4. CASE COMPANY ANALYSIS
4.4. Summary of the demand forecasting process of the case company
Because the performance measurement has not received much of a focus, steps 6 and 7 (figure 3.3) have also been neglected. Thus far, there has not been any change of parameters of statistical models or even forecast models themselves. Additionally the feedback that the people involved are provided with is, in most cases, only the information provided by the software, which includes the aforementioned error calculations and the control card. Afterwards, it is up to the people involved to find the relative data that could be beneficial in their estimation of the quality of the forecasts and the whole forecasting process. (Case company material [5]) Because of the external approach of this study, it cannot be said with utmost certainty that there are no other sources of performance feedback. However, based on the fact that the whole demand forecasting process of the case company is centered on the forecasting software, it can be suggested that it is also the main conduit for feedback.
4.4.
Summary of the demand forecasting process of the
case company
This chapter summarizes the analysis of the previous subchapters and additionally uses the literature review of chapters 2 and 3 as a benchmark to ascertain the problems of current forecasting practices. It must be remembered though, that prior studies and research focus usually on forecast practices applied in consumer markets, and the case company is not only engaged in consumer markets but also in industrial markets, within which some variation exists. In addition, there are some other shortcomings of previous studies, which are discussed in this chapter.
The characteristics of the case company also have an effect on the forecasting practices, and the suitability of prior research to this particular situation. For example, the heterogeneous customer base also means that there are number of sources of information available and not just the historical demand data, which is usually the focus in the prior studies. This is true especially in the industrial markets, but in this case also in consumer markets. For example, the demand for outdoor paints can be depended on weather conditions.
It should also be noted that because of a vast product mix of the case company, it is not always possible to incorporate all of the demand forecasting procedures to all of the different products. Therefore, some of the forecasting practices presented in the literature review are not always possible because they do not always distinguish the difference between an ideal and a real-life situation. The difference between those two
is that in real-life resources are limited, and as it is in this case, there are number of different individual products in the mix which require at least partially different approaches. In addition, prior studies are often merely generalizations about the best practices, which means that they are not necessary applicable to all of the situations. Even though there are some shortcomings in the literature review, it is still applicable, when comparing the demand forecasting process of the case company (figure 4.1) to the one described in chapter 3.2 (figure 3.3).
Figure 4.1. The demand forecasting process of the case company.
When comparing the actual situation of the demand forecasting process of the case company to the ideal one presented in subchapter 3.3, some general differences can be noticed. The different statistical models that have been used have always been the same, as well as their parameter combinations. There combinations are used by the forecasting software to calculate the systematical forecast and are used for all of the product
segments. Even though there are twelve different models available, including specific models for certain kind of demand patterns, only two of the available models are used. These two models are used for all of the products regardless of their product segments or demand patterns.
Another problem of the case company is the incorporation of judgmental input to the forecasting process: the rules and previous guidelines relating to adjustments focused only on how they should be done in the forecasting software, not in which situations or for which products. However, this problem is also a shortcoming in some of the previous studies: even though they present different ways on how judgmental adjustment of forecasts can be done, they do not define properly, when it is needed and in when not. An additional problem related to judgmental input is that it is not always possible to do because, based on the guidelines of the case company, it should be done for individual or at least a small number of products, which obviously takes considerable amount of time because of the vast product mix.
The problem related to the performance measurement is the fact that even though the calculation of forecast errors is done automatically every period by the software, this information has not been properly used by the people involved in forecasting. The same goes for performance feedback. This problem was also mentioned in literature review, where it was stated that too often the errors are calculated but their impacts on performance are not measured in any way. One reason for this is the fact that the entire demand forecasting process of the company works around forecasting software, which means that the so-called feedback is in the raw data calculated by the software. This could also mean that, even though the implementation of the forecasting software helps in most of the forecasting procedures, it can also have a negative effect because it handles everything automatically, which can lead to the fact that the users of the software rely too much on it.