Flowcasting The Retail Supply Chain PDF

0 introduction 4/4/06 1:00 PM Page 1
Flowcasting: Introduction
I N T R O D U C T I O N
Winning the Retail

Supply Chain Race
➫
I magine that you’re a race car driver, and you’re minutes away from
the most important event of your Formula 1 racing career. Your
world class pit crew has prepped your car from front to end, and
everything in between. All systems are go. Your team pushes your car
into place at the starting line, and you climb in. The steering wheel
is locked into place. The countdown begins, and the first starting
light flashes. Then the second, followed by the third, fourth, and
finally the fifth. The lights simultaneously flicker off indicating the
start of the race. Your engine roars, and then you discover something
terrible: the steering wheel isn’t connected to anything, and you’re
out of the race.
This is an apt analogy for what happens in retail supply chains
today. The retailer may have great wholesale and manufacturing
partners, but if the retail store is disconnected from its trading part-
ners, the retail supply chain will never win the race to cut costs while
offering the best customer service. Instead, the retail supply chain
will function like the race car with the disconnected steering wheel -
- it will have tremendous brute force to push forward, but it will have
little ability to steer a course, react to changing "road" conditions, or
carry out a well-executed plan.
Because of this disconnect, the retail store is often considered
the weakest link in the retail supply chain -- a notion supported by
19
numerous surveys conducted over the past 10 to 15 years regarding

out-of-stocks. Retail store out-of-stocks (usually in the 5 percent to 8
percent range) are indeed much worse than the percentage of out-of-
stocks that occur elsewhere across retail supply chains. As you’ll
learn in this book, the nodes in a retail supply chain are highly inter-
dependent. And when they’re managed accordingly, the benefits that
accrue to the trading partner are staggering in terms of reducing
costs and improving customer service -- two keys in winning today’s
retail race.
Steering Linkage for the Retail Supply Chain

If you work within a manufacturing organization, you’ve probably
heard of DRP (Distribution Resource Planning) and may deploy it in
your company right now. DRP is a means of ensuring that goods pro-
duced at the manufacturing end of a supply chain reach the retail end
in the most cost-effective way possible. The results from DRP, how-
ever, will be only as good as the ability to forecast what retail stores
will actually need for routine and promotional sales activity. If you
can’t accurately forecast consumer demand, you’ll need to rely on
safety stock as a hedge against uncertainty. Whereas in the "good old
days" inventory and capacity were cheap, today, no company can be
competitive if a substantial portion of its inventory is held in limbo
in case the forecast is incorrect. The only way to truly remove uncer-
tainty about what you’ll sell across a given retail supply chain is to
forecast at the store level. Moreover, forecasting at the retail store
level "drives" the entire retail supply chain; all plans for every trading
partner can be derived from that single forecast.
Now wait a minute. According to conventional wisdom, you
can’t forecast at the store level! That’s correct, as far as yesterday’s
conventional wisdom goes and before the advent of a revolutionary
new concept called Flowcasting1. Today’s DRP software packages,
which do a superb job of planning for distribution and logistics man-
agement, are designed for manufacturing, not store-level, volumes.
1
Flowcasting is a new concept and is derived from the DRP Process created and
implemented by André Martin at Abbott Laboratories in the mid 1970’s
20
Software packages for Flowcasting, in contrast, are specifically

designed for forecasting, inventory management and replenishment,
and driving the retail supply chain from the point of sale back, elim-
inating all other levels of uncertainty.
Flowcasting is a multi-echelon retail inventory management
business process designed to 1) forecast products at the retail store
level and 2) plan inventory, replenishment, people requirements,
space, and equipment resources throughout the retail supply chain
in a time-phased manner. Flowcasting continually monitors and
controls the flow of inventory from beginning (the store) to end (the
factory) -- the entire trading network. As inventory at store level
increases or deceases above or below a set level, Flowcasting auto-
matically recommends the adjustment of the flow and level of inven-
tory across the Distribution Centers (DCs) and factories that service
the store. The result is a balanced trading network, the elimination
of shortages, increased sales, reductions in supply chain operating
costs, and greatly increased inventory velocity (fewer inventories,
greater turns).
The Flowcasting business process works as follows:

1. Flowcasting generates a baseline forecast that extends a year
into the future for every item in every store.
2. Flowcasting allocates total planned promotion quantities to
participating stores, thereby creating a total forecast of con-
sumer demand.
3. Flowcasting takes the total forecast, deducts on hand inven-
tories, considers on order quantities, applies the retailer’s
business rules, and creates a 52-week, time-phased model of
how much product should flow into both retail stores and
Retail Distribution Centers (RDCs), and when the product
flow will occur.
4. Flowcasting calculates how much inventory will be necessary
across all other nodes in the retail supply chain to support
the total store and RDC’s level forecasts every day for the next
year.
21
5. Flowcasting enables the communication of replenishment

schedules for both retail stores and distribution centers to
wholesale and manufacturing trading partners over a 52-
week planning horizon.
6. Flowcasting plans all required people, space, equipment, and
capital resources necessary to acquire, transport, store, and
deliver products from the final point of manufacture to the
final point of sale.
The implications for retail supply chains are significant.

Flowcasting not only eliminates the need for suppliers to forecast
their retail trading partners’ needs, but it ensures that the entire
retail supply chain will be refreshed and automatically resynchro-
nized on a daily basis, depending on whether sales at store level come
in under or over forecast. In doing so, Flowcasting generates a
detailed and realistic model of inventory based on the store-level
forecasts of consumer demand. This model factors in all retailer spe-
cific constraints, schedules and rules, such as store shelf capacity,
minimum shelf displays (including safety stock), minimum ship
quantities to stores and DCs, transit lead-times between nodes, and
shipping schedules.
The Flowcasting business process is made possible by several
developments: advances in forecasting techniques that marry intrin-
sic and extrinsic forecasting at store level; the availability of new
DRP-based, time-phased planning logic that can crunch and model
retail store volumes on low-cost hardware; and advanced networking
technology that enables trading partners to communicate on a level
playing field. Of the many benefits that Flowcasting provides retail-
ers, the following are particularly compelling:
1. Retailers and manufacturers can "flowcast," rather

than forecast.
Flowcasting is a business process that gives retail supply chain trad-
ing partners the ability to model the stocks and flows of inventory, in
a time-phased manner, from factories to final points of sale.
Consequently, sales forecasting and a myriad of other forecasting
22
activities are completely eliminated at every node of a given supply

chain except the only place where it matters: the final point of sale.
The end result is the creation of a model of the business inside the
computer and a totally synchronized retail supply chain aligned daily
to pure end user demand. This process, upon reaching critical mass,
enables manufacturers and their trading partners to transition from
a make and ship-to-stock environment to a make and ship-to-order
environment. Resulting benefits from this process are greatly
reduced store out-of-stocks, inventories and cost reductions for all
trading partners, ranging from 1 percent to 6 percent of sales
depending on the company.
2. Better planning and execution of promotions.

Promotions may be the lifeblood of most retailers today, but they fre-
quently become cost-intensive exercises in futility. Too often, manu-
facturing operations ramp up production in anticipation of a promo-
tion -- in effect inventorying their capacity -- then deploy the inven-
tory to their distribution networks where it sits until the retailers
actually order. Since there’s no visibility into the retail side of the
supply chain (that is, into what and when trading partners will
order), this strategy is akin to rolling the dice at Las Vegas or Monte
Carlo. Maybe 50,000 cases of product wind up in the Chicago or
Rome distribution center, but only 40,000 are actually needed.
London and Milwaukee need another 10,000 cases, so the company
ships the 10,000 cases from Chicago or Rome, incurring unnecessary
added costs.
On the retail side, the situation is likely worse, because buyers,
category managers, and planners are thinking about how many items
they can sell in their 3,000 locations rather than about how many
items they can sell in a given promotion. The result is that some
stores wind up with excess inventory and the associated carrying and
return costs, while others incur lost sales due to insufficient inven-
tory.
In a Flowcasting environment, all of these problems vanish.
Guessing and blind spots give way to a synchronization of the flow of
inventory across the entire retail supply chain. This, in turn, makes
23
it possible to plan and execute promotions that are profitable for all
trading partners involved in the exercise.
3. A common language for all trading partners in the supply chain.

Traditionally, entities involved in creating, distributing, and selling
goods have been nations unto themselves, each speaking a different
mother tongue. MRPII and DRP each eliminated a few floors of this
"Tower of Babel," and ERP further promoted consistency throughout
the manufacturing side of supply chains. But several key disconnects
among trading partners still exist on the retail side. Disconnects
occur between store inventory replenishment systems, RDC replen-
ishment systems, merchandising systems, and trading partners
replenishment systems.
Flowcasting resolves these major disconnects and introduces a
means of consistent communication for all trading partners, one that
enables them to easily share required people, space, equipment, and
capital resources necessary to acquire, transport, store, and deliver
products. Flowcasting makes anticipated people, space, and equip-
ment constraints highly visible to all supply chain partners so they
can deploy effective resolutions before the constraints become a
problem. For example, take the after-the-fact-capacity considerations
that often compromise the appointment systems of large retail
chains. All too often, the retailer end of the supply chain orders a
product and the manufacturer is ready to ship it immediately to an
RDC. Unfortunately, the RDCs’ receiving docks are all booked, and
there won’t be a delivery window for three days. The result is that
stores may have stock outs which, in turn results in lost sales and
reduced customer service levels. The manufacturers also pay a price,
since they’ll have to carry the inventory and replan logistics so that
the order can eventually reach the RDC in a cost-effective way. With
Flowcasting, receiving capacity considerations are taken into
account before orders are released to suppliers resulting in the elim-
ination of delivery delays.
Flowcasting creates a tight and natural linkage between con-
sumer demand at the retail store shelf and output from manufactur-
ing partners in the retail supply chain. Because Flowcasting models
24
the flow of product in a time-phased manner, product will be pulled

across the supply chain to distribution centers and retail stores only
when it’s needed. And because all trading partners are communicat-
ing in the same language, the flow will take place in an optimal way
from a logistics standpoint.
4. Linking the supply chain to true consumer demand.

With Flowcasting, all trading partners win as forecasting moves from
the individual nodes to the place where it counts: the point of pure
consumer demand (the retail store). This is critical, because the
biggest cost of doing business in a retail supply chain environment is
the management of uncertainty on a daily basis. Flowcasting enables
retailers to adopt a truly consumer-centric approach and focus on
pure consumer demand. In this regard, Flowcasting might be called
"Reality Planning and Modeling," since it synchronizes retail, whole-
sale and manufacturing plans to the reality of consumer purchases.
5. Instant business simulations, better reaction time.

Flowcasting enables retail supply chain trading partners to model the
way they want to do business. The focus is on how product should
flow from factories to store shelves. In seconds, trading partners can
perform "what if" calculations that, at one time, would have required
hours on costly hardware systems. By using Flowcasting, retailers
can simulate their entire business and Flowcast their total supply
chain on inexpensive computers and workstations. This enables them
to react quickly to anticipated changes in demand and replenishment
flows of products, and take appropriate actions.
So, rather than being caught by surprise with excess or insuffi-
cient inventories, retailers can respond to change in a way that
enables them to minimize problems and seize new opportunities.
The key is that trading partners are dealing with instant information,
rather than information that is days old and out of synch with activ-
ity at the retail shelf level. Flowcasting enables buyers at all levels of
the supply chain to truly make informed decisions, because the
future has already been modeled and future flows of inventories fore-
25
casted on the basis of real-time consumer demand. What better way

could there be to run a business?
6. Cutting costs at the core.

We recently analyzed 20 of the largest consumer goods (CG) manu-
facturers and 5 of the largest retailers in the world, comparing the
cost of producing product to the cost that a consumer actually pays
in a store. Our analysis showed that for those 20, depending on the
product, consumers pay on average anywhere from 2 to 4 times the
manufacturing cost. (Refer to Appendix G for complete details about
the aforementioned calculations.)
While every retailer and manufacturer has its margin goals and
pricing strategies, 45 percent (on average) of the cost a consumer
pays represents the costs of manufacturing the product. Another 37
percent represents the way these companies market, sell, and dis-
tribute products. Flowcasting enables manufacturers and their retail
trading partners to dramatically reduce their manufacturing, selling,
distribution, and administrative costs. How much? We believe that
the above mentioned 45 percent of the costs of manufacturers can be
reduced as follows: materials 1 percent to 3 percent, labor 3 percent
to 10 percent, and overhead 5 percent to 15 percent. We also believe
that the cost of marketing, selling, distributing, and administration
can be reduced in the range of 5 percent to 10 percent.
The bottom line is this: Trading Partners can reduce their cost
of doing business in the range of 1 percent to 6 percent of their total
sales volume. We calculated ranges because implementing the
Flowcasting business process is a journey and is usually done one
supplier or one retailer at a time, depending on who initiates the
process. Therefore, the speed with which the process is implemented
and the volume of business done by the participating trading part-
ners will dictate the size of benefits achieved. Second, some compa-
nies are more efficient than others and, therefore, you have to allow
for that.
According to the Economist, the Global Consumers Goods mar-
ketplace is a $10.36 trillion Industry. Reducing the costs of manufac-
turing, marketing, selling, and distributing products to consumers
26
by 6 cents on the dollar (which represents the high end of the range)
would generate savings of over $600 billion (USD) to be shared by
trading partners and the consumer.
As you’ll see throughout this book, Flowcasting opens remark-
able opportunities for cutting costs while boosting profitability and
customer service.
But…Does It Work
At this point you might be thinking, "Yes….but can you really elimi-
nate forecasting everywhere but at the retail store level?" The answer
is a resounding "yes." To test the Flowcasting concept, we conducted
several simulations and pilots. Our latest simulation involved gener-
ating 378 different sales forecasts at the store/SKU level at a Fortune
100 Consumer Goods Manufacturer. To provide a context for the
results of the simulation, we chose a study conducted by Georgia
Tech’s Marketing Analysis Laboratory. The study provides data,
summarized in Figure 1, about the state-of-the-art in forecasting
accuracy.
Figure 1: Benchmarking Sales Forecasting Performance by Kenneth B Khan,

Journal of Business Forecasting, Winter 1998-’99.
As shown in Figure 1, our ability to forecast tends to improve at

higher levels of aggregation. This makes good sense; it is far easier to
27
forecast total corporate sales for the year than it is to forecast sales at
the strategic business unit or specific product (SKU) level. Yet fore-
casting at the business unit, SKU, and SKU by location (retail, whole-
sale and manufacturing DCs) must be done. And the lower the level,
the less accurate the forecast. Now what if, as a manufacturer, you
had to forecast at the store/SKU level as well? Would your forecast-
ing accuracy improve? Actually, it would most likely continue to
deteriorate below the level reported in the study (67 percent to 70
percent), even though no statistics on forecast accuracy are available
at the store/SKU level.
Can we improve on the current state-of-the-art? Let’s return to
our simulation for an answer. The objective for the manufacturer in
our simulation was to use Flowcasting to forecast what one of its
retail trading partners, also a Fortune 100 company (which we’ll
refer to as "Retailer X"), would sell during a given month for a sam-
ple of six products. The products included two high-volume SKUs,
two medium-volume SKUs, and two low-volume SKUs sold across a
group of 63 stores. The simulation would then subsequently forecast
what Retailer X would purchase from the manufacturer for that
month.
The simulation was performed on November 12, 2005. First, a
sales forecast for December 2005 and the following eleven months
was created for every product (SKU) in every store for a total of 378
separate forecasts. Then, the Flowcasting system netted store inven-
tories (inventory on store shelves and back room) the morning of
November 12, considered store/SKU safety stock requirements, and
used each store’s ordering rules (minimum ship quantities, lead
times, and shipping schedules) to create a model of what each store
would most likely order by day and week for the next twelve months.
The results were accumulated for the six products across all 63
stores and aggregated to show what Retailer X would specifically buy
from the manufacturer during December. The information was kept
for seven weeks. Then, in early January 2006, when the December
2005 orders from Retailer X had been received and shipped, a com-
parison was made between orders received in December and the fore-
cast made seven weeks earlier for December.
28
Figure 2: Simulation Results for Retailer X.
The level of forecast accuracy ranged from 83 percent to 97 per-

cent for the 6 products as shown in Figure 2. In addition, Flowcasting
gave hard evidence that Retailer X had ordered more inventory than
it needed in anticipation of heavy sales during the holiday season.
Had this not been the case forecasting accuracy at the store/SKU
level would have been higher.
Hard evidence now exists to prove that sales forecasting accu-

racy at the store level can consistently be in the 80 percentage plus
range. With the Flowcasting approach to retail supply chain man-
agement, forecasting uncertainty only exists at the store, so that the
balance of the retail supply chain becomes a mere calculation.
Therefore, with a Flowcasting business process in place, sales fore-
casting at all other levels in retail, wholesale, and manufacturing
companies can be completely eliminated. The implications for retail
supply chains are staggering, which brings us to the final point in
29
this introduction: it’s time to start thinking of Flowcasting as the

gold standard for retail supply chain management.
The Flowcasting Imperative

If you look back at the introduction of new planning and execution
business processes such as MRPII, DRP, and ERP, you’ll see a similar
pattern. Early adopters understand the vision and potential, leap in,
and serve as guinea pigs while bugs and glitches are hammered out.
And while they might be on the "bleeding edge," they’re also the first
to reap the benefits. Others soon follow, recognizing that the busi-
ness processes will be essential to running a profitable business.
Lastly, a group of companies adopt the business processes and sup-
porting technologies because they’ve found themselves at a compet-
itive disadvantage.
Where are we with Flowcasting today? A number of companies
-- some at the Fortune 100 level -- are already implementing the
Flowcasting business process and harnessing the immense power of
planning based on a single forecast of consumer demand to gain a
cost and competitive advantage. It’s not a matter of if your competi-
tors will adopt Flowcasting, it’s a matter of when they’ll begin using
Flowcasting to improve planning, reduce costs, boost customer ser-
vice, and reap all of the benefits of selling through finely-honed and
well-synchronized supply chains.
And while Flowcasting represents a new paradigm for manag-
ing the supply chain, it’s also an adaptation of existing systems that
enables companies to easily transition to a new and highly efficient
way of doing business. Unlike so many systems that require massive
investments of time and money and the scrapping of existing sys-
tems, Flowcasting allows for the gradual adoption of new planning
and modeling capabilities. Perhaps most appealing, Flowcasting sim-
plifies the management of the entire retail supply chain.
Thirsting to know more? This book is designed to help you
understand why it’s critical to be on the earlier side of the adaptation
curve and how you can use Flowcasting to gain a competitive advan-
tage. It’s divided into three sections that provide you with the essen-
30
tial information you need to know to adapt Flowcasting effectively in

your organization.
Section 1 covers the current difficulties and challenges of fore-
casting and planning the retail supply chain, focusing on the differ-
ence between independent and dependent demand. It also reveals a
little-recognized fact: most of the forecasting in the retail supply
chain is unnecessary! Section 2 explains how to use Flowcasting to
forecast demand at the store and integrate replenishment schedules
among stores, DCs, and manufacturing plants. The third and final
section discusses how problems that were previously tackled with
disjointed forecasting can be solved more effectively using
Flowcasting. It shows how Flowcasting, which creates a valid simu-
lation of every future product movement, makes it easy to plan pro-
motions, products’ phase-ins and phase-outs, seasonal set-ups and
take-downs, labor scheduling, capacity planning, load building,
transportation, and financials.
Given the enormous benefits of Flowcasting, we hope you’ll join
in the new race for greater retail supply chain efficiencies, lower
costs, higher customer services, and greater profitability. Don’t wait
-- some of your contenders are already rounding the first lap!
André Martin
Mike Doherty
Jeff Harrop
May 2006
31
Flowcasting
32
Flowcasting
Section 1:
Flowcasting Basics
33
Flowcasting
34
Flowcasting: Chapter 1
C H A P T E R 1
The 21st Century

Retail Supply Chain
Doing Business with a Single Forecast
➫
B ack in 1958, Jay Forrester, then a professor at MIT, wrote a
groundbreaking article stating that a volume increase of 10 per-
cent at the retail store level actually cascades and translates into a 40
percent increase at the manufacturing level. His findings, which
originally appeared in the Harvard Business Review, were later doc-
umented and published in a book entitled Industrial Dynamics.
Forrester’s work focused on the behavior of the flows of money,
orders, materials, personnel, and capital equipment across what we
refer to today as "retail supply chains." He found that these five flows
were integrated by an information network. This information net-
work gave retail supply chains their own dynamic characteristics.
Interestingly, Forrester calculated that the four-fold increase at
the manufacturing level would take six months to manifest itself.
This makes sense, given the fact that in those days, most business
was done by snail mail.
Fast forward to the 21st century. Remarkably, the multiplier
between retailer and manufacturer is still the same: four-fold. The
only difference is that the ripple effect takes days or weeks, not
months, to materialize thanks to the use of communication tech-
nologies that form the life blood of today’s networked supply chains.
The elapsed time may have shrunk, but the volume amplifier effect
remains because we still manage the nodes in a supply chain inde-
35
pendently, almost as if they’re discrete islands or fiefs, blind to each

other’s needs. Because of this disconnect, pure consumer demand
takes a back seat to business incentives, such as promotions, which
create forward buys and diversions. Optimal shipping formats can
also play a role in the multiplying demand. For example, the retailer
may only require three cases of an item. But the manufacturer offers
incentives to ship in pallet or truckload quantities. The retailer then
winds up with more inventories than needed. Hence 10 percent at the
retail store level becomes 40 percent at the manufacturing end of the
supply chain.
This amplification of demand, coupled with the various discon-
nects across the supply chain, inevitably leads to uncertainty about
the items, quantities, and delivery dates that various trading partners
will need over the course of days, weeks, months, and beyond. That
uncertainty, in turn, results in excess inventories and added costs.
The good news, as you’ll learn in this chapter, is that there are now
solutions for gaining better visibility into the retail supply chain, so
you can eliminate surprises without using inventory as a hedge. The
first part of this chapter explores the problems with forecasting in
retail supply chains today, while the second shows how forecasting at
the retail store level is a far more accurate way to predict demand
from the factory to the store shelf. If you believe that retail store level
forecasting is impossible or just a dream for tomorrow, read on.
PART 1: FORECASTING IN A VACUUM
Every Node for Itself

The largest cost of doing business across a retail supply chain is deal-
ing with the uncertainty of demand. This is directly related to not
having visibility into what customers will actually buy, what quanti-
ties they’ll purchase, and when they’ll make their purchases. This
lack of visibility and associated uncertainty results in major operat-
ing disruptions and significantly increased costs of doing business.
The lack of visibility is manifest in the fact that, in most supply
chains, each node does its own forecasting. A retail store may have
some sort of forecasting and replenishment system, but that system
36
usually functions in a vacuum. The same holds true for the RDC,
which looks at the needs of the stores it supports as an aggregate sum
that’s computed on the basis of history such as warehouse with-
drawals. It is only a coincidence if the aggregate forecasts at the RDC
level relate at all to what is actually happening on the retail shelves.
At the next downstream level in the supply chain, from the
manufacturing DCs to the factory floor, you’ll typically see tighter
planning and data integration due to the use of MRPII and ERP. But
even the most exquisite choreography among the manufacturing
nodes won’t improve supply chain management unless all the various
nodes are connected with actual demand at the retail store level.
Nonetheless, few retailers today actually forecast at store level;
instead, forecasting typically starts at the RDC, and is ultimately done
throughout the supply chain. The farther from the consumer that
the forecasting takes place, the less accurate the forecasts are likely
to be. And depending on what’s being forecasted, inaccuracies can
have a significant impact on a business. Inaccurate sales forecasts
across a given retail supply chain translate into increased operating
costs and lower customer service levels for all trading partners. In
addition, inaccurate sales forecasts lead directly to:
• Lost sales
• Dissatisfied customers
• Too much inventory
• Increased selling costs
• Increased distribution costs
• Increased manufacturing costs
• Increased purchasing costs
• Obsolescence
To appreciate the roots of forecasting inaccuracies, consider a
typical retail supply chain (see Figure 1.1), which consists of two sep-
arate legal entities, a retailer and a manufacturer.
Figure 1.1: A typical retail supply chain.

37
The retailer owns the first two nodes, which consist of the retail
store and the RDC and the manufacturer owns the last two nodes, the
manufacturing DC (MDC) and the plant. In an effort to generate the
best forecast, both the retailer and manufacturer must forecast at
multiple levels. Let’s start with the forecasting on the retailer’s side.
The Retail Nodes

Retailers typically begin forecasting at the total business level, based
on a plan with a timeline that usually stretches out one year and
beyond. This business plan normally includes a forecast of total sales
expected for the business period, as well as assumptions about new
store openings and closings, store renovations, new product intro-
ductions and deletions, and anticipated promotions. In addition,
business volume forecasters will compare their forecasts to those of
prior years, compare their data with key competitors’ data by buying
syndicated data, examine market share, and consider other critical
economic and competitive factors as well. Figure 1.2 lists the various
functional areas that create forecasts and the type of forecasting
activities they typically perform in a typical retail operation; in aggre-
gate, the number of different forecasts that are created is staggering!
Regardless of how many forecasts are created in a given retail
company, they’re done for high-level planning purposes. And the lit-
Figure 1.2: Retailer’s functional forecasting activities.

38
mus test of a "good" plan is financial. Will the plan generate enough
cash from operations to meet our targets? Will we have to borrow to
support store expansions, and store renovations and closings?
In other words, the planning, by definition forecasting, is relat-
ed more to the blessings or condemnation of Wall Street than to the
planned flow of product from manufacturer to retailer to consumer.
And therein lies the problem; these higher-level forecasts are not
linked to the day-to-day business activities and lower level forecasts
that drive replenishment into the stores, which in turn drives replen-
ishments (purchases from suppliers) into the distribution centers. As
a result, it would be a remarkable coincidence if the sum of the store
and DC forecasts were to add up to the corporate sales forecast!
The Manufacturing Nodes

Like retailers’, manufacturers’ business plans must also include a
forecast of total sales expected for the business period, which means
forecasting at multiple levels. Typical forecasts must take into
account the proposed marketing and sales targets, promotional
plans, new lines of products, and product deletions. Given expected
sales volumes, manufacturing must make adjustments to production
capacities (adding or eliminating production shifts), and the addi-
tion, expansion, or closing of distribution centers. Figure 1.3 below
depicts the major forecasting activities that functional areas of a
manufacturing organization undertake as they plot the future.
Higher level plans in many manufacturing companies are dis-
connected from day-to-day business activities and the lower level
forecasts that ultimately determine factory output and replenish-
ment into manufacturing distribution centers (MDCs). What are the
chances that the sum of the sales forecasts that drive both distribu-
tion replenishment and production will add up to the corporate sales
forecast? Slim to none!
The Wrong Stuff

The problem with so many independent forecasts across the supply
chain is two fold. First, the very fact that they’re independent and dri-
39
Figure 1.3: Manufacturer’s functional forecasting activities.
ven by internal metrics means that they do not mirror changes in

actual demand at store level.
Second, the independent forecasts are expressed in mutually
exclusive units of measure. People in sales will be forecasting in
terms of the number of units that will be sold and the currency those
units represent. People in distribution will be forecasting in terms of
how many trucks will be coming and how many people will be need-
ed to unload them, how much warehousing space will be needed,
what kind of resources will be necessary for order picking, and so
forth. And in manufacturing, people will be forecasting in terms of
how many batches, pieces, or units the plants must produce, how
many shifts will be needed, and so on.
The fact is, on both the retail and manufacturing side, there’s a
whole underworld of forecasts below the radar screen -- and none of
them are connected.
PART 2: A BETTER WAY OF DOING BUSINESS
Doing—and Forecasting—Business Your Way

Given the patently clear shortcomings of the current "forecast every-
where but where it counts" approach, it’s tempting to ask, "Why don’t
40
we just forecast the way we conduct business?" The question is so

obvious that we rarely, if ever, bother to ask it. In part, doing busi-
ness as usual is easier than questioning whether our planning tools
map to our actual business. As we saw earlier, with marketing, sales
and management people, sales forecasting is usually done at the cor-
porate, regional, category, or product group level. On the operating
side of the business (distribution, manufacturing, and purchasing),
sales forecasting is typically done at the product and raw material
level. Although these methods of forecasting have value, they bear
no resemblance to the way we do business. Consumers do not buy
products at the corporate, regional, category or product group level;
consumers buy specific products in specific stores on specific days.
Distribution people do not ship and receive products at the corpo-
rate, regional, category or product group level -- companies build dis-
tribution centers and ship products to support retail stores with the
products they need. And factories produce and ship products to dis-
tribution centers. Until retailers, wholesalers and manufacturers step
back and take a reality check on their planning practices, forecasting
will continue to be out of synch with the actual flow of product
throughout their supply chains.
Another reason that we forecast at the wrong place is our
assumption that since we’re dealing with supply chains, the answer
to our problems lies in technological fixes and practices. So we keep
developing new systems and practices that try to improve supply
chain management. During the last several years, a number of initia-
tives such as Quick Response, Vendor Managed Inventory and
Efficient Customer Response have improved overall supply chain
performance. As good as some of these technologies and practices
may be, none focus on the core issue: you only need one unique sales
forecast to drive a retail supply chain. The retail store is both the
beginning and the end of retail supply chains. It’s the beginning of
the information flow and the end of product delivery. So why not
start there? As you’ll see below, if you forecast at the retail store, the
need to forecast anywhere else in the supply chain completely van-
ishes. The magic behind this lies in a concept called "dependent
demand," which we’ll turn to next.
41
Dependent Demand
Years ago in the auto industry, people realized that once you fore-
casted how many of a particular car model you would assemble and
then sell, you could easily calculate the demand for tires, steering
wheels, hubcaps and a variety of other parts. These item level fore-
casts were based on "dependent demand" -- that is, they depended
entirely on another item’s forecast (the assembly schedule).
Dependent demand is important in retail planning as well.
Consider the supply chain shown in Figure 1.4. It consists of a dis-
tribution channel (or network) with four levels: a factory, a manufac-
turer’s distribution center (MDC1), two retail distribution centers
(RDCs 1 and 2) and four retail stores. At every node of this distribu-
tion channel, a customer/supplier relationship has been created. For
example, the factory has one customer, MDC1. MDC1 plays a dual
role -- it is the customer of the factory and the supplier to RDC1 and
RDC2. RDCs 1 and 2 also play dual roles; each is a customer of MDC1,
and each is a supplier to a specific number of stores (two each in this
example).
Figure 1.4: Dependent demand in a retail environment.
42
Suppose that when store 108 needs products, it orders from

RDC1. When RDC1 needs products, it will order from MDC1. The
product demand that RDC1 experiences will always be generated by
stores 108 and 602. In other words, the demand on RDC1 is depen-
dent on the needs of stores 108 and 602.
Another way to look at this distribution channel is to think of
the way products will normally flow from factories to store shelves.
Once you have forecasted what consumers will buy at the store level,
you can calculate the demand flow through every node and trading
partner within this distribution network. And this makes perfect
sense. After all, we build distribution centers to serve the demands of
other DCs and stores. So why not have forecasting and planning
processes and systems follow in the same path that supports the way
we actually do business?
One Retail Supply Chain, One Forecast

All retailers and their supply chain trading partners who sell and dis-
tribute products must answer the following "universal logistics ques-
tions":
1. "What am I going to sell?"

2. "Where will I sell it?"
3. "What do I have?"
4. "What do I have on order?"
5. "What do I have to get?"
As shown in Figure 1.5, the first question need only be

answered at the final point of sale, the retail store. The calculated
demand at each level, from retail stores through the supplier’s facto-
ries, is the one set of numbers that can be converted into meaning-
ful units within each functional area of the supply chain. (This is
quite a contrast to the traditional approach in which each functional
area has its own currency and is left to its own devices to develop a
forecast that enables it to hit its bogey. The forecasts are done on the
basis of history, application of "creative" formulas, and rules of thumb
developed over the years through hard work and experience.)
43
Figure 1.5: One sales forecast only at store level.
The answer to each of the subsequent questions lies in calcu-

lating and communicating demand to the next level of the supply
chain where inventory is maintained. What sells at the retail shelf
drives how much the retail store needs. What the retail store needs
drives what the retail DC needs to provide, and so on until the entire
retail supply chain is calculated and synchronized.
An important benefit of this approach is that the universal
questions are answered not just for today, but for a planning horizon
of 52 weeks into the future, in daily time increments. Imagine not
only knowing your current inventory balance, but your projected
inventory balance weeks and months into the future. Or what your
expected purchases will be weekly for the next 52 weeks!
As desirable as this may sound, it’s also tempting to ask,
"Wouldn't it be more accurate to forecast at the retail distribution
center level instead of the store level? Isn't this the law of large num-
bers?" The "law of large numbers" may be intuitive, but several pilots
and simulations have shown that forecasting at the store-level only
and Flowcasting back yields a more accurate RDC demand plan than
44
forecasting at the RDC level explicitly, even though you’re dealing

with smaller numbers. This is because the law of large numbers still
applies -- the only difference is that you’re taking a sum of the store
forecasts rather than a forecast of the sum.
Moreover, forecasting at the distribution center level actually
introduces a significant source of error: store-level inventories. If the
stores have too much inventory, any RDC level statistical forecast
based either on aggregate sales history or warehouse withdrawals will
yield forecasts that are larger than what is actually needed. The result
is even more excess inventory. If the stores don't have enough inven-
tory any RDC level statistical forecast based on the aggregate sales
history or warehouse withdrawals will generate a forecast that is too
low, resulting in out-of-stocks.
A multitude of factors make forecasting at the RDC level differ-
ent from the sum of the POS forecasts at the RDC level supporting
these stores. These include store on hand balance, shelf resets, deliv-
ery schedules, minimum shipping quantities, supplier ordering rule
changes, product phase in/phase out, and so on. Figure 1.6, which is
based on actual numbers from a simulation, shows the sum of the
POS forecast for a product at a number of stores supported by a retail
DC, as well as the projected demand (dependent demand calculation)
on the distribution center from the same stores. It clearly shows the
effect of one of these factors: inventory imbalances at store level.
The solid line is the sum of the POS forecasts for the more than
100 stores supported by this distribution center. The dotted line is the
actual demand, or dependent demand, that the distribution center
will experience once store level inventories are taken into account.
Notice that in the first week, the dotted line demand is significant as
the stores that are below the minimum display quantities are brought
up to the minimum.
The differences between the dotted and solid lines indicate the
degree of error that can exist between the POS forecasts and an accu-
rate orders forecast (what a retailer will buy from suppliers). The typ-
ical orders forecast resembles the solid line, while the orders fore-
casts we are proposing would resemble the dotted line.
45
Figure 1.6: POS Forecast based on retail DC that supports more than 100 stores.
Figure 1.7 shows the effect of yet another factor that diminish-
es the value of RDC level forecasts: shelf resets at store-level.
The area of the graph on the left, beginning just before 9/28,
represents the additional demand on the RDC as the number of fac-
ings for this product is increased at the stores. These changes are
scheduled to occur on different dates at the different stores.
The area in the middle of the graph, beginning at 11/23, repre-
sents the depressed demand on the RDC that results from returning
the number of facings to the original level at the end of the selling
season. Again, these changes are scheduled to occur on different
dates at the different stores.
Notice that there is a significant difference between the two
curves, showing that the product forecasted has excess inventory at
the stores that will take some time to sell off. This situation is repre-
sented by the lower dependent demand curve which, over the course
of a year, finally catches up to the POS forecast curve. The point is
that any calculation that does not take inventory at the stores into
account on a store-by-store basis will provide the supplier with an
incorrect picture of the future.
46
Figure 1.7: Shelf resets at the store level.
Summary
In this first chapter, we’ve looked at the traditional retail supply chain
model in which hundreds of forecasts are generated by various func-
tional trading partner groups. We’ve also commented on the prob-
lems associated with that model. Here are the key points to remem-
ber:
1. Traditional forecasting in retail, wholesale, and manufactur-

ing companies is done to satisfy high-level business plans,
which are based on aggregate numbers across a company’s
operations, rather than calculated on the basis of actual con-
sumer demand.
2. Forecasting is done everywhere but where it counts: at the
retail store level. The result is a collection of disconnected
forecasts.
3. When forecasting is done in the traditional disconnected
fashion, stock outs and excess inventories are inevitable and
promotions become logistical and financial nightmares. As a
47
result, all trading partners experience higher operating costs.

4. Contrary to conventional wisdom, forecasting can, and
should only, be done at the retail store level. Everything else
throughout the supply chain can be calculated on the basis of
the retail store-level forecast.
5. By forecasting at the retail store level, the need for hundreds
of forecasts generated elsewhere in the supply chain vanish-
es. That provides visibility into the entire retail supply chain,
which means that retailers, wholesaler and manufacturers
can model the way they actually do business! The net result
is that uncertainty is totally removed from the retail supply
chain except the only place where it truly exists: the retail
store.
6. By forecasting at the retail store level, all trading partners are
poised to improve the way they do business.
The next chapter introduces a new concept, "Flowcasting,"

which provides unprecedented insight into consumer demand and
every activity in a retail supply chain, from the moment product
leaves the factory to the time it is purchased at the retail store.
48
chapter 2 4/4/06 1:05 PM Page 1
C H A P T E R 2
From Forecasting to
Flowcasting
The New Art and Science of
Managing a Retail Supply Chain
➫
A Brief History of Supply Chain Technology
Since the advent of the industrial revolution, manufacturers built
distribution facilities to match their unique needs. As they grew,
manufacturing businesses found that it made sense to build distrib-
ution centers in geographic areas with heavy customer concentra-
tions. Doing so enabled them to achieve economies of scale; they
could manufacture products in a specific city, then ship full or near-
ly full truckloads to a given distribution center in another city, there-
by reducing transportation, handling and warehousing costs. This
approach not only saved money through newfound efficiencies, but it
enabled manufacturers to improve customer service.
Wholesalers and retailers followed a similar course. Retailers,
for example, found that it made sense to build distribution centers in
geographic areas with a heavy concentration of retail stores. The
motivation was the same as with manufacturers: achieving
economies of scale in transportation, handling and warehousing
costs, and improving service to retail stores and consumers. As mar-
kets further expanded, wholesalers and retailers began to look for
additional opportunities to reduce their distribution costs. This led to
the beginning of warehouse automation projects, many of which are
still going on today.
49
During the 1950s and 1960s, companies also began to make

investments in inventory management and replenishment systems
designed to support the burgeoning distribution networks. The first
inventory management systems were manual and included the visu-
al two-bin system, followed shortly after by the visual reorder-up-to
system. In the early 1950s, we saw the emergence of the manual Min-
Max Kardex system, which was subsequently replaced by the manu-
al, and then computerized, ReOrder Point system. For manufactur-
ing, wholesale, and retail companies, the primary concern in inven-
tory management was determining how much to order and when.
Today, the emphasis is more on the "when" part of the equation. The
attention has shifted to getting the right material at the right place
and at the right time.
The mid-1970s saw the emergence of a new approach to inven-
tory management and replenishment on the distribution side of a
manufacturing business: Distribution Resource Planning (DRP),
which revolves around the concept of dependent demand. DRP arose
from the recognition that 1) within a manufacturing company, the
capabilities of production facilities were out of synch with the needs
of distribution facilities, and that 2) inventory management and
replenishment practices had to accommodate inventory and produc-
tion dependencies throughout the company’s internal supply chain.
Today, DRP is universally accepted in modern manufacturing and dis-
tribution operations in consumer goods manufacturing companies.
But what about managing inventory on the retail side of the
supply chain? Back in the 1980s, it was recognized that DRP and
dependent demand had applications in retail and that a complete
retail supply chain was indeed interdependent. But the sheer volume
of product/location combinations within large retail operations made
the DRP approach to inventory management and replenishment
impractical at the retail store level. (By contrast, the DRP system in
a large consumer goods manufacturer may have to plan for, and man-
age, 250,000 product/location combinations. A large retailer may
have upwards of 400 million product/location combinations that
need to be planned and managed.) As we stated earlier in the book,
50
advances in technology and forecasting have made Flowcasting a

reality.
In this chapter, we’ll explore the benefits of forecasting at the
store level through Flowcasting, and show how these accrue to retail-
ers, wholesalers, and manufacturers alike.
The Power of One (Forecast)

Imagine how different the management of your company would be if
you could operate and manage your total business on the basis of the
single forecast described in Chapter 2. That one forecast would pro-
vide an unprecedented opportunity to reduce costs. Remember, the
greatest cost of doing business today is managing uncertainty, and
sales forecasting uncertainty only exists at the final point of sale. By
forecasting only at the retail store (the underlying premise of
Flowcasting), the entire retail supply chain benefits in two major
ways: first, from simplicity (demand throughout the supply chain can
be calculated from the store-level forecast), and second, from the
ability to model several weeks into the future the flow of product
from the store shelf to the factory.
This modeling capability is extremely powerful and places users
in a proactive mode by establishing a "rack and pinion"
customer/supplier relationship that starts at the retail store and con-
nects every trading partner in the supply chain. By time-phasing rec-
ommendations for product over a period of several days, weeks, and
months into the future, retailers can simulate the future in the way
they actually do business. Such modeling greatly improves decision
making and customer service while significantly reducing costs. In
addition, it gives retail supply chain partners greater control of their
business activities. The key is to think in terms of "Flowcasting" --
the ability to calculate, in a time-phased manner, the flow of products
in and out of each node in a given supply chain from the factory to
the store shelf.
The Flowcasting Concept

Flowcasting starts at the head of the retail supply chain (the store)
by forecasting what consumers will buy each day over the forecast
51
period -- typically one full year to capture the entire business cycle.
It calculates dependent demand to predict how much inventory RDCs
must ship to the stores, and when specified quantities of product
must arrive in order to meet consumer demand from several days to
weeks into the future. Flowcasting repeats this process for every sup-
ply chain node that a product will flow through on its way from the
factory floor to the retail store shelf. Figure 2.1 shows the power of
this approach within the first two nodes of a retail supply chain. The
detailed logic and mechanics of flowcasting will be discussed in later
chapters, beginning with Chapter 3.
The ability to Flowcast the inflow and outflow of products
across each node in a given retail supply chain enables the transla-
tion of information into the various languages of the key functional
areas within a retail company. For example, as shown in Figure 2.2,
planned receipts (into a store or RDC) can be converted to receiving
hours in order to plan receiving capacity. The capacity plan for
receiving can also be expressed in terms of the number of trucks that
need to be received in a given retail store or RDC, thereby trans-
forming the typical appointment system into a forward looking sys-
tem in which valid delivery dates can be stated on planned purchas-
es before purchase orders (or supplier schedules, covered in Chapter
Figure 2.1: Flow of products in and out of retail stores and DCs.
52
Figure 2.2: Conversion of planned receipts into receiving hours.
5) are sent to suppliers. By using this approach, suppliers would no

longer have to call in for appointments before making a delivery --
Flowcasting creates a receiving capacity plan for the planned
inbound flow of traffic and can be used to match daily receiving
capacity before sending out purchase orders.
Planned shipments to stores and RDCs can also be converted to
shipping hours to plan shipping capacity as shown in Figure 2.3. This
Figure 2.3: Conversion of planned shipments into shipping hours.

53
Figure 2.4: Total capacity plan for an RDC.
makes it possible, for example, to generate a capacity plan of total

manpower for a given retail DC (see Figure 2.4).
Output from Flowcasting can also be used to make financial
projections of planned product receipts from suppliers in dollars or
euros (or any other currency) by week, by month, or one year into the
future. These projections become excellent input for cash flow plan-
ning purposes. Companies can even take the projected product
receipts from suppliers and offset them by their payment terms to
predict the amount of accounts payable that product purchases will
represent. Projected inventory levels can be converted to projected
inventory investment in the currency of choice, as shown in Figure
2.5. The projected inventories can also be converted into cases and
pallets in order to calculate how much warehousing space will be
needed. Since information is generated by a single consumer sales
forecast and is time-phased a year out, it drives every key function in
a retail company. In other words, the consumer sales forecast
becomes a "universal" set of numbers that can easily be trained to
speak the various functional languages of the company.
In Chapter 11, we delve into significant detail on the benefits of
using Flowcasting as a financial planning tool.
54
Figure 2.5: Projected inventory levels converted to inventory investment

and space requirement.
Flowcasting for Lower Costs

By Flowcasting demand in a retail supply chain, companies can cre-
ate numerous opportunities for reducing the cost of selling and dis-
tributing products. To appreciate the magnitude of these opportuni-
ties, consider the manner in which most companies order and deliv-
er products today, as depicted in Figures 2.6 and 2.7.
Figure 2.6: Ordering product in typical retail supply chains.
55
Figure 2.7: Delivering product in typical retail supply chains.
Normally, retail stores order from retail DCs, which in turn

order from manufacturing DCs (MDCs) that place orders on the fac-
tories. Typical delivery cycles follow the same pattern, but in reverse:
factory to MDC, MDC to RDC, and RDC to the store. (Note that we say
"typical," because there are some variations. A retailer might, for
example, cross-dock through a MDC or through an RDC. Or a retail-
er might bypass either the MDC or RDC.)
To Flowcast this supply chain, we would start with daily
requirements at the head of the supply chain -- the store. Once those
requirements are known, the entire retail supply chain can be syn-
chronized to meet those needs. As a result, ordering and delivery
requirements to retail stores will be completely visible to supply
chain participants that manufacture and distribute those products.
Trading partners will know the specific retail store needs today,
tomorrow, next week and well into the future. The same visibility will
apply to all other nodes in the supply chain. As a result, when condi-
tions change (selling over or under forecasts), product flow require-
ments can be recalculated daily and communicated to supply chain
participants.
Imagine how much supply chain partners could reduce selling
costs if they had daily visibility into their customers’ inventory levels,
what and how much product their customers sell every day, and
when their customers will order again. Now imagine this informa-
tion being refreshed daily and shared among supply chain trading
partners. The efficiencies and savings would be enormous!
56
Consider how the following trading partners would benefit if

Flowcasting and the resulting Flowcasting process were implement-
ed in their retail supply chains:
Retailers
• Retail distribution centers would become, over time, cross-docking

and repackaging facilities. Products would arrive from supply
sources and would then be repackaged when necessary in units of
weekly store sales. Initially, most products would be cross-docked
and inventory turn rates in RDCs would rise to 50 or more annual-
ly. Repackaging operation costs would be shared among supply
chain trading partners. Over time, fewer retail DCs would be
required.
• Retail stores would receive deliveries so that trucks would be

unloaded in shelf placement sequence. Since repackaging would be
done in RDCs (or in 3PL provider facilities), retail stores would be
able to free up significant shelf space that management could use
to fulfill other consumer needs. Backroom inventories would
become a thing of the past. Retail store sales would increase 2 to 8
percent, while store inventory turns, over time, would rise to 50 or
more annually depending on products’ selling velocities.
• Retailers would no longer need separate forecasting and replenish-

ment systems for their stores and RDCs. One common system
would manage inventory in both stores and RDCs and interface
with suppliers on a daily basis.
• The internal theft portion of product shrink would be reduced by at

least 50 percent due to the hourly, and up to the minute, visibility
of what is on the shelf, what is sold, and what is coming into the
store.
• One common system would span cross-organizational boundaries.

Store operations, merchandising, buying, category management,
57
distribution, finance, and management would all be working from

one set of numbers.
Wholesalers
• Flowcasting would enable wholesalers to become value-added part-

ners for their manufacturing suppliers and retailing customers.
Their businesses would transform as they become professional
cross-dockers and repackagers for those retailers or manufacturers
that cannot or do not want to take on those activities themselves.
• Wholesalers would only need to hold a few days of inventory, com-

pared to the volumes they carry today.
• Wholesalers would require far less warehousing space. Many of

their current activities in purchasing, distributing, marketing, and
selling would be significantly reduced and, in some cases, elimi-
nated.
Manufacturers
• A manufacturer’s way of doing business would change completely.

The timeframe for the change would depend on the size of their
retail and wholesale customers, and how rapidly those customers
adopt Flowcasting. For example, if the major global retailer of a CG
manufacturer we’ve researched adopts Flowcasting, that manufac-
turer would no longer have to forecast 30 percent or more of its
business. Over time, as more retailers adopt Flowcasting, the man-
ufacturer would gradually convert from a manufacture-to-stock
(MTS) strategy to a manufacture-to-order (MTO) strategy for most
of their business, and reap all the economic and productivity bene-
fits that derive from the MTO approach. Gone would be the uncer-
tainty of demand, associated safety stocks, and warehousing and
operating costs as well as last minute and very costly production
schedule changes.
58
Finally, supply chain-wide inventory investment would drop

significantly for those retail supply chain partners that adopt the
Flowcasting way of doing business. We predict that once critical mass
is achieved, finished goods inventories in the global consumer goods
industry would drop by two-thirds. These inventories, which now
range from 80 to 120 days on hand in the consumer goods industry,
would drop to 30 to 45 days on hand, which would represent a sub-
stantial opportunity to reduce costs, product obsolescence, and
returns.
Improving the Execution of Sales Promotions

As vital as promotions are to a retail business, they often result in lost
sales and excess inventory. The root cause is almost always improper
deployment. It’s one thing to forecast how much you will sell in total
for a given promotion -- retailers working with the supplier offering
the promotion usually do a good job of this task. But problems begin
when the retailer has to spread the promotion quantities over partic-
ipating stores. Retailers usually ship the complete promotion quan-
tity to a store just ahead of the promotion. Since most promotions
last a week, there is little time to make adjustments. By the time the
promotion is over, retailers inevitably end up with out-of-stocks in
some stores and overstocks in others.
Promotions are also critical for manufacturers in the consumer
goods sector. As much as 40 to 70 percent of a CG company’s annual
sales volume is derived from products that are promoted to retailers
and wholesalers. In certain product categories, many wholesalers
only buy from promotion to promotion. Unfortunately, manufactur-
ers typically have significant problems with planning and executing
the promotions. The major challenges center on production con-
straints. On average, most manufacturers promote their products
three to four times per year for a period of three to four weeks per
promotion. That translates into an enormous amount of inventory
created in advance of promotions. By way of example, the U.S. gro-
cery industry (as of 2005) manufactures and distributes roughly 50
percent of their annual volume to retail DCs over a 12-week promo-
tional period as promotional inventory.
59
Regardless of the industry, promotions are scheduled well

ahead of time. Since manufacturing people need advance informa-
tion about the scheduling of promotions, they are normally able to
plan production and accommodate capacity constraints accordingly.
For example, let’s assume that the maximum amount of product a
manufacturer can produce is 1,000 units per day. Let’s also assume
that the manufacturer knows a promotion for the product will start
in three months and that the promotion forecast predicts demand of
10,000 to 12,000 units per day (or 10 to 12 times the plant’s daily pro-
duction capacity). This means that the manufacturer has no choice
but to start producing product and holding it as inventory several
weeks before the promotion begins. In fact, many consumer goods
manufacturers typically start making product 10 to 12 weeks ahead
of a promotion. In other words, they produce very large amounts of
inventories ahead of the stores’ needs and deploy it to their own DCs
in anticipation of receiving retail/wholesale orders for a particular
promotion.
Manufacturers must resort to this strategy because they have
little flexibility to add or reduce production capacity without creating
a major financial impact. That means for 40 to 70 percent of their
annual production volume on promoted products, manufacturers
have already produced 100 percent of what will be sold during a pro-
motion! That inventory is normally deployed into the manufacturer’s
own DCs six to eight weeks before that product will be sold in retail
stores. So for promoted products, the problem is not with produc-
tion -- the real issue is one of executing proper product deployment
from MDCs to wholesale or retail DCs and stores. If there are con-
straints during promotional periods, they will be found outside the
four walls of the factory, anywhere within the retail supply chain.
Impact of Flowcasting on Demand

and Business Planning
The introduction of Flowcasting will bring with it the need for new
approaches to planning. Like manufacturing systems thirty years
ago, the planning systems in retail organizations as mentioned earli-
er are isolated today from one another. As a result, the top-level cat-
60
Figure 2.8: Disconnected retail systems.
egory plans are not directly connected to the systems that purchase
and replenish products (see Figure 2.8).
Top-level plans are developed, but a reorder point system orders
products when the inventory at a particular location drops to the
trigger point. If the sum of these purchases adds up to the dollar
amount on the category plans, it’s just coincidental.
A Flowcasting system creates a direct connection between top-
level plans and the execution systems that buy products. The same
direct connection is also in place with respect to the systems that are
used to make decisions about hiring full- or part-time associates,
negotiate freight rates and/or increase the fleet size, and project prof-
it and inventory levels into the future.
In a Flowcasting environment, the business planning process is
similar to that used in manufacturing operations. Flowcasting, how-
ever, adds components to both demand management and resource
planning that make it possible to link business plans with a single
time-phased forecast that starts at the retail store end.
The left column of Figure 2.9a shows the overall business plan-
ning process. First the business plan, which is developed in the cur-
rency of choice, would be agreed upon. Next, product plans at a high
level in the merchandise hierarchy or at the category level would be
developed to support the business plan. These high-level plans would
be translated into the demand side of the business; that is, what the
consumers are expected to buy. At the "atomic" level (the store/SKU
level), the plans are aggregated to confirm, or change and realign,
higher level plans.
The right column of Figure 2.9a shows the four basic compo-
nents of demand management that Flowcasting integrates. (These
components, which will be described in detail in Chapters 3 through
5, enable effective demand management in a retail setting.)
61
Figure 2.9a: Demand

Management components
of a Flowcasting system.
Figure 2.9b shows the resource planning components that

Flowcasting injects into the planning process. These five components
(see Chapters 6 through 10 for details) enable retail companies to
effectively manage all their key resources -- inventory, people, space,
and equipment.
Figure 2.9b: Resource

Planning components
of a Flowcasting system.
62
The Demand Management and Resource Planning components

of Flowcasting create a direct connection between the top-level plans
of the business and the operating plans that actually order merchan-
dise, schedule people's time, arrange for transportation, etc. For
example, if a buyer decides to make a large purchase of an item, that
quantity will immediately show up in the merchandise plans, along
with:
• the amount of storage space needed (in the space plans)
• the weight and cube needed (in the transportation plans)
• the hours required to unload the shipment (in the capacity
plans)
• the cost of this large purchase
• the inventory increase
• the change in gross margin (in the financial plans)
In short, everyone will be in the game using the same playbook.

Best of all, there’s still time to change the plans if business conditions
change. From a planning perspective, this is vastly better than look-
ing in a rear-view mirror and piecing together "what we should have
done."
Flowcasting Will Spawn New Modes of Distribution

As a result of Flowcasting, distribution, and logistics professionals
will finally be able to apply their expertise in their field and generate
significant savings for their companies; distribution people have long
known how to save money, but they’ve lacked the timely information
necessary to do so. Flowcasting gives distribution professionals the
visibility they need to take advantage of opportunities for bypassing
nodes in retail supply chains when applicable. This will result in
increased inventory velocity and reduced operating costs for trading
partners.
As more companies adopt Flowcasting as a way of replenishing
inventory across retail supply chains, new modes of distribution will
emerge. Manufacturers, for example, will be able to collaborate and
combine their efforts to reduce the cost of distribution and increase
product velocity (see Figure 2.10).
63
Figure 2.10: Collaborative distribution model.
Such collaborations will make it possible for manufacturers to

partner with other manufacturers and deliver to a common customer
at a specific store. Transportation volumes for a group of manufac-
turers could also be combined by store, affording additional
economies. For example, imagine a group of ten stores which has a
combined volume for one week that adds up to ten trucks. Say there
are ten manufacturers involved in the collaboration and each has the
equivalent of one full (or nearly full) truck to ship weekly to this
group of stores. The Flowcasting system communicates the store
orders to all parties involved. Trucks are bulk loaded at the factories
and shipped to an RDC or to a third party logistic provider (3PL) loca-
tion. The trucks are then unloaded and the products are cross-docked
to store specific trucks. The stores receive their deliveries on the day
requested, and each truck will have a mix of product from each of the
ten participating manufacturers.
In large manufacturing companies where products are manu-
factured in multiple factories, collaboration will be done internally to
deliver to a common customer at a specific store. This will entail
using a corporate portal, as shown in Figure 2.11.
64
Figure 2.11: Corporate portal model for large manufacturers.
Making It Happen: The Flowcasting Team of the Future

For Flowcasting to succeed, retailers, wholesalers, and manufactur-
ers will need to create Flowcasting teams. The teams’ mandate will be
to put new business processes in place. These new business process-
es will support dynamic models of complete product flow from the
time products leave factories to the time it ends up on the store
shelves. In other words, a whole new world of communications and
collaboration will evolve. Retailers will create data repositories, as
Wal-Mart and K-Mart did when they created Retail Link and
Workbench. These retailer data repositories will contain:
• daily POS, daily store, and DC on hand balances and

in transits
• daily store and DC cycle count adjustments
• replenishment lead-times
• minimum store shelf and DC safety stock quantities
• minimum store and DC ordering quantities
• shipping schedules
65
• new store openings

• new product introductions
• product deletions
Retailer data will flow daily from their repository into the
Flowcasting systems, which will contain distribution patterns (facto-
ry to MDC to RDC to store, for example) for every product found on
retail store shelves. Flowcasting will use the distribution patterns to
model the total product flow from factories to store shelves, thus
serving as a bridge between the retailer and the wholesaler or manu-
facturer’s ERP system. Figure 2.12 offers a preview of the information
flow.
The Composition and Role of Flowcasting Teams

We anticipate that Flowcasting teams will consist of a retailer team
and a supplier team. The retailer team will include representatives
Figure 2.12: Communication patterns in a Flowcasting environment.
66
from the stores, distribution, category management and purchasing.

If the supplier is a wholesaler, the team will include representatives
from sales, customer service, demand planning, logistics, and pur-
chasing. And if the supplier is a manufacturer, the team will be made
up of representatives from sales, customer service, demand planning,
logistics, and manufacturing.
The Flowcasting teams will have two roles. The first will be
making sure that foundational data is accurate, complete, and up-to-
date. By foundational data, we mean the data repository mentioned
previously. It consists of daily POS, inventory balances, lead-times,
shelf configurations, safety stocks, MOQs, and shipping schedules.
Specific team members will continually monitor and update the
pool of foundational data as business conditions change. The updat-
ing will be critical; if the goal is to create accurate models of the way
a company wishes to do business, the foundational data must mirror
the current environment as well as the rules and schedules used to
manage the business on a day-to-day basis.
The second role of the flowcasting team will be to jointly devel-
op and agree upon (the retailer with the wholesaler or manufactur-
er) store-level sales forecasts that also include promotion planning,
product introductions, and product discontinuations.
Finally, Flowcasting teams will act as a buffer for the new busi-
ness paradigm. You can’t overhaul the data structure of a retail sup-
ply chain overnight anymore than you can turn an aircraft carrier on
a dime. There will be pockets of resistance as well as the normal fears
of change that accompany the introduction of new ways of doing
business, no matter how much better the outcome may be. Part of
the methodology entails preparing people for change. And change
there will be! The enormous visibility provided by Flowcasting will be
the precursor to the demise of forward buys and product diversions.
In addition, over time we will see the reduction of back rooms in
retail stores, and a reduction in the number of warehouses and dis-
tribution centers across retail supply chains.
67
Summary
In this chapter, we demonstrated the power of Flowcasting, a time-
phased system for planning the flow of product throughout the entire
retail supply chain. Key points include the following:
1. True uncertainty only exists at the final point of sale: the

retail store. Uncertainty can be eliminated by Flowcasting all
product movement throughout the supply chain, beginning
at the retail store.
2. Flowcasting provides unparalleled supply chain modeling
capabilities, made possible by dependent demand.
Flowcasting introduces new opportunities to plan capacities
for people, space, and equipment from the same numbers
used to purchase and replenish products into retail DCs. The
flowcasting projections from Flowcasting can be converted
into desired currencies as well as into units of capacity for
people, space, and equipment, thereby providing a much-
needed bridge between manufacturing and retail trading
partners.
3. Flowcasting introduces new approaches to demand manage-
ment and resource planning in a retail company. It enables a
retailer to operate the business from a single set of numbers.
4. In the future, Flowcasting teams will introduce new business
practices and processes that will ultimately transform how
retailers, wholesalers, and manufacturers interact to achieve
improved customer service and greater profitability.
This concludes Section 1 of this book. In the second section,

we’ll focus on the details of Flowcasting and explain how Flowcasting
is used in actual retail settings. You’ll quickly see that Flowcasting is
far more than a theoretical construct. It’s here. And companies are
using it to gain a competitive advantage, today.
68
Flowcasting
Section 2:
Flowcasting the Retail
Supply Chain:
From Store to Factory
69
Flowcasting
70
C H A P T E R 3
It All Begins at the Store

The Only Forecast You’ll Ever Need
➫
I n the previous chapters we introduced a revolutionary concept:
forecasting can be done at the retail store level, and the results of
a store-level forecast can be used to drive the entire supply chain in
a time-phased manner. The results of such Flowcasting are impres-
sive, and can include massive lead time reductions (70 percent or
more for some hard lines), a doubling of inventory turns, and major
reductions (30 to 50 percent) in inventory investment. And these
benefits accrue while improving customer service levels to as high as
99 percent. Because Flowcasting is a completely different approach,
the "service/inventory tradeoff" of old no longer applies. Simply put,
there can be major improvements in both.
In the next three chapters, we’ll take a closer look at how the
numbers all "foot" – how the bottom line numbers of the retail store
forecast become the top line inputs to the DC forecasts, and how the
bottom line DC forecasts become inputs to forecasts at the manufac-
turing plants. This chapter focuses on where the forecasting all starts
-- the retail store. Before we delve into the specifics of forecasting at
the retail store, it’s important to understand how the fundamental
nature of operational forecasting changes in a Flowcasting environ-
ment. While the basic act of forecasting remains the same – using
current data and history to predict the future based on patterns – in
71
a Flowcasting environment, there is a fundamental shift regarding

which customer is being forecasted.
Rethinking Forecasting: Meet The "New" Customer

In today’s supply chains, the first big challenge in operational fore-
casting is to answer the question "Who’s my customer?" The com-
mon sense answer, of course, is "My customer is the end consumer -
- the end consumer is the reason I’m in business." In reality, most
partners in an extended supply chain (especially those who are
upstream from the store) have split loyalties. As much as they’d like
to devote all of their forecasting energy to understanding the end
consumer, the fact remains that consumers don’t buy from manu-
facturers -- distributors do. A factory’s activity is driven directly by
activities in the manufacturing DCs. Likewise, the activity in a retail
distribution center is driven by the store’s ordering behavior. In
short, the various partners in the supply chain must create their own
forecast because, practically speaking, the answer to the question,
"Who’s my customer?" is different for each of them (see Figure 3.1).
Figure 3.1: Forecasting in a traditional retail supply chain.
In a traditional retail supply chain, each partner has its own invento-
ry, lead-time offsets, policies, and constraints. As a result, everybody
spends a great deal of time and effort trying to forecast the demands
of the immediate customers. This leaves little time to really think
about the end consumers, because the link to them is indirect at best.
As much as a manufacturer would like to stay on top of store-level
demand for their products in real time, the fact of the matter is that
the orders that will fulfill consumer demand in a few weeks need to
72
be shipped to the DCs today -- that’s the split loyalty. The manufac-
turer is forced to try to forecast the needs of its DCs and its customers
DCs explicitly, just to keep the supply chain running.
By contrast, in a Flowcasting environment, all of the forecast-
ing focus is squarely on the end consumer. By modeling the "chain
reaction" of demand from the store shelf back to the factory floor, it’s
no longer necessary to forecast the needs of the next supply chain
partner/customer. The reason for this is that the visibility afforded by
a Flowcasting process allows us to answer the question "Who is the
customer?" separately from "Where to ship?" (see Figure 3.2).
Figure 3.2: How Flowcasting separates demand from supply.
Flowcasting provides high-quality projections that can be

translated and rolled up to any level in the supply chain. All product
flow needs are unambiguously scheduled around the consumer, and
all partners’ inventory, lead-time offsets, policies, and constraints are
explicitly modeled around projected consumer demand. The net
result is that a great deal of unnecessary forecasting is eliminated, as
shown in Figure 3.3. This, in turn, means that individuals responsi-
ble for creating forecasts can work on more value-added activities
within their areas of expertise, rather than spending time trying to
predict the future.
If you’re skeptical about whether forecasting can really take
place at the retail store level, consider this: there has always been
forecasting at the retail store level -- it just hasn’t been articulated.
There’s a reason that a retail store orders 10 units of an item and not
100 units when it’s time to replenish the shelves for that item. The
decision comes from "knowing" the product, the season, and other
73
Figure 3.3: Some of the forecasting activities that can be eliminated by Flowcasting.
critical factors. The problem is that, until recently, even if people

could articulate their forecasting rationale, there was simply no way
to practically do so for every item in the store on a fifty-two week
horizon. In other words, the problem hasn’t been particularly com-
plex, just very large. With low-cost computing power readily avail-
able, and software that can efficiently crunch millions of SKUs, the
size issue becomes a technicality. This brings us to the central issue
of this chapter: just how do you forecast at the retail store level?
Of Plans and Forecasts

Before delving into the actual steps used to forecast at the retail store
level, let’s review the context in which forecasting takes place. In
Chapter 2, we outlined a series of high-level business processes that
all retailers perform to one degree or another (see Figures 3.4 and
3.5). Company business plans are developed in dollars. Next, overall
merchandise category strategies are developed that support the busi-
ness plan. These high-level plans are then translated into supply
chain demand plans -- product specific projections of what we expect
consumers to buy.
In most retail organizations, these plans are created in the
aggregate and are totally disconnected from day to day operations. In
74
Figure 3.4: Demand Management components of a Flowcasting system.
a Flowcasting world, Demand Management activities (Baseline

Forecasting, Promotion Forecasting, Pricing and Assortment
Changes) take place at the most atomic level, where the only true
demand exists -- by product by store in units. As a result we can take
our product/store level forecasts and aggregate them to the category,
region, or national levels in any unit of measure desired. Therefore,
Flowcasting makes it possible for a retail organization to work from
one set of numbers and drive the entire supply chain.
After the product/store level demand plans have been created
and projected well out into the future, all of the resources employed
from the point of manufacture to the final point of sale (including
inventory, labor, space, equipment, and cash) are aligned to the ulti-
mate purpose of satisfying consumer demand -- from the retail store
shelf right back to the factory.
The remainder of this chapter will illustrate how a basic
demand plan for a single item at a single store feeds into a store level
replenishment plan for that item. Subsequent chapters will build on
this illustration and fully explore the other aspects of Demand
75
Figure 3.5: Resource Planning components of a Flowcasting system.
Management and Resource Planning at all echelons of the retail sup-

ply chain.
As you read on, you will find the logic of Flowcasting to be very
simple, intuitive, and easy to understand. And as your level of under-
standing deepens, you will learn to appreciate just how powerful it
can be to plan the entire retail supply chain with a single store-level
consumer forecasting process.
Retail Store-Level Forecasting in Action

While there are numerous specific software systems available to help
you with the forecasting process, you’ll need to carry out five basic
steps to build a time-phased projection that can be used to drive the
entire retail supply chain:
Step 1: Get your foundational data in order
Step 2: Forecast consumer demand at the store shelf
Step 3: Plan the first store replenishment
Step 4: Plan future store replenishments
Step 5: Calculate planned orders to support the
arrival-based plan
76
Each step is described in detail below.
Step 1: Get Your Foundational Data in Order

The old saw, "garbage in, garbage out," rings true with Flowcasting
and any enterprise business process. For Flowcasting to generate a
single forecast that can be used to drive the entire supply chain, it’s
critical to have the right data in the right form at the right time. In
other words, the foundational data must be in good order, and should
include:
• Daily POS
• Daily store and DC on hand balances and in transits
• Stores’ and DCs’ replenishment lead-times
• Minimum store shelf displays and DC safety stock quantities
• Minimum store and DC ordering quantities
• Shipping and receiving schedules
Daily POS
Most retailers today use barcode scanners to track daily sales by item.
Ideally, an historical archive of two to three years’ worth of scan data
should be available as inputs to the Flowcasting process. The data
should be tracked by product and by store, in daily or weekly time
periods. Also, each item would be scanned discretely by size, color,
flavor, and other key replenishment attributes.
Daily Store and DC On Hand Balances and In Transits

Every stocking location – whether at the retail store or the DC --
must have a system-calculated perpetual inventory balance that fac-
tors in all sales, receipts, and inventory count adjustments. Ideally,
the system-calculated inventory balance at any location will be 95
percent accurate2 or higher when compared to a physical count. In
transits are quantities that are already in route; that is, they have left
2 The accuracy measure is stocking location specific. 95 percent accuracy means

that the system calculated inventory balance at item/location level matches (with-
in a reasonable tolerance) the physical count 95 times out of 100. See Appendix B
for more information on how to achieve high levels of inventory record accuracy.
77
the source, but have not yet been received into stock at the destina-
tion.
Replenishment Lead-Times
For each product, source location, and destination location, the
repository should store the elapsed time from the time an order is
received at the source until the time the product is received at the
destination. For example, if a retail store can get a delivery 3 days
after placing an order on its DC, then the lead-time between the store
and DC is 3 days. Similarly, if the DC can get a delivery 10 days after
placing an order with a supplier, then that is the lead-time between
these points.
Minimum Store Shelf Display and DC Safety Stock Quantities

In either case, for each product/location, this represents the lowest
level to which you would ever want your inventory balance to fall, in
units. At store level, this may mean the minimum amount of stock
required for the display to be attractive. For example, if it’s desirable
to have a minimum of 4 facings at least 2 units deep for a particular
product at all times, then the minimum shelf display quantity3 for
that product would be 8 units. At the RDC level, attractiveness is not
an issue. Instead, safety stock would be used to hedge against vari-
ability in the demand from the stores.
Minimum Store and DC Ordering Quantities

For each product at each location, the minimum order quantity rep-
resents the smallest amount of product that will be ordered, in sell-
ing units. A retail store, for example, may sell items individually to
consumers, but order cases of 12 units from the RDC. When the RDC
orders from the MDC, however, it may do so in pallet or truckload
quantities for the same item to maximize efficiency. For simplicity in
3 In a majority of cases, the minimum required mathematical safety stock (based

on demand variability) to maintain a 95 percent service level at the retail store
shelf is significantly less than the minimum display requirements calculated by
shelf management systems. Flowcasting makes this information highly visible
and, over time, retailers will have opportunities to work with suppliers to reduce
manufacturing packs by exploring repackaging and remerchandising opportuni-
ties.
78
the upcoming examples, we assume that each location also orders in

multiples of the minimum ordering quantity, but in practice it’s not
uncommon to have multiples set differently from the order mini-
mums.
Shipping and Receiving Schedules

Many locations may only have specific days when they are open for
shipping or receiving. This information is required to make sure that
product movements are planned to happen only on the days these
activities are allowed. For example, while a retail store may be open
for business 7 days a week, its receiving crew may be on duty only
Tuesday and Thursday mornings. Similarly, DCs may ship or receive
product only Monday through Friday, at non-peak times. Holidays
and shutdowns should also be incorporated into the schedules.
Again, for simplicity in our examples, we only focus on receiving
schedules, but shipping schedules can be used as well.
This information may come from a variety of sources (e.g., on
hand balances and in transits may come from the ordering system,
whereas the shipping and receiving schedules may reside with the
transportation department), but it represents a minimum require-
ment for getting a Flowcasting process running. If you don’t already
have this information available to run your current replenishment
process, then this is the first thing to work on.
Step 2: Forecast Consumer Demand at the Store Shelf

For purpose of illustration, let’s consider the task of forecasting
demand for a single item in a single store. Our goal is to predict how
many units of that product we will sell in that store each day for the
next 52 weeks. The demand forecast will consist of three compo-
nents:
• The baseline forecast stream is a forecast of what you
would expect to sell on an ongoing basis.
• The promotional forecast stream contains forecasts of
expected demand increases (or decreases) due to
promotional activity.
• The local judgment stream contains forecasts of expected
demand increases (or decreases) due to local events.
79
The Baseline Forecast Stream

The first step in creating a baseline forecast stream is to retrieve 2 or
3 years’ worth of POS history from the foundational data repository
for the item/store being forecasted. If this data is stored in daily time
periods, it is aggregated to weekly time periods.
Once the weekly POS history stream is created, it can be fed
into pattern-based time series forecasting software4 that:
1. Filters out the impact of any unusual historical demand

periods, such as those caused by past promotions, large
special orders, out-of-stocks, and weather-related events
2. Determines if there is a positive or negative sales trend
3. Detects a repeating pattern over a one-year business cycle
caused by the inherent influence of changing seasons on
that item/store combination
4. Develops a computational model that will project the his-
toric pattern (including trend and seasonality) 52 weeks or
more into the future
As illustrated in Figure 3.6, the software generates a weekly

forecast of "what would normally happen" in the absence of the effect
of promotions and special events.
Most time series forecasting systems also allow users to over-
ride the calculations if the trend or seasonality doesn’t seem reason-
able based on assumptions about future market conditions. In gen-
eral, the more good POS history data that’s available, the less the
planner will need to intervene.
But what if an item has been recently introduced and no POS
history exists? A common approach in this instance is to simply copy
history from a similar item, with a volume scaling factor applied, if
necessary.
It’s important to reiterate that the baseline forecast stream
modeled from history only exists at the product/store level. Because
4 Many forecasting packages of this type exist on the market. While they can vary
in complexity and user functionality, they all have some method of performing
these basic calculations.
80
Figure 3.6: Illustration of how a baseline forecast stream is generated from

POS history.
we are working solely with pure end consumer demand at the store
shelf, there will be specific forecasting rules and guidelines that will
apply (see Appendix D for more details).
The Promotional Forecast Stream

A baseline forecast provides an initial forecast every week for a year
into the future. Because it’s computer generated, it takes little effort
to produce all the numbers. To the extent that future trends and sea-
sons will unfold in a similar way as the past, the baseline forecast is a
good start. In the case of promotions, however, we know in advance
that the future will be different from the past. This is where human
intervention is required.
In the example in Figure 3.6, we have promoted the item on dif-
ferent schedules in each of the prior 3 years and will be promoting it
again this year. Once the details about the upcoming promotions
have been decided, then a forecast can be created. Most retailers plan
promotions in the following way for any given item (promotional
planning will be covered in more detail in Chapter 6):
1. Finalize the overall promotional plan (schedule, pricing,
advertising, etc.) for the item.
81
2. Review and analyze the POS sales results from past promo-
tions for the item.
3. Develop new promotional consumer forecasts for the item,
ideally in collaboration with suppliers and a sampling of
store managers.
4. Allocate the forecast to participating stores, based on each
store’s volume contribution by item or category.
Once this planning process has been completed, the item/store

promotional forecasts are overlaid onto the baseline forecast, as
shown in Figure 3.7.
Figure 3.7: Illustration of an integrated baseline and promotional forecast.
The Local Judgment Stream

Like the promotional forecast stream, local judgment is a change to
the baseline forecast based on knowledge that the future will be dif-
ferent from the past. The difference is that this information is sup-
plied at the local level, rather than derived from a national or region-
al number. Examples of local judgment would include in-store spe-
82
cials, markdowns, and community events. These are specific to a par-

ticular item in a particular store or group of stores and therefore only
known at the item/store level.
Putting It All Together

Once the baseline forecast stream has been appropriately modified to
reflect what is planned in the future, we will have a complete picture
of what we expect our POS sales to be for the item/store each week
for a year into the future (see Figure 3.8).
Figure 3.8: Expected POS sales.
Creating Daily Forecasts

Recall earlier that our goal was to predict how many units of that
product we will sell in that store each day for the next several weeks.
We’ve generated the weekly numbers by using a baseline forecast and
factoring in promotional plans and local market knowledge. The last
step is to take our total weekly forecast numbers and allocate them
to each day within the week. This can be done in many ways, but a
simple and sensible approach is to apply daily percentages based on
how sales and traffic generally unfold in the particular store. For
example, a store that is open 7 days a week may have an overall sales
and traffic profile in any given week similar to the one shown in
Figure 3.9:
Figure 3.9: Sample weekly traffic profile.
83
In this store, 45 percent of the weekly sales happen on the

weekend. Monday is the lightest day in the week, representing only 5
percent of total weekly sales.
By applying this simple logic to the total forecast derived earli-
er (see Figure 3.8), we can generate the daily sales projection shown
in Figure 3.10:
Figure 3.10: Converting weekly forecasts into daily projections.
Now that we have a daily demand forecast for this item/store,

we are ready to plan the rest of the supply chain for this item. As will
be demonstrated in the remainder of this book, this forecast will
drive all other planning activity in the supply chain -- it’s the only
forecast you will ever need.
At this point, you may be wondering about the amount of effort
that will be required on the part of Flowcasting teams to manage a
store-level forecasting process -- particularly when we extend the
example to include hundreds of stores and thousands of products
across a year-long time horizon.
What makes it possible is the fact that properly designed
Flowcasting tools will do most of the "grunt work" (generating a 52-
week mathematical baseline forecast from POS history) after they
have been initially set up. In addition, the people responsible for fore-
casting within the Flowcasting teams will work by exception, only
reviewing forecasts that are tracking out of tolerance. And finally, as
you will see in the following chapters, this one forecasting process at
store level completely replaces all other forecasting activities in the
84
extended supply chain -- not only for replenishment purposes, but for
capacity planning, resource scheduling, and budgeting as well.
Step 3: Plan the First Store Replenishment

When trying to determine what a given store will need to order, a
forecast is a good start. After all, you need to know how much of an
item you’re going to sell before you can decide how much to replen-
ish! But replenishment decisions aren’t based solely on a sales fore-
cast. You need additional store/item information, including:
• Current on hand balance and in transits

• Replenishment lead-time from RDC to store
• Minimum store shelf display quantity (time-phased,
if necessary5)
• Minimum store ordering quantity
• Store receiving schedule
Let’s step through the process by first organizing our informa-

tion in an example (see Figure 3.11).
Figure 3.11: A typical

retail supply chain.
5 An example illustrating the use of time-phased minimum display quantities is

shown in Figure 3.12.
85
The remainder of this chapter will illustrate how the flow of

demand, supply, and inventory for a particular product between store
108 and RDC1 can be planned for the next 52 weeks using a single
forecast of demand at store level. (In Chapters 4 and 5, this illustra-
tion will be extended all the way back to the factory, and Appendix A
will show how this same planning process can be applied to other
network configurations, such as web retailing, direct-to-store/VMI ,
and centralized distribution.)
Figure 3.12 shows the results of retrieving the following key
information from the foundational data repository pertaining to
product 01234567 in store 108:
• The current on hand balance
• The minimum store ordering quantity
• The store’s receiving schedule
• The replenishment lead time from order to receipt from
RDC1
• The planned minimum display quantities,
time phased into the future
• The open in transits that are en route from RDC1
In addition, we have generated a total forecast of demand for
this item/store over the next year as described in Step 2. We now have
all the information we’ll need to plan a year’s worth of replenish-
ments to the store from RDC1.
Notice that the minimum display quantity will increase to 48
units on day 7 and drop back down to 12 units on day 14. That is
Figure 3.12: Total forecast and basic replenishment information for an

item/store.
86
because the item will be on promotion and an end aisle display of 36

facings will be used in conjunction with the normal 12 facing shelf
slot. On day 7, we are planning to set up our end aisle display one day
in advance of the start of the sale. On the last day of the sale (day 14),
we plan to empty our end aisle display for this item, so we revert back
to our standard 12 facing shelf.
Determining When More Inventory Is Needed

We now have enough information to figure out when we need the
first replenishment to arrive at the store (see Figure 3.13). We can do
so by taking our Current On Hand Balance and subtracting our Total
Forecast for each day, until we ascertain the day that we’ll drop below
our Minimum Display Quantity. This is accomplished through the
use of Projected On Hand, which takes the known current on hand
balance and, similar to a bank account, simulates what the future
balances will be based on when inventory will be deposited (supplied)
or withdrawn (demanded) at the location.
Figure 3.13: Determining when more inventory is needed.
In Figure 3.13, you can see that our current on hand balance is
18 units. On day 1, we are forecasting that we will sell 1 unit, leaving
us with a Projected On Hand of 17 units at the end of that day.
Our forecast is to sell an additional 2 units on day 2, but we also
have an in transit quantity of 12 units that’s expected to arrive on
that day from RDC1. Therefore, our projected on hand at the end of
day 2 is 17 – 2 + 12 = 27 units, and so on. Using this logic, we can see
87
that our projected on hand will drop below our minimum display
quantity on day 7, if more stock does not arrive on, or before, this day.
So now we also know than on, or before, day 7, we will need to sched-
ule a Planned Arrival of stock into store 108.
Determining How Much Inventory Is Needed

We now know that we need more product to arrive by day 7. We have
also determined that our projected on hand on that day will be 8
units, which is a shortfall of 40 units below our minimum display
quantity of 48 on that day. Therefore, the quantity we need to arrive
on the 7th day is 40 units, which will ensure that we don’t drop below
the minimum display quantity. However, our store only orders this
product in cases of 12 (the minimum ordering quantity), so we have
to round up our quantity to 48 units (see Figure 3.14).
Figure 3.14: Determining how much inventory is needed.
In Figure 3.14, we can see that if we plan for 48 units to arrive

on the 7th day, our projected on hand will again be above our mini-
mum display quantity.
Our store, however, is only open for receiving on Mondays,
Wednesdays, and Fridays (as indicated by the white boxes). Since we
will drop below our minimum display quantity on the 7th day, we
must plan our arrival of 48 units on the first available receiving day
before our needed arrival date. This would be on day 5 (see Figure
3.15).
88
Figure 3.15: Scheduling the first planned arrival.

Based on our revised arrival date, our projected on hand bal-
ance on day 5 has been recalculated as follows:
In bank account fashion, the projected on hand has been recal-

culated for days 6 and 7 as well, based on the new schedule.
Step 4: Plan Future Store Replenishments

At first blush, the replenishment approach we’ve just described (sub-
tracting demand from available on hand and in transits) may sound
similar to a reorder point system. The key differentiator between a
standard reorder point system and the initial calculation in a
Flowcasting system is the calculation and tracking of the projected
on hand (projected time-phased inventory balance) as described in
the previous step.
89
It is the projected on hand that allows us to extend the replen-

ishment logic for as far out into the future as a forecast exists and to
continually critique open purchase orders (in transits) and re-plan
future planned arrivals. Instead of merely knowing what will be on
the next order, we now have a planning system that will tell us for
each and every day over the next year:
• How much we plan to sell
• How much we will need to order
• How much inventory we will have on hand
In addition, retailers need only one system to plan the flows of

inventories, replenishments, and purchases for all stores and RDCs -
- rather than one system for the stores and another for the RDCs.
This will be covered in greater detail in Chapter 4.
To illustrate the power of keeping track of our projected on
hand balance, consider where we left off in Figure 3.15. We have only
calculated our plans out until the end of the 7th day, but there’s no
need to stop there. Since we are keeping track of our projected on
hand day by day, we can extend our replenishment logic out as far as
the forecast allows. Starting with our projected on hand of 56 units
at the end of day 7, we can simply pick up from where we left off.
If we have calculated a projected on hand of 56 units at the end
of day 7, and we think we will sell 14 units on day 8, we’ll have only
42 units left over at the end of day 8. Furthermore, the store won’t be
able to receive any product on day 9, so we need to cover an addi-
Figure 3.16: Determining the timing of the second replenishment.

90
tional 28 units of sales, which will put our projected on hand balance
at 14, as illustrated in Figure 3.16.
As Figure 3.16 shows, the arrival on day 8 needs to be at least
34 units, in order to ensure our projected on hand is at least 48 at the
end of day 9 (see Figure 3.17).
Figure 3.17: The second planned arrival.
But since we order in multiples of 12, the arrival in day 8 needs

to be rounded up to 36. As a result, our projected on hand at the end
of day 8 is now 78 units.
In like fashion, the remainder of the forecast horizon can be
planned (see Figure 3.18):
Figure 3.18: A 52-week arrival-based plan.
Notice that, at no point, does the projected on hand dip below

the minimum display quantity. We’ve now worked out each and every
time stock will need to arrive at store 108 for the next year.
91
Step 5: Calculate Planned Orders to Support the Arrival-

Based Plan
Now that we have determined when inventory will need to arrive at
the store, it’s a simple matter to determine when orders will need to
be placed on RDC1 to meet the plan. This is where the replenishment
lead-time comes into play.
If it takes 2 days for product to arrive at the store after it has
been ordered from RDC1, then we simply need to subtract 2 days
from each store’s arrival date to determine when it will need to be
ordered in the future (see Figure 3.19).
Figure 3.19: A 52-week ordering and arrival-based plan.
In Figure 3.19, if we want 48 units to arrive on day 5, then it

will need to be ordered from RDC1 on day 3. The difference repre-
sents the 2-day replenishment lead-time between the store and
RDC1. If this store were to get shipments from another location (e.g.,
another RDC or even a manufacturer’s MDC) that is further away,
steps 1 through 4 (figuring out when stock will need to arrive) would
be exactly the same. The only difference is how far in advance the
store would need to place orders.
6 Once the replenishment lead time is reached, there is only enough time left to
get it on the truck at RDC1 and transport it to the store. At this point, the ship-
ment is considered to be en route to the store, so it’s too late to change the dates
or quantities.
92
The planned orders will continue to be recalculated until they

are within the replenishment lead-time, at which point they will
become firm in transits.6
Continuous Re-Planning
Of course, not all plans will happen exactly as expected. On any given
day, you may sell more than your forecast or you may sell less – and
if you get it exactly right, it’s probably a coincidence. That’s why, in
order to Flowcast properly, the foundational data needs to be
refreshed daily. On a net change basis7, if any of the foundational data
has changed for the item/store, then the rest of the planning steps
will be executed again, resulting in a plan that not only stretches out
an entire year, but is constantly being kept up to date with the most
recent information.
Recall, for example, the plan we just created for Product
#01234567 in store 108 (see Figure 3.19). Suppose that on day 1,
instead of selling the 1 unit that was forecast, you actually sold 5. In
other words, you oversold your forecast by 4 units. Here’s what would
happen: after day 1 had passed, you would begin a new planning cycle
(day 2). Between day 1 and day 2, your Current On Hand Balance
would have dropped from 18 units to 13 units. Assuming that the
forecast for the next 52 weeks doesn’t change overnight (typically the
forecast would change weekly if required), the Flowcasting logic
would completely refresh the plan as shown in Figure 3.20):
Because the forecast was oversold on day 1, the impact of this
change automatically updates the entire 52-week planning horizon,
starting now from week 2. In this particular example, the planned
arrival on day 8 increased from 36 units to 48 units as a result.
Flowcasting the Entire Supply Chain

So far we have used a simple example of planning a single item in a
single store to illustrate the Flowcasting concept. In fact, the same
7 Net change planning is what makes huge retail data volumes manageable.
The time-phased plans will only change if one of the input variables changes.
Therefore, it’s only necessary to re-plan product/locations that have experienced a
change from one day to the next.
93
process (on a much larger scale, executed by computer) can be used

to plan replenishment projections for the rest of the products and
locations in the retail supply chain. This can be done with no addi-
Figure 3.20: A recalibrated 52-week plan after overselling day 1.
tional forecasting beyond the store-level forecast of consumer

demand discussed in this chapter!
Summary
We have illustrated how Demand Management and Resource
Planning converge at the retail store level in the supply chain, and
have discussed some of the key inputs required to make Flowcasting
work. We also described a high-level forecasting process used to pre-
dict consumer demand at the store level. We demonstrated how,
using this forecast, on hand balances, planned arrivals, and planned
orders can be time phased into the future. Here are the key points to
remember:
1. Every retail supply chain exists to fulfill the demands of the

consumer. In a traditional retail model, the consumer buys
products from stores, so it only makes sense to focus all
Demand Management activities at store level.
2. The first step in making Flowcasting work is to have an up-
to-date repository of common demand and logistics data at
the store and RDC level, such as: POS history, inventory bal-
94
ances, lead-times, store shelf display configuration, safety

stock, order minimums, and activity schedules.
3. In a Flowcasting environment, all demand forecasting is
done only at the store level, in selling units, and by item, by
day for several weeks and months into the future. Starting
with baseline forecasts generated by a computer, modifica-
tions can be made both top-down (for activities such as
national promotion and assortment strategies) and bottom-
up (for local promotions and events), as required.
4. By using the aforementioned forecasts as a foundation, some
simple real-world constraints can be applied to generate
time-phased projected flows of arrivals, orders, and invento-
ry balances for the entire forecast horizon for any item at any
store.
5. Outside the replenishment lead time, the time-phased
planned flows are fluid and are continually updated with
fresh information. In this way, new information from con-
sumers is being input directly into the store (and as we’ll see
later, the entire retail supply chain) to make sure that no plan
is more than 24 hours out of date.
The following chapter extends the Flowcasting process to the

next echelon in the supply chain. It demonstrates how RDCs can be
planned with the same forecast that drives the stores through the
simple power of dependent demand.
95
Flowcasting
96
C h a p t e r 4
Supplying the Stores

Dependent Demand and
Replenishment at the Retail DC
➫
T o paraphrase the old saw, "no RDC is an island." The RDC’s activ-
ities are wholly dependent on the needs of the retail stores.
Without stores, there would be no need for RDCs to exist. Yet today,
most retailers today have one system for replenishing their stores
and another for replenishing their RDCs. Flowcasting reconnects
the stores and RDCs, as they should be connected. This means more
than drawing from a centralized data repository; it means that the
final replenishment plans from the stores actually become the top-
line inputs for the RDCs. In this way, the forecasts cascade through
the first two levels of the supply chain, starting with the only point of
true demand: the store shelves.
In this chapter, we’ll extend the logic described in Chapter 3
from the store level to the RDC. As in the examples used in Chapter
3, we’ll focus on an individual item. Remember, it doesn’t matter
whether we’re forecasting one SKU or 10,000 SKUs; the logic
remains the same. The Flowcasting process crunches the numbers
and makes the complexity manageable.
Passing the Demand

Before diving into forecasting at the RDC level, we’ll review the
process of forecasting at the retail shelf level, as described in Chapter
97
3. In a Flowcasting environment, replenishment demand and supply

planning at store level for any given item requires 5 steps:
1. Get your foundational data in order.
2. Forecast consumer demand at the store shelf.
3. Plan the first store replenishment.
4. Plan future store replenishments.
5. Calculate planned orders to support the arrival-based plan.
The output of this activity is a store-level plan for when product

needs to be ordered from the RDC to support demand for the next
year into the future. Let’s return to the fully calculated example from
Chapter 3, for store location 108:
Figure 4.1: A 52-week ordering plan for product 01234567 in store 108.
If store 108 was the only retail sales venue in the supply chain,
life would be simple. But, of course, supply chains are dynamic enti-
ties consisting of numerous nodes. For the sake of illustration, we’ll
expand the model (see Figure 4.2) to include two stores to demon-
strate how Flowcasting can tightly integrate retail stores and RDCs
within the supply chain.
In Chapter 3, we determined the replenishment needs of store
108 (Figure 4.1). But to compute RDC1’s shipping requirements,
we’ll also need a time-phased plan for another retail node in the sup-
98
Figure 4.2: Using Flowcasting to integrate retail stores and retail DC.
ply chain, store 602. We can determine store 602’s requirements by

deploying the same process used to calculate a plan for store 108, but
instead using store 602’s unique demand forecast, supply constraints,
and lead-time:
Figure 4.3: A 52-week ordering plan for product 01234567 in store 602.
99
Now that we’ve calculated the planned orders for the next 52
weeks for stores 108 and 602, it’s very easy to determine the demand
that RDC1 will need to satisfy over that time period. All we need is
the following information:
• Daily DC on hand balances and in transits

• RDC replenishment lead time (from the manufacturer DC)
• RDC safety stock quantities (minimum desired inventory
balance)
• RDC minimum order quantities
• RDC shipping and receiving schedules
With that information in hand, we can plan replenishment for

RDC1 the same way that we planned replenishment for the stores,
with one critical difference: no additional forecasting will be
required to generate a time-phased inventory and replenishment
plan for RDC! Remember, the bottom line numbers from the store
replenishment plans cascade to the next level of the supply chain and
become the top-line numbers for the RDCs. This makes it possible to
determine RDC supply requirements through three easy steps:
Step 1: Calculate RDC shipment projection

Step 2: Calculate arrival-based plan for the RDC
Step 3: Calculate planned orders to support the arrival-based
plan
Step 1: Calculate RDC Shipment Projection

By using the Flowcasting process at each store, the work of deter-
mining when product will need to ship from RDC1 has already been
done. The planned order stream from each store represents how
much the stores want shipped and when.
To calculate RDC1’s shipment projection, you just need to add
up the Planned Orders for each store (see Figure 4.4, which shows
the Planned Order lines for stores 108 and 602. The full details for
stores 108 and 602 are shown in Figures 4.1 and 4.3, respectively).
100
Figure 4.4: RDC1 shipment projection calculated by summing store planned

orders
The total store planned orders line now becomes the dependent
demand that RDC1 must satisfy.
Step 2: Calculate Arrival-Based Plan for the RDC

With a demand projection that’s been given by the stores, a purchas-
ing plan for RDC1 can now be calculated, through a process similar
to the store-level planning described in the previous chapter.
Figure 4.5: Dependent demand and basic replenishment data for RDC1.
In Figure 4.5:
• The Minimum DC Ordering Quantity represents the mini-

mum amount that will appear on a purchase order between
RDC1 and MDC1. In this case, the volume through RDC1 for
product 01234567 is sufficient to order in pallet quantities
of 144 units.
• The DC Receiving Schedule describes when the receiving
department is open for business at RDC1 (also denoted by
the white boxes in the Planned Arrivals line).
• The Replenishment Lead-Time from MDC1 represents the
101
amount of time required for product to be received into

stock at RDC1 after a purchase order has been created on
MDC1.
• The Store Planned Orders line represents the sum of the
Planned Orders line for all stores that are serviced from
RDC1. This is dependent demand that is used in place of a
shipment forecast for RDC1.
• The Safety Stock line represents the minimum level of
inventory for this item that we want to hold at RDC1 to
cover for demand uncertainty.
The terminology is slightly different, but the process is the

same for calculating an arrival-based plan for an RDC:
Starting with a current on hand balance of 250 units, we’ll
determine when the projected on hand is planned to fall below the
safety stock. This is the date when the first planned arrival is needed
(day 6, in our example, as shown in Figure 4.6).
Figure 4.6: Determining when the first arrival is needed at RDC1.
As you can see, if no stock arrives on, or before, day 6, the

inventory at RDC1 will drop to -26 units. Since our safety stock
should be 100 units, 126 units will be required. That number will be
rounded up to 144 units, the minimum ordering quantity.
One other factor must be taken into account: RDC1 is not open
for receiving on day 6, which means that the first arrival of 144 units
102
will need to be scheduled for day 5, the first available date prior to
when the arrival is actually needed. Figure 4.7 below summarizes
this simple logic:
Figure 4.7: Scheduling the first arrival at RDC1 to maintain projected on hand
above safety stock.
By calculating and tracking the projected on hand at RDC1, we
have the ability to continue planning forward for as far out as we have
Store Planned Orders. Figure 4.8 below illustrates a 52-week arrival
based plan for RDC1:
Figure 4.8: A 52 week arrival based plan for RDC1.
Step 3: Calculate Planned Orders to

Support the Arrival-Based Plan
As in the case of planning for the retail stores used in our ongoing
example, the bulk of the planning activity for RDC1 has been dedi-
103
cated to determining when supply needs to arrive in order to support

the dependent demand. We know that the replenishment lead-time
(from order to receipt into stock) is 4 days between RDC1 and MDC1.
We can, therefore, create RDC planned orders from the planned
arrivals by offsetting the planned arrival dates ahead by the lead-time
of 4 days, as illustrated in Figure 4.9.
Figure 4.9: A 52 week ordering and arrival based replenishment plan at RDC1.
As Figure 4.9 shows, on day 1, RDC1 will need to order 144 units
from MDC1 to ensure that it arrives on day 5, and so on through the
year.
We now have a complete 52-week plan for RDC1, including
planned store demand, projected on hand balances, planned arrivals
into stock, and planned ordering activity. All three were determined
without having to forecast requirements at RDC1. Put another way,
the item/store level forecast of consumer demand that we created in
Chapter 3 is now directly driving two full echelons of the supply
chain.
Whether you use a Flowcasting business process or not, there is
a chain reaction of demand from consumers through to the factories.
Consumer buying behavior directly impacts store replenishment
which, in turn directly impacts RDC replenishment. By modeling
this reality explicitly, not only does the amount of energy required for
forecasting decrease, but the plans will be far more accurate as well.
104
The Chain Reaction of Demand

Because the store and DC echelons are being planned in an integrat-
ed fashion by using dependent demand -- and all within the same sys-
tem – the impact of changing the single input variable (the consumer
demand at store level) will be clear at both the stores and the RDC.
For example, starting in week 1, our integrated plans for stores 108
and 602 and RDC1 would be as shown in Figure 4.10.
By using time-phased planning, the complete chain of events
that needs to take place to satisfy consumer demand is simulated in
advance. If the forecast is oversold or undersold at store level, the
entire 52-week plans are recalculated and shared with RDC1, which
in turn recalculates its planned orders to MDC1. At any time, the
impact of yesterday’s POS sales at store level will be reflected in
today’s plans at both the stores and the RDCs.
Figure 4.10: The chain reaction of demand from consumers to retail

stores to the RDC.
105
What If: Re-Mapping Demand to Supply

Flowcasting makes timely, robust planning feasible and practical, and
gives all trading partners in a retail supply chain a powerful tool for
predicting and meeting demand in a cost effective way. That ability
makes Flowcasting an essential tool for managing in the real world,
where change is a constant and it’s often necessary to make decisions
"on the fly." For example, with traditional means of managing sup-
ply chains, hours of laborious analysis, reforecasting and order real-
location are required to facilitate network changes. With
Flowcasting, such activities can be done effortlessly simply by updat-
ing the system with the new information.
Think about the model of the supply chain that we have been
building in this chapter and the previous chapter. It extends from the
retail store shelf to the retail DC. Because the model begins with a
forecast of consumer demand at the store shelf, downstream depen-
dent demand can be re-directed through different network configu-
rations as business needs dictate.
Suppose that an analysis of the network reveals that trans-
portation and inventory costs would be better optimized if, say, store
760 were supplied from RDC1, as illustrated in Figure 4.11.
Figure 4.11: A network change resulting in store 760 being replenished from RDC1.
106
In a traditional supply chain environment (without

Flowcasting), the impact of this network decision would be signifi-
cant for the following reasons:
• Historical demand for RDC1 prior to the network change

would be understated, as it would not include orders from
store 760. Historical demand for RDC2 similarly would be
overstated, as it would include past orders from store 760. As
a consequence, it would be necessary to restate demand his-
tory for both RDC1 and RDC2 for every product impacted by
the change and reforecast at both places as well.
• It would be necessary to revise existing promotional shipment
forecasts that were entered into RDC1 and RDC2 before the
network change.
• Because of the changes in demand rate, it would be necessary
to adjust inbound orders created before the network change
for RDC1 and RDC2.
When you consider that every product in store 760 would be

affected by this change, this could represent days of work, not to
mention the potential for making many errors.
By contrast, in a Flowcasting environment, it’s simple enough
to re-plan the entire supply chain following a network change. That’s
because a) there is a single demand forecast -- the one showing con-
sumer demand at the store shelf and b) the network change is only a
supply decision that does not impact demand and therefore requires
no change to the forecast.
Basically, all that’s required is to map store 760’s planned orders
to RDC1, as shown in Figure 4.12
Figure 4.12: RDC1 shipment projection revised to include planned

orders from store 760.
107
With demand re-mapped to a new network path, all that we

need to do is recalculate the plan for RDC1 (and RDC2) using the new
streams of Store Planned Orders and the procedures described in
Steps 2 and 3 earlier in this chapter. Figure 4.13 shows a revised
replenishment plan using this approach.
Figure 4.13: Revised replenishment plan for RDC1 using new store
planned orders stream.
Once a decision has been made to change the network, it can be

implemented in a matter of minutes -- not days or weeks -- and the
entire supply chain will be updated accordingly.
The ability of a Flowcasting system to map demand at will vast-
ly simplifies other supply chain challenges as well, including:
• Adding new RDCs to the network or removing existing RDCs

from the network
• Opening new stores, or closing existing stores
• Temporarily re-directing demand to alleviate overstocks
• Simulating the short-term operational impact of planned net-
work changes at every node
• Planning direct-to-store deliveries or VMI programs with sup-
pliers
• Carrying out multi-echelon replenishment (ordering in bulk
into one RDC from overseas, then breaking up the order for
shipments to other RDCs for distribution to stores)
108
Again, this remarkable level of flexibility is made possible

through less forecasting, not more.
It’s clear that Flowcasting can vastly streamline and simplify a
retail enterprise. But to realize the full benefit of Flowcasting, the
process must be extended from the retail shelves all the way back to
the factory. Since most retailers rely on other entities (with their own
separate planning systems) to manufacture the products they sell, a
standardized form of communication is required to manage the myr-
iad supplier relationships within a Flowcasting environment. The
vehicle for doing so is called "supplier scheduling."
Supplier Scheduling
The output of the Flowcasting process at RDC1 (and any other RDCs)
is a projection of what suppliers must sell and ship to the retailer, 52
weeks into the future (i.e., a stream of Planned Orders). By sharing
the Planned Orders with wholesalers and manufacturers, the retailer
can eliminate the need for these suppliers to forecast what the retail-
er is going to buy – information about demand is disseminated and
refreshed daily as conditions change in retail stores and RDCs.
Flowcasting also not only reduces forecasting efforts, but it pro-
vides vastly superior forecast quality at the downstream echelons.
Less effort is required because the entire supply chain is planned with
the consumer demand forecast at the store shelf. The forecast quali-
ty is superior, because every constraint and inventory pool upstream
from the consumer is considered in the projection -- there’s no need
to guess the demand at each echelon of the supply chain.
The sharing of planned orders is commonly known in manu-
facturing and distribution circles as Supplier Scheduling, and will be
discussed in more detail in Chapter 5. Even without further discus-
sion, it should be obvious that supplier scheduling represents a radi-
cal departure from the way retailers, wholesale distributors, and
manufacturers have traditionally conducted business.
Summary
In this chapter, we extended the Flowcasting business process from
the retail store to the RDC echelon in the supply chain. We demon-
109
strated that planning replenishment at the RDC is very similar to

planning for demand at a retail store. Here are the key points to
remember about managing in a Flowcasting environment:
1. There is only one echelon in the supply chain where con-

sumption occurs. Until a product is taken from the store
shelf, all other inventory movements prior to that point are
simply value-added transfers that are dependent on the store-
level consumption. Hence the term "dependent demand" for
consumer demand that will cascade through the supply
chain.
2. As long as all of the retail stores serviced by an RDC are cal-
culating and sharing planned orders, the RDC receives its
demand plan "for free" -- no additional forecasting is required
beyond the store level consumer demand forecast discussed
in Chapter 3.
3. While the terminology is different, the Flowcasting tech-
nique used at the RDC is the same as that used in the store.
This entails:
i. Determining when supply must arrive to satisfy the
demand
ii. Offsetting the arrivals by the replenishment lead-
time to determine when product needs to be
ordered from the next echelon in the supply chain.
4. Because dependent demand (in the form of store planned
orders) is being shared between the retail stores and the
RDC, the impact of changes in consumer demand at the store
shelf can be seen at the retail stores and the RDC -- virtually
in real time.
5. The shared 52-week replenishment plans at the stores and
DCs are not static. Every day, the information is updated, and
the chain reaction of demand is cascaded downstream and
recalculated for the next 52 weeks into the future.
6. Once store-level consumer demand forecasts exist, demand
can be flexibly mapped and re-mapped to support any supply
network configuration with minimal effort.
110
7. The sharing of highly relevant and directly actionable infor-

mation with suppliers will allow retailers, wholesalers, and
manufacturers to eliminate unnecessary forecasting from
the supply chain. This, in turn, will not only reduce costs and
improve supply chain dynamics, but it will completely trans-
form business relationships for the better.
The next chapter moves the Flowcasting process outside the

retailer's domain and into the supplier's distribution and manufac-
turing processes. It shows that through supplier scheduling, the ben-
efits of sharing dependent demand are not confined within organiza-
tional boundaries.
111
Flowcasting
112
C h a p t e r 5
Supplying the Retailer

Wholesale Distribution and
Production Scheduling
➫
S o far, we’ve seen how Flowcasting links retail stores and RDCs
through a single store-level forecast. The key is dependent
demand, which specifies how much product must be shipped and
when is the product needed, taking into account all relevant con-
straints. That information is "actionable" at every node in the retail
supply chain, which means that the dependent demand can be used
directly by the next partner upstream, with no translation or inter-
pretation necessary.
In this chapter, we’ll describe how dependent demand can
extend beyond the retailer’s organization (stores and RDCs) and
transform business relationships and planning processes with whole-
sale distributors, manufacturers and raw material suppliers. It will
also describe how manufacturers that utilize ERP systems can inter-
nalize Flowcast-generated dependent demand provided by key cus-
tomers. If all this seems like a tall order, bear in mind that
Flowcasting reduces uncertainty between nodes in the supply chain.
And the more you can reduce uncertainty, the less the need for indi-
vidual nodes to guess (forecast) what’s needed today, tomorrow, and
a year out.
113
Flowcasting for Upstream Supply Chain Partners

Recall how, taking the Store Planned Orders (dependent
demand) from store 108 and store 602 as givens, we were able to
Flowcast planned orders from RDC1 on MDC1 (See Figure 5.1).
Figure 5.1: The network elements of RDC1’s product flow plan.
The completed product flow plan for RDC1 based on this net-
work is shown in Figure 5.2.
Figure 5.2: A 52-week flow plan at RDC1.

114
But as we can see from Figure 5.1, a planned order stream also
needs to be created for RDC2 so that MDC1 can have complete visi-
bility into its dependent demand. To do so, we need to create a
Flowcasting plan for RDC2 in addition to the one for RDC1. First, we
collect and sum the planned order streams from the stores that are
serviced by RDC2 (see Figure 5.3):
Figure 5.3: Calculating total dependent demand at RDC2.
As in the case of RDC1, the total store planned orders stream

now becomes the "top line" of RDC2’s time-phased plan. With RDC2’s
replenishment data, the Flowcasting process can be employed to pro-
duce a complete product flow plan for RDC2 (Figure 5.4):
Figure 5.4: A 52-week flow plan at RDC2.
With RDC2 now calculating planned orders in addition to

RDC1, we can continue the chain reaction of demand down to MDC1.
In the same way that the RDCs calculate their dependent demand by
summing store planned orders, MDC1 can calculate its dependent
demand by summing up planned orders from RDC1 and RDC2 (in
115
our example RDC1 and RDC2 are the only two customers of MDC1 -
- see Figure 5.5):
Figure 5.5: Calculating total dependent demand at MDC1.
Once again, this dependent demand stream from the RDCs

becomes the enabler for the Flowcasting process in MDC18. With a
dependent demand stream given to it, MDC1 is simply another dis-
tribution node that can be planned in exactly the same way as RDC1
and RDC2 (see Figure 5.6).
Figure 5.6: A 52-week flow plan at MDC1.
116
Figure 5.7: Integrating MDC1 with the factory.
The Planned Orders line in MDC1’s plan can then be given to

the factory to plan production as shown in Figure 5.7. While the fac-
tory is the last trading partner in the supply chain, the Planned
Orders generated by the Flowcasting system will still drive a number
of processes within the manufacturer’s operation -- think of them as
"sub-nodes" driven by dependent demand. We’ll now turn our atten-
tion to the processes affected by Flowcasting within the factory.
While the stores, RDCs, and MDC all redistribute finished
goods, it is the factory that sits at the end of the retail supply chain,
charged with the responsibility of producing the products that will
eventually reach the consumers’ hands. Rather than passing a stream
8 Our example assumes that MDC1 is a distribution center of the manufacturing

company, but this need not be the case. MDC1 could also represent a third-party
wholesaler/distributor that sits between the manufacturer and the retailer.
Because each node in a Flowcasting supply chain produces the same type of
universally actionable information (how much is needed and when must it ship),
it does not matter how many different organizations are involved in getting the
product from the factory to the store shelf.
117
of planned orders to the next node in the retail supply chain, the fac-
tory instead takes its dependent demand from MDC1 and produces a
Recommended9 Master Production Schedule (MPS)10.
The now familiar Flowcasting method of planning supply to
meet demand is no different for a factory than it was in the retail
stores or distribution centers discussed in Chapters 3 and 4. The only
difference is a change in terminology, specifically:
• "Manufacturing Lot Size" represents the amount of product
that is made in a single production run and is used instead
of "Minimum DC Ordering Quantity."
• "Manufacturing Lead Time" represents the amount of time
needed to produce a single lot and is used instead of
"Replenishment Lead Time."
• "Work in Process" represents product that has begun a pro-
duction run but not yet finished. It is used in place of "In
Transits."
• "MPS Available" represents when finished goods are needed
to be available off the production line and is used instead of
"Planned Arrivals."
• "MPS Start" represents when production needs to begin in
order to have finished goods available by the MPS Available
date and is used instead of "Planned Orders."
Because the factory in our example runs lean, it only produces

what it needs to ship, and no safety stock is planned.
Assume for purposes of this example that the recommended
MPS is feasible (we have an agreed-to level of production for a spe-
cific product stated in specific times and quantities and both required
9 We say "Recommended" because the determination of what needs to be pro-

duced, how much, and when, involves the need to plan required labor and equip-
ment capacities as well as materials needed to support production. This planning
requires a balancing act between the availability of both materials and capacities.
Consequently, human beings are required to create this critical balance and main-
tain it through a master schedule policy supported by executive management.
10 The subject of Master Production Scheduling is extremely important but is

beyond the scope of this book. Over the years, the accumulated body of knowledge
on MPS has been well documented. The American Production & Inventory
Control Society (APICS) is an excellent resource for information on MPS and
related topics.
118
materials and capacities are or can be made available). With the MPS
stated and agreed upon within a given planning horizon, the pur-
chasing department within the factory knows when production runs
for the 01234567 Deluxe Widget are planned to start. By taking this
projection and multiplying it through the bill of material (BOM) for
the Deluxe Widget, the factory can plan its purchases of the raw
materials that will be necessary to support the master production
schedule .
For example, suppose that a finished 01234567 Deluxe Widget
is comprised of the following components:
• 6 oz. blue plastic pellets
• 1.5 ft. nylon tubing
• 3 metal washers
• 1 packages boxes
That means each production run of 576 Deluxe Widgets will

require the following:
• 216 lbs (3,456 oz.) blue plastic pellets
• 864 ft. nylon tubing
• 1,728 metal washers
• 576 packages of boxes
With the quantities and timing provided by the "MPS Start" line
as shown in Figure 5.8 and the conversion table provided by the
Figure 5.8: A 52-week recommended master production schedule (MPS) at the factory.
119
Figure 5.9: Using the MPS and BOM to drive dependent demand for
raw material components.
BOM, we now know when raw materials and packaging supplies will
be needed by the factory to produce the Deluxe Widgets11 (see Figure
5.9).
We have now closed the loop by creating a 52-week plan of
demand, supply, and inventory for every retail store, distribution cen-
ter, and factory in the supply chain. And remember, all of these plans
were generated without any forecasting beyond the forecast we cre-
ated at store level in Chapter 3.
While everything that has been described from the beginning of
Chapter 3 until now may seem intuitive, it is assumed that the com-
munication links between nodes have already been firmly estab-
lished. Between retail stores and RDCs, communication is relatively
easy to set up, since all of these nodes are generally a part of the same
company. Within the retailer’s organization a single, self-contained
system would be used to plan both the stores and the RDCs simulta-
neously.
11 In a factory that produces several products sharing some (or all) of their raw
materials or component parts in different proportions, the MPS will need to be
exploded through all bills of material before a purchasing plan can be calculated.
120
When it comes to the manufacturing and wholesale distribu-

tion activities, however, multiple organizations come into play as
shown in Figure 5.10. At the beginning of this chapter, we stated that
the process must not end with the retailer; for Flowcasting to become
a reality and change the way retailers and their suppliers do business,
the information must extend from the point of sale back to the pro-
curement of raw materials at the factory. But when organizational
lines are drawn between functions in the supply chain, it becomes
necessary to have a standardized, efficient way of sharing actionable
information. In order to plan the supply chain seamlessly (that is, as
if those organizational lines didn’t exist), retailers need to speak the
language of their suppliers, and vice-versa. This can be accomplished
through a process called "supplier scheduling," which is the topic of
the next section.
Supplier Scheduling
Without the Flowcasting process, the manufacturer must inde-
pendently determine the items and quantities it must ship out to
meet the retailer’s demands. Specifically, it must figure out which
Figure 5.10: Typical lines of ownership and the role of supplier

scheduling in the retail supply chain.
121
days it must ship on and the location is must ship to, several weeks
or even months into the future. This is done today at great time and
expense through key account sales forecasting and order booking.
But as we’ve already seen, planned order streams can be effort-
lessly calculated at any level in the supply chain with a Flowcasting
business process. And no matter where you look in the supply chain,
planned orders always represent the same thing: which items need to
ship, the quantities needed, and the specific ship dates and locations,
52 weeks into the future. Flowcasting answers all of these questions
-- what, where, how much, and when, at the required level of detail.
And the information produced at one echelon in the supply chain
requires no additional interpretation by the next.
In a Flowcasting world, the demand requirements on the sup-
plier have already been scheduled within the retailer’s planning
process (hence, the term "supplier scheduling"). The ramifications of
having this information on tap are highly significant for all suppliers:
• No longer will manufacturers need to guess about what their

retail customers will need; the retailers will tell them what
they are planning to buy and when for the next several
months, thereby eliminating the need for manufacturers to
forecast sales and shipments for key accounts. This will serve
to completely redefine the role of the manufacturer sales rep
(see Appendix H).
• No longer will manufacturers be blindsided by demand
shocks that increase in urgency and severity as they ripple
through the supply chain over a period of several weeks. The
moment something unexpected happens at the retail shelf,
the chain reaction of demand is refreshed, and all trading
partners in the supply chain (including the manufacturer)
can see the precise impact on their operations the very next
day.
• No longer will manufacturers need retailers to lock in their
orders weeks or months in advance as a buffer against forecast
error. By virtue of the fact that retailers explicitly state their
intentions many weeks into the future, demand uncertainty
122
for the manufacturer is drastically reduced, if not eliminated

completely. Buffer can be stripped out of lead-times, allowing
postponement of product flow until the last possible moment.
In a Flowcasting environment, the ordering lead-time
between retailers and manufacturers represents only the time
required for picking, packing, loading and shipping to the des-
tination.
• No longer do retailers and manufacturers behave as indepen-
dent fiefs with disparate planning approaches. The fact that all
entities use the same Flowcasting process means that the
lines between retailer, wholesaler, and manufacturer dissolve;
for satisfying demand, they act as one entity with the sole pur-
pose of getting product into the hands of consumers.
Transfers of product ownership are appropriately treated as
mere administrative formalities.
• No longer are supply issues uncovered and discussed when it’s
too late to do anything about them. Implicit in the act of shar-
ing supplier schedules for joint planning is the notion that
silence implies agreement. In other words, in exchange for
keeping the supplier schedules up-to-date and accurate, the
retailer can expect the supplier to use this information, inter-
nalize it and spot potential supply problems while there’s still
an opportunity for contingency planning. Flowcasting teams
(with representative from both the retailer and the supplier)
will solve potential problems rather than react to problems
that have already happened.
Supplier Scheduling Redefines Business Relationships

Supplier scheduling is far more than a process for efficiently moving
data among trading partners. It represents a drastic transformation of
the way the trading partners work together and do business. Over
time, familiarity and trust breeds efficiency between supply chain
partners, thereby eliminating the need for retailers to constantly
change supply sources on the basis of who offers the lowest price this
week. With the shared Flowcasting information in hand, the trading
partners can turn to other opportunities for lowering costs, such as
123
renegotiating order minimums, reducing lead-times, and reducing

inventory along the entire extended retail supply chain.
When partners share projections in the form of supply sched-
ules, legal purchase transactions become "non events." At the end of
every day, any planned orders that are within the lead-time release
horizon could be sent to the retailer’s transaction system for auto-
matic purchase order creation. Suppliers receive constantly updated
schedules from their customers, so by the time the purchase order is
received, the product has already been made and just needs to be
shipped.
As the long-term retailer/manufacturer relationship strength-
ens and matures, the need to have a purchase contract for every ship-
ment will diminish. Instead, companies will annually agree on busi-
ness terms (pricing, order quantities, lead-times, etc.), which will be
reflected in the supplier schedules by the retailer. On a day-to-day
basis, manufacturers will automatically ship everything that’s on the
supplier schedules within the agreed-upon lead-time. Tracking num-
bers may be generated for each shipment, but formal purchase orders
will become a thing of the past. At the end of the month, a single
statement, detailing the month’s shipping activity, will be sent to the
retailer for payment.
Simplifying the transactions and removing red tape will enable
the retailer and manufacturer to focus on new ways of improving
their processes and satisfying customers. (See Chapter 11 for a spe-
cific example of supplier scheduling, its financial benefits, and the
elimination of purchase orders and invoices.)
Interfacing Supplier Schedules into the Supplier

Processes
Our example thus far assumes that each inventory pool in the supply
chain will have 100 percent visibility into its immediate customers’
demands. For a typical retailer that owns both its stores and RDCs,
this is likely to be the case. Between the stores and the RDCs, the
supply chain is self contained.
124
But what about the manufacturers and wholesalers? Most often,

a manufacturer will have many different retailers as customers for its
products. To the extent that some of these retailers provide supplier
schedules and others do not, the manufacturer will be have multiple
demand streams:
In Figure 5.11, a distributor/manufacturer supplies product to
three different retail customers. Each of these retailers has a differ-
ent planning approach:
• Retailer A operates in a complete Flowcasting environment.

Based on a consumer demand forecast, the stores create a 52-
week stream of planned orders. These planned orders become
the demand stream at the RDCs which, in turn, create 52
weeks of planned orders. The planned orders from the RDCs
are shared with the Distributor/Manufacturer as supplier
schedules.
Figure 5.11: A distributor/manufacturer receiving different inputs from its retail

customers.
125
• At the other end of the spectrum, Retailer C uses a tradition-

al ordering strategy at the stores and the RDCs12. No visibility
exists beyond the next order between any two supply chain
entities. The Distributor/Manufacturer is required to forecast
Retailer C’s demand explicitly.
• Retailer B is in transition to a full Flowcasting environment.
Like Retailer C, its stores send only immediate replenishment
requests to the RDCs. Like Retailer A, the RDCs produce a
stream of planned orders which are, in turn, shared with the
MDC/factory as supplier schedules, but the planned order
stream is based on a forecast of total demand from the stores,
not dependent demand from the stores.
From the Distributor/Manufacturer’s point of view, both
Retailer A and Retailer B are sending the same type of actionable
time-phased information. However, Retailer A’s supplier schedules
will be of much higher quality, because all inventory and constraints
are being considered at the lowest level of granularity (from the store
shelf back). Retailer A has virtually removed uncertainty from the
most important supply chain echelon.
So, how can a Distributor/Manufacturer benefit from
Flowcasting, given the "mixed bag" of information it will receive from
its retail customers (that is, some retailers provide supplier schedules
and others don’t)?
As Figure 5.11 shows, Retailer A and Retailer B both share sup-
plier schedules (of differing quality). Therefore, the Distributor/
Manufacturer no longer needs to forecast these accounts. Instead,
the only forecasting effort required is for Retailer C, who does not
share a supplier schedule. The approach for planning MDC1 would
include a mixture of summing dependent demand from retailers who
provide supplier schedules and forecasting shipments for retailers
who do not, as illustrated by the "Total Retailer Demand" line in
Figure 5.12.
12 At the RDCs, a traditional reorder point strategy would likely be used. At

store level, orders may be generated manually based on a visual shelf review or
with a simple reorder point system as well.
126
Figure 5.12: Planning MDC1 for a mixed demand stream.
From the manufacturer’s point of view, the ideal situation

would be for all of its retailer customers to provide supplier schedules
that represent 100 percent of their demand, not only for their man-
ufacturing and distribution needs, but for Sales and Operations
Planning (S&OP) as well (S&OP will be discussed in more detail in
Chapter 11). How likely is this to happen? As Flowcasting gains
acceptance within the retail community, more of the manufacturers’
demand will be handed to them in the form of supplier schedules.
And as this plays out over time, a critical mass will be reached.
Manufacturers will begin to transform their businesses from make-
to-stock to make-to-order; every activity, from raw material acquisi-
tion, selling, and distributing finished goods, will be subject to review
and changed for the better. It is not a question of if, but when.
While the inventory and service-level benefits of the
Flowcasting approach provide retailers with ample incentive to
adopt, suppliers can hasten this adoption by agreeing to protect sup-
ply for customers that operate in a Flowcasting environment.
Available To Promise and Protected Supply

When retailers invest in the ability to provide supplier schedules to
their partners, they create a true win/win situation. By treating
planned orders from retail customers as promised orders, the suppli-
127
er has a much clearer picture of what can be sold to other customers

without harming its long-term relationships with retailers who pro-
vide schedules.
A good way to do this is to use Available To Promise (ATP), a
capability that ensures quality order promising in a make-to-order
business. ATP is an excellent approach for protecting the supply of
retailers who provide schedules. In addition, with time-phased infor-
mation at hand, the ATP calculation can be continually recalculated
for as far out as needed to make decisions.
Consider the example in Figure 5.12. We know that Retailers A
and B both provide planned order streams to MDC1, while Retailer C
does not. We also know that the current on hand balance at MDC1 is
650 units and that in transit quantities and planned arrivals from the
factory are scheduled well out into the future. With this information,
we are able to do a time-phased cumulative available to promise cal-
culation for deluxe widgets at MDC1.
In Figure 5.13, the Promised Orders line represents the sum of
the planned orders from Retailers A and B only. Retailer C is exclud-
ed from this calculation because it does not provide a schedule. Since
the supplier is forced to forecast its demand, supply will not be pro-
tected for Retailer C.
Figure 5.13: Calculating cumulative Available To Promise (ATP) at MDC1.
The Planned Supply line represents all scheduled supply,

including in transits and planned arrivals in MDC1’s Flowcasting
plan. The Available To Promise line represents supply that will be
available for MDC1 to sell to other customers, after Retailer A and B’s
demands have been met.
In our example, our current on hand is 650 units and an addi-
tional 576 units is in transit, so our Planned Supply on the beginning
of day 1 is 1226 units. The next Planned Supply is 576 units, due to
arrive on day 4.
128
Our Promised Orders (that is, supply that has been protected
for Retailer A and Retailer B) between now and day 4 is 432 units (288
on day 2 and 144 on day 3). The difference between the Planned
Supply and Promised Orders over the next 3 days represents the
Available To Promise: 1226 – 432 = 794.
Put another way, if a customer other than Retailer A or Retailer
B wants product from MDC1 over the next 3 days, MDC1 can only sell
794 units to that customer, even though they will have 1226 units in
supply.
Because the Promised Orders and Planned Supply are calculat-
ed over a long time horizon in a Flowcasting environment, the ATP
calculation can be extended as far out as is needed. For example, on
day 4, additional supply of 576 units is planned, and the Promised
Orders are 288 units. The ATP in the previous period was 794 units.
So the cumulative ATP on day 4 is 794 + 576 – 288 = 1082, and so on.
The key point is that a Flowcasting relationship between part-
ners requires a high level of trust and co-operation to make it work.
If retailers make the necessary investments to provide schedules to
suppliers, they will expect a high level of service with no excuses from
suppliers. Conversely, if suppliers are to reserve inventory without a
firm purchase contract locked in, they expect retailers to follow their
schedules.
Flowcasting and CPFR

Those who are familiar with CPFR will recognize the steps that com-
prise CPFR have been totally institutionalized as part of the day-to-
day activities of the Flowcasting business process. Some of the word-
ing may be somewhat different, but the steps and objectives are iden-
tical. To recap Chapters 1 through 5, thus far, you have seen how:
1. Flowcasting teams (Retailer and Manufacturer or

Wholesaler) have been created.
2. The teams have met, and Collaborative Arrangements and
Joint Business Plans are set and agreed upon.
3. Product flow plans (from the factory floor to the store shelf)
have been developed and agreed to.
129
4. Store-level sales forecasts have been created, aggregated,

and agreed to.
5. Store and retail DC inventories have been considered.
6. Inventory targets, service-level objectives (required safety
stocks), lead-times and minimum ship quantities and trans-
portation requirements have been considered.
7. Resulting supplier schedules have been internalized and
integrated (S&OP, DRP, MPS, and ATP) by the manufactur-
er.
8. Constraints (production, distribution, and transportation)
have been identified and resolved, and supplier schedules
are now protected for delivery to the retailer.
9. Exception Management and Performance Assessments are
ongoing.
The difference between Flowcasting and CPFR is the following:

With CPFR, you can have a working arrangement with one or more
trading partners and continue to do business as usual with all other
trading partners. If you are a retailer, Flowcasting will drive all your
forecasting, inventory management, and replenishment activities. If
you are a manufacturer, to get the full benefits of Flowcasting you
must internalize the processes and integrate them into what we call
the four main touch points. The last three touch points below were
addressed in this chapter, and the first touch point, Sales &
Operations Planning, is covered in Chapter 11:
1. Sales & Operations Planning (S&OP)

2. Distribution Resource Planning (DRP)
3. Master production Scheduling (MPS)
4. Available-to-Promise (ATP)
With a Flowcasting business process in place, trading partners

have built a complete product flow model of the way they have agreed
to do business inside the computer. This model now resides inside
the same system, and both the retailer and the manufacturer or
wholesaler have access to this ecosystem. We cannot think of a bet-
130
ter way of doing business and have trading partners jointly

Collaborate, Plan, Forecast, and Replenish their complete supply
chains from the factory floor to the retail store shelf.
Summary
In this chapter, we "closed the loop" by continuing the same
Flowcasting process that began in Chapter 3 beyond the walls of the
retail organization to their upstream supply partners. We demon-
strated that the inventory planning problem can be universally solved
with the same planning method, whether the operation is a retail
store, a distribution center or a factory. Here are the key points to
remember:
1. The concept of dependent demand is not limited to a partic-

ular type of trading partners; it’s relevant to every echelon
in the supply chain.
2. Store-level forecasts of consumer demand can actually drive
every planned product movement, from the extraction of
raw materials to the production of finished goods through
the distribution network and into consumers’ hands.
3. Supplier scheduling is the bridge that will connect
Flowcasting team members from separate organizations. It
represents a fundamental change in the business relation-
ships between retailers and their suppliers.
4. The visibility provided by retailers in the form of supplier
schedules requires that retailers and manufacturers act in
true partnership -- reducing cycle time and waste, simplify-
ing terms and conditions, and making long-term commit-
ments on the demand and supply side.
5. Each retailer that comes on stream with supplier schedul-
ing is one more key account that the supplier no longer
needs to forecast.
6. By respecting supplier schedules from retailer customers
that provide them, suppliers can more accurately calculate
their Available To Promise over a long time horizon. By pro-
131
tecting supply in this manner, suppliers can reward retailers

for sharing their planning information.
7. When a critical mass of retailers adopt the Flowcasting busi-
ness model, manufacturers can transform their operations
from make-to-stock to make-to-order. For manufacturers
with a small number of large customers, this may happen
very quickly.
This concludes Section 2 of this book. In the third and final sec-
tion, the proverbial rubber will "hit the road." The chapters in
Section 3 show how Flowcasting can be used to solve some of retail's
most common (and most frustrating) challenges: managing promo-
tions, product introductions and discontinuations, seasonal prod-
ucts, slow moving items, and operational and financial planning.
Finally, we present a high-level plan for implementing Flowcasting in
today's networked retail supply chains.
132

Flowcasting The Retail Supply Chain PDF

Uploaded by

Copyright:

Available Formats

Flowcasting The Retail Supply Chain PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flowcasting The Retail Supply Chain PDF

Uploaded by

Copyright:

Available Formats

0 introduction 4/4/06 1:00 PM Page 1

Winning the Retail

numerous surveys conducted over the past 10 to 15 years regarding

Steering Linkage for the Retail Supply Chain

Software packages for Flowcasting, in contrast, are specifically

The Flowcasting business process works as follows:

5. Flowcasting enables the communication of replenishment

The implications for retail supply chains are significant.

1. Retailers and manufacturers can "flowcast," rather

activities are completely eliminated at every node of a given supply

2. Better planning and execution of promotions.

3. A common language for all trading partners in the supply chain.

the flow of product in a time-phased manner, product will be pulled

4. Linking the supply chain to true consumer demand.

5. Instant business simulations, better reaction time.

casted on the basis of real-time consumer demand. What better way

6. Cutting costs at the core.

Figure 1: Benchmarking Sales Forecasting Performance by Kenneth B Khan,

As shown in Figure 1, our ability to forecast tends to improve at

Figure 2: Simulation Results for Retailer X.

The level of forecast accuracy ranged from 83 percent to 97 per-

Hard evidence now exists to prove that sales forecasting accu-

this introduction: it’s time to start thinking of Flowcasting as the

The Flowcasting Imperative

tial information you need to know to adapt Flowcasting effectively in

The 21st Century

pendently, almost as if they’re discrete islands or fiefs, blind to each

PART 1: FORECASTING IN A VACUUM

Every Node for Itself

Figure 1.1: A typical retail supply chain.

The Retail Nodes

Figure 1.2: Retailer’s functional forecasting activities.

The Manufacturing Nodes

The Wrong Stuff

Figure 1.3: Manufacturer’s functional forecasting activities.

ven by internal metrics means that they do not mirror changes in

PART 2: A BETTER WAY OF DOING BUSINESS

Doing—and Forecasting—Business Your Way

we just forecast the way we conduct business?" The question is so

Figure 1.4: Dependent demand in a retail environment.

Suppose that when store 108 needs products, it orders from

One Retail Supply Chain, One Forecast

1. "What am I going to sell?"

As shown in Figure 1.5, the first question need only be

Figure 1.5: One sales forecast only at store level.

The answer to each of the subsequent questions lies in calcu-

forecasting at the RDC level explicitly, even though you’re dealing

Figure 1.7: Shelf resets at the store level.

1. Traditional forecasting in retail, wholesale, and manufactur-

result, all trading partners experience higher operating costs.

The next chapter introduces a new concept, "Flowcasting,"

During the 1950s and 1960s, companies also began to make

advances in technology and forecasting have made Flowcasting a

The Power of One (Forecast)

The Flowcasting Concept

Figure 2.2: Conversion of planned receipts into receiving hours.

5) are sent to suppliers. By using this approach, suppliers would no

Figure 2.3: Conversion of planned shipments into shipping hours.

Figure 2.4: Total capacity plan for an RDC.