$2.$ The demand planning process

The demand planning has to be done for both the local market $($Italy$,$ France, Germany$)$ and the export $($USA$,$

Canada, UK$)$; for these reasons, there are three San Pellegrino demand planners dedicated to forecast the

customer demand in the local market, and other three assigned to the latter. For each market $($local and

export$),$ demand forecast $($and communication of the information along the chain$)$ is done differently for the

different product families $($i$.$e$.,$ Waters, Beverages, Juices$).$

In particular, each product typology is distinguished in models $($e$.$g$.,$ for beverage: tea and orange juice$)$; for

each model, units are grouped according to the capacity and the material of the bottle in which they are sold

$($e$.$g$.,$ PET $16\mathrm{.}9$ oz; Glass $22$ oz; Can $12$ oz$...).$

A different forecast is generated, within the same family, for products with similar packaging.

For example, among beverage typologies, a common format is PET bottle of $16\mathrm{.}9$ oz each: the demand

planner group historical data about the demand of all the beverage products sold in this package, aggregate

them and generates a forecast according to this aggregated demand. The same happens with PET bottle of

$12$ oz each: all the beverage products sold in this package are clustered, historical data collected, and forecast

generated on their aggregated demand $($see Sheet $1$ of the Excel file for detail$).$

In both cases, the forecasting process for the next four months $($t$+4)$ starts one month in advance $($t$-1),$ and

numbers are generated having "week" as a unit of analysis.

SCM $301$ Introduction to Supply Chain Management $3$

Figure $2$: San Pellegrino Demand planning organization

$3.$ Demand management for PET $12$ oz Beverages

As the main objective of $2018,$ the demand planning department wants to improve the nature and the

characteristics of the demand for PET $12$ oz Beverage in the Local market, to generate a more reliable forecast

model.

Demand for PET $12$ oz has always shown unpredictable patterns, which is critical because beverage of this

size can be considered "not systematic" purchases from customers, meaning that customers tend to buy the

same brand, but they have no problems in switching to another one in case the product they are looking for

is not present on shelf.

For this reason, on one side San Pellegrino Supply Chain needs to keep an eye on cost$,$ given the fact that

beverage products usually have margins no higher than $5\%/10\%($the average price paid by the customers for

a Sanpellegrino bottle is $$1\mathrm{.}13$; Supply Chain cost for purchasing, production and distribution are, on average,

$$0\mathrm{.}95,$ but also be sure the product is highly available, as a stock $$ out immediately translates in a lost sale

$($which can be a long term lost sale if the customer decides not to repurchase Sanpellegrino products after

having tried other ones$).$

To do that, data about quantity requested by customers during the last $104$ weeks $(2$ years$)$ have been

collected $($see Sheet $2$ of the Excel file for detail $$ data are expressed in $,000$ bottles$).$

You are asked to support the demand planning department in doing this.

SCM $301$ Introduction to Supply Chain Management $4$

Questions:

$1.$ According to the information provided in the text, comment about the level of aggregation chosen

for the demand management process

Market: Low Medium High

Product: Low Medium High

Time: Low Medium High

$2.$ According to the data reported in the Excel, characterize the demand. San Pellegrino is particularly

interested in understanding main univariate and bivariate statistics $($especially if there is a

correlation between demand in specific weeks over the year$)$

$($Refers to the Excel graphs and calculation$)$

Univariate statistics:

How you consider the demand?

Stable Unstable Stable in some periods, unstable in others

How data are distributed?

Bell curve$/$normal distribution Other

Where data are skewed?

Left $($low values$)$ Right $($high values$)$ Not skewed

Bivariate statistics:

Is there correlation between demand in a week and demand in another week?