The CPFR concept assumes the cooperation of enterprises in the logistics chain in three areas of operation: planning, forecasting, and inventory replenishment. However, in the author’s opinion, the stage of collaborative sales forecasting is of the greatest importance for the idea of cooperation being undertaken. The level of maintained inventory and the associated level of customer service depend on the quality of the forecasts. Therefore, the effort of cooperation in this area undertaken by partners brings multilateral benefits to the involved parties.
The idea of collaborative forecasting within the CPFR strategy will be presented using the example of two cooperating enterprises: a manufacturer and a retail chain. This concept has already been implemented globally, but in Poland, no retail chain has yet taken up this challenge. For this reason, the presented results are based on research and simulations conducted by the author. Two simulation models were built for the research. The first, called the transactional system, corresponds to the currently prevailing type of cooperation. The relationships connecting both enterprises and the processes they implement are characterized by:
-
a high degree of independence in decision-making
-
an independent process of planning one’s own activities
-
a limited scope of mutual agreements
-
a lack of open information exchange
-
individual forecasting systems
-
inventory replenishment by the manufacturer based on orders placed by the retail chain.
The second simulation model, called the collaborative system, represents an extremely different approach to the established cooperation. It is based on:
-
joint decision-making
-
shared goals and action plans
-
a wide scope of mutual agreements
-
extensive information exchange
-
collaborative creation of sales and order forecasts
-
inventory replenishment by the manufacturer based on a joint “frozen forecast”.
Historical data from one of the retail chains operating in Poland were used for the simulation research. Based on the volume of recorded sales in previous periods, 10 variants of possible consumer demand patterns for the following year were randomly generated. The inverse transform sampling method was used to generate random numbers.
The conducted simulation research concerned three potential situations:
-
Situation I: cooperation between the manufacturer and the retail chain is carried out without disturbances – no calendar events (e.g., organized promotions), partners do not take any actions affecting the realized sales volume
-
Situation II: the manufacturer takes marketing actions aimed at stimulating the sales of the studied product
-
Situation III: the retail chain, in order to intensify sales in its branches, takes additional marketing actions that cause an increase in sales.
A product characterized by sales with an upward trend was subjected to the study. For this reason, the Holt method, used for forecasting this type of product, was chosen to determine forecasts for subsequent periods.
The forecasting cooperation of partners in the CPFR concept concerns two determined quantities: the sales forecast to consumers carried out by the retail chain and the order forecast submitted by the retailer to the manufacturer. For each of these forecasts, it is necessary to establish the framework of the jointly implemented process. For this reason, partners are obliged first to define the preliminary conditions that allow them to compare the created forecasts. Among the issues to be established, particular attention should be paid to three aspects.
1. The organizational level for which joint forecasts will be created. The determined level decides the degree of data aggregation used for forecasting and must be common to both partners so that the forecasts can be compared. This is a particularly important issue, as retailers generally forecast at a higher level of detail regarding product sales locations than manufacturers. Conversely, manufacturers forecast at a higher level of detail regarding products than retail chains do. Therefore, it is very important that the manufacturer and the retail chain choose one of the possible levels of joint forecasting, which may concern the chain’s distribution center, a region, or a store (branch).
2. Criteria for identifying exceptions for the created forecasts and handling them. In jointly created forecasts, there may be so-called “exceptions” concerning products that are not fully covered by joint forecasting or for which special procedures are provided. In their case, partners must specify the mode of action in detail so that each party knows what its responsibilities are. Exceptions may also be partner forecasts for which the recorded discrepancy is unacceptable. However, to be able to identify such an exception, it is first necessary to establish acceptable limit discrepancies between the forecasts created by the partners for subsequent periods. The set values will then determine the classification of a given forecast as an exception. Criteria for identifying exceptions are set for both sales forecasts and order forecasts.
3. Time. Within the joint forecasting system, partners must determine the unit of time for which the forecast will be created. This is necessary because significant discrepancies exist here in the case of a retail chain and a manufacturer. However, for forecasts to be comparable, they must refer to the same unit of time. In the author’s opinion, it is most beneficial in this case to create forecasts for weekly data. The level of aggregation is small enough to enable the effective use of the created forecasts for planning supply deliveries to the chain. At the same time, it does not require the daily transmission of detailed data between partners, which would be associated with high costs of the forecasting process. Such a determined level of aggregation is also confirmed in practice, as in most implemented joint forecasting projects, the adopted unit of time was indeed a week. The element of time is also associated with the need to determine the forecasting horizon.
The article comes from the “Logistyka” magazine 4/2009.