which typically looked at availability only at a particular
location. Furthermore, it may be important to look not just
at availability within a specific time bucket, but also at the
dynamics of future demand for a given location. If demand
at a specific location is expected to be high in the near
future, it may be better to source from a different location.
Companies want to make source-of-supply fulfillment
decisions based on a number of factors that create the most
profitable decision at a given point in time as well as over
the time horizon of their plan. For that reason, ATP, which
has been known to manufacturers for decades, will become
a big focus area in retail. In an omnichannel operation,
locations such as warehouses that previously dealt only with
dependent demand (calculated demand from downstream
warehouses and retail locations) will now also have to deal
with independent demand (end consumer demand from
the e-commerce channel). These forecasts will also be used
to drive labor planning at retail stores and other fulfillment
locations.
Demand planning, therefore, will increasingly be driven
by big data analysis. This includes shopper data from point-of-sale systems, and unstructured social media data gathered from various sources, including Twitter and Facebook.
Demand planning processes must be increasingly dynamic
and data-driven, using machine learning techniques. For
example, in the past, demand forecasting systems required
causal factors to be configured based on response models
established through offline analysis. In the future, machine
learning capabilities will dynamically analyze social media
data and provide an early warning of changing market conditions and shopper behavior, while automatically updating
the demand forecasting systems. Likewise, the analytics
described earlier can be used to understand a shopper’s
propensity to return certain merchandise—information
that can then be used to forecast returns.
5 Synchronize distribution planning, warehouse man- agement, transportation management, and store
operations. Fulfillment operations, including factory distribution, warehouses, transportation resources, stores,
and associated labor, have traditionally been planned and
executed in silos. This creates time latency between each
area, a situation that historically has resulted in the “
bullwhip effect,” a phenomenon in which upstream operations
become out of sync with downstream customer demand.
Successful omnichannel execution requires zero latency
across core fulfillment processes. Previously siloed func-
tions such as distribution, warehousing, and transportation
should become aware of upstream and downstream con-
straints. For example, distribution and transportation plans
should be aware of downstream warehouse dock, space,
and labor constraints. This prevents things like shipments
showing up at warehouses that don’t have the dock space
or the labor to process them, causing the warehouse to be
“overrun.” Synchronization of these operations provides
much more flexibility in responding to customer demand
for any product, anywhere, at any time.
6 Implement a distributed order and inventory man- agement system to provide a single view of orders and
inventory, and to execute orders across assets. Traditional
brick-and-mortar operations assort and forecast for physi-
cal store locations, and then generate replenishment orders
upstream into the supply chain. This results in orders
against warehouses, transportation, and ultimately factories
that produce the ordered goods. In addition, traditional
order management systems source from single points in the
supply chain. An order comes in and is immediately pegged
to a predetermined inventory location; if the inventory is
not there, it generates an ATP based on the next scheduled
availability of the inventory at that location. These systems
are not designed for sourcing from multiple locations or
to handle orders with multiple line items that must be
sourced from one location and delivered in different ways
(all together, separate, or grouped).
Distributed order management (DOM) and distributed
inventory management systems, by contrast, provide a sin-
gle view of inventory across multiple inventory management
systems. If a company has multiple divisions with different
order management systems, the DOM system can provide
a single ordering view to the customer, and can then gather
information and make decisions using the various back-
end systems. These solutions, which have been available
for about 15 years, are very useful in omnichannel, where
retailers need a consolidated view of orders across brick-
and-mortar and e-commerce nodes as well as the ability to
source orders from multiple inventory-stocking locations.
DOM systems must be configurable to handle a host
of complex scenarios, including sourcing from multiple
locations both within and outside of a supply chain (for
example, a partner’s products sold through a company’s
own website), as well as orders that are sourced from one
location but returned to another. In order to operate, there-
fore, the DOM system must be integrated with different
systems that are processing orders and managing inventory
along the supply chain, including warehouse management,
transportation management, and distribution planning.
And finally, a DOM system must also be armed with the
information and logic necessary to make profitable sourc-
ing decisions.
7Deploy next-generation profit-, constraint-, and allocation-based available-to-promise (PCA ATP).
