The projection of
demand, to cover the cumulative planning lead time that is not
covered by (firm) customer orders, is the role of the forecast, be
it manual or computer based. Forecasting has always been the most
difficult application to implement effectively, even in those
businesses not impacted by fashion, taste or seasonal variations.
The traditional statistical intrinsic forecast required
modification by judgmental and extrinsic factors, or models that
allowed for the external factors. It is shortcomings in these areas
that usually degraded forecast effectiveness.
variety, choice, fashion, new retailing channels, promotions and
impact of competition have "destabilized" the forecast. The
influence of extremely large retailers, and their requirement for
compliance in a partnership, have invalidated much of the
forecasting logic of the 1970's. Today, a forecast has to be more
customer oriented (i.e., multiple projections of the same SKU or
group by customer). Extrinsic models and parameters must be more
easily assimilated, and directly managed. In some cases, the best
forecast is to take the projected sales plan (i.e., buyer's
forecast) directly into the manufacturer's planning and control
system. More emphasis must be given to monitoring, and modifying the
projection, on an ongoing basis.
The typical order
entry/control system is based on firm or confirmed orders for
discrete dates/quantities of a specific product, either as
individual orders or releases against a blanket order or contract.
There is a need to change such systems, to allow additional types of
order that can reflect the information from a partnership, and
afford the manufacturer a valid information base of demand by
customer. These additional types of orders could be described as the
Unconfirmed/projected orders for a specific product/ SKU
• Confirmed order of a product group (e.g. style) that will be later
amplified by specific identification of variants (e.g. color,
pattern, width, size)
• Unconfirmed/projected orders for a product group (e.g., style)
• Sample or demonstration orders that are not part of true (retail)
unconfirmed orders need to be consumed by the discrete confirmed
orders, just like a traditional forecast. The objective is to
provide a combination order/forecast that not only reserves
capacity, but also assists the user by being easier to convert into
a normal order.
Even with a
forecast in place, or with projected orders as described above, the
sales department is often required to commit to delivery promises
while on the phone. The representatives do not have the time to
confer with the manufacturing managers involved with the master
schedule. The traditional "available to promise" information,
typically by product group, is of only limited value in such cases.
It is usually periodic, forecast based, and therefore out of date in
Also, it may not
have considered the key resource or material constraints involved in
the final definition of product or variant (e.g.,
dyeing/finishing/printing in textiles, color/size in apparel).
systems can do better. Entry of customer request date/quantity,
explosion against key resources or materials, comparison with up to
the minute capacity or material availability can be performed
interactively, while still on the phone. Simulation of suggestions
for alternatives can also be provided, and negotiated with the
customer. Responsiveness is achieved, the customer is satisfied,
and manufacturing schedules have not been impacted by the misguided
guesses of the sales person. In other words, we employ the resource
requirements planning techniques of classical master scheduling on
an order by order and cumulative basis.
To Be Continued