistical fallacies that exist which work against today’s forecasting
models as follows:
1.
The
fallacy of disaggregation.
Disaggregation simply means breaking up of a
total (aggregate),
integrated whole, or a conglomerate,
into smaller elements,
parts, or units,
usually for easier handling
or management. This fallacy has no impact on variation, but
it does have a negative impact on forecasts.
2.
The fallacy of the mean. This fallacy relates to the wrong interpretation
of data. Huge mistakes are made in
almost every organization because they don’t understand statistics. If, for example, the average demand for a
product was 20 units, do you think exactly 20 units would be sold in each store? The answer is no because of the next fallacy.
3.
The fallacy of the standard deviation. Many forecasting
algorithms present the data as an average demand without considering the
standard deviation. And one thing we
know is that the standard deviation is a measure of variation implying that the
average without knowledge of the standard deviation is meaningless. And as Amir rightfully points out, how many
people conceptually can estimate a standard deviation and determine its impact
on sales?
4.
The fallacy of sudden changes. Most forecasting methods can track changes in
demand, but when the change in demand is sudden, the accuracy of the forecast deteriorates.
As Amir points out, all four
of these fallacies severely impact the forecast of a single SKU and therefore
provide a very poor base for determining the required stock level of the SKU’s. So with this in mind, we need another way to
make stocking decisions. In my next
posting, we will look at the TOC Distribution and Replenishment Model and
discuss why it is a superior method to use.
Bob Sproull
No comments:
Post a Comment