Understanding the characteristics affecting the classification over time within combined MTO-MTS production systems.

(1)

Understanding the characteristics affecting the classification over

time within combined MTO-MTS production systems.

Master Thesis

M.Sc. Technology and Operations Management

University of Groningen

Faculty of Economics and Business

Mathijs Peter Zandberg

m.p.zandberg@student.rug.nl

S3834689

First assessor: Dr. O.A. Kilic

Co-assessor: Prof. Dr. D.P. van Donk

Date: 22-06-2020

(2)

1

Abstract

A big challenge in combined MTO-MTS production is the classification of MTO and MTS products. Previous research shows that industry-specific characteristics of production systems have a strong influence on MTO-MTS classification and makes it more complex. However, the literature mainly focuses on classification at a certain moment in time and do not address the need to revise and the factors that urge a revision of classification. This study looks at how food processing companies decide on a classification over time within combined production systems. Due to the explorative nature of this research, a multiple case study was conducted on food processing companies exercising combined MTO-MTS production. Our study demonstrates that companies use periodic reclassification to deal with uncertainty and risk or to increase competitive advantage along a product life cycle. Classification over time appears to be much more of a continuous process that balances industry-specific characteristics and customer wishes, rather than it being a straightforward calculation. This study provides insights into the influencing characteristics of classification over time and provides a generic approach that aids companies to decide on how often they should revise a product classification along the product life cycle.

(3)

2

Acknowledgements

This research paper marks the completion of the M.Sc. in Technology and Operations Management at the University of Groningen. The completion of this paper would not have been possible without the help of others. Therefore, I would like to give a sincere thank-you to my supervisor’s prof.dr. D.P. Van Donk and dr. O.A. Kilic. Their comprehensive feedback, suggestions and guidance during this project were vital for its completion. Thank you both for the valuable and stimulation feedback on my work.

Additionally, I would like to thank my fellow students Jacob Berger and Rick Cramer for the interesting discussions regarding our subjects which kept the research process motivating.

Next, a special thanks goes to the participating companies for their openness and the interesting data they provided for this study. Their insights on challenges within the food processing industry have been indispensable for this research and have motivated me, even more, to pursue a future career in the food processing industry.

The completion of this study marks the end of my time at university. A period in my life where I have grown both personally and professionally. I am very thankful for this opportunity for which I want to thank my mother and father. Their everlasting encouragement and operational support to pursue my wide interests have been a true privilege.

(4)

3

List of Figures

Figure 2.1: Hierarchical approach to MTO–MTS problem (Soman et al., 2004). ... 7

Figure 2.2: Five characteristics that affect the length of the classification interval . ... 289

Figure 2.3: The research framework ... 2910

Figure 5.1: Five characteristics that affect the length of the classification interval ... 27

Figure 5.2: New affecting characteristics on the product classification over time. ... 288

Figure 5.3: Generic structure to determine a classification interval ... 299

List of Tables

Table 2.1: Food processing characteristics. ... 10

Table 3.1: Sample characteristics of the cases ... 14

Table 3.2: Interview details ... 15

Table 3.3: Example coding tree. ... 16

Table 4.1: Overview of FPI characteristics... 19

Table 4.2: Overview of findings on classification over time ... 22

(6)

5

1. Introduction

A challenge in combined make to order (MTO) and make to stock (MTS) production is to determine which products to MTO and which products to MTS (Soman et al., 2004). This strategic decision is defined as the MTO-MTS classification and is influenced by product-, market- and process-oriented characteristics (Aitken et al., 2003; Van Kampen et al., 2012). Because some of these characteristics change over time (Hemmati & Rabbani, 2010; Tanco et al., 2013; Van Donk, 2001) one would expect that the classification changes accordingly. For instance, when the demand is seasonal, companies may use an MTO strategy in periods of low demand, and switch to an MTS strategy in periods with high demand (Aitken et al., 2003). This gives them the ability to adapt to demand peaks. However, adapting a classification solely based on demand might not be advisable, as opportunities are missed to either reduce risks or to increase competitiveness (Van Kampen & Van Donk, 2014b). In the presence of these opportunities to increase the value or to reduce risks, the question remains when to adapt a classification and how to determine an appropriate period for reclassification.

This paper investigates the aforementioned question in the context of the food processing industry, as industry-specific characteristics make combined MTO-MTS production systems quite common (Dora et al., 2016; Soman et al., 2004). The classification within the food processing industry is surrounded by conflicting characteristics, as the demand of products is volatile, the products have limited shelf life and production lead times are long (Van Kampen & Van Donk, 2014b). A poor classification could, therefore, result in loss of sales, penalties or the decay of products, hence it is important to properly determine and adapt the classification for each product (Van Kampen et al., 2012). As a wrong classification has big consequences for food processing companies, managers need an understanding of the important food processing industry (FPI) characteristics in classification and reclassification to provide a structural approach for classification over time.

(7)

6 been acknowledged in other research areas as inventory and spare part management (Aitken et al., 2003; Bacchetti & Saccani, 2012; Nallusamy et al., 2017; Syntetos et al., 2009). Nallusamy et al. (2017) and Aitken et al. (2013) show that periodically reviewing the ABC-inventory classification could increase control of the raw materials more efficiently. Baccetti & Saccani (2017) and Syntetos et al. (2009) even suggest that reclassification should be organisationally embedded. Van Kampen & Van Donk (2014b) also acknowledged this gap within the classification literature and executed a single case study to investigate how a classification should be adapted over time. Their results suggest that the decision to revise a classification should balance between competitive advantage and reduced risk on stock obsolete on the one hand and the effort to adapt and communicate the reclassification on the other. While this paper provides insights into reclassification trade-off it lacks an understanding of how companies deal with this trade-off when they adapt their classification.

The aim of this paper is, therefore, to understand which factors influence the MTO-MTS classification over time within food processing companies. While current literature mainly focuses on the classification at one single moment in time, this paper will provide new insights about when, how and how often organizations revise their classification. Given that there is only little known about classification over time, an explorative research design was chosen. A multiple case study is conducted to gain a deeper understanding of the decision-making process of food processing companies in the context of classification over time in combined MTO-MTS production. Therefore, we aim to contribute and expand the current knowledge base on the dynamics of the classification over time.

(8)

7

2. Literature background

In this section literature on classification over time in combined production systems will be discussed, starting with the general research directions, followed by the classification over time. Next, the characteristics of the food processing industry are discussed to highlight their potential influence on the classification over time.

2.1 Classification within a combined MTO-MTS production system

More and more production systems are following a combined MTO and MTS production mode (Beemsterboer et al., 2016; Hemmati & Rabbani, 2010; Köber & Heinecke, 2012). Therefore, an increase in academic attention has been dedicated to combined MTO-MTS production systems (Hemmati & Rabbani, 2010; Perona et al., 2009), and researchhas shown the potential benefits of combined MTO-MTS production systems (Beemsterboer et al., 2016; Zhang et al., 2013). One of the more qualitative studies on combined MTO-MTS was conducted by Soman et al. (2004). They provided a framework with the hierarchy of decisions involved in combined MTO-MTS production situations, which is presented in Figure 2.1.

Figure 2.1 Hierarchical approach to MTO–MTS problem (Soman et al., 2004).

(9)

8 (Hemmati & Rabbani, 2010). The classification is complicated due to the conflicting market, product and process characteristics that need to be taken into account (Soman et al., 2004). For example, long lead times (process) and short delivery times (market) might lead a product to be classified as MTS, while dealing with high product perishability (product) an MTO classification is preferred. Therefore, in some cases, finding and maintaining a suitable classification for each product is not straightforward, but is surrounded by trade-offs and compromises between different departments (Romsdal et al., 2013; Soman et al., 2004; Van Kampen et al., 2012).

A company has to decide which product, process and market characteristics are most relevant to use in their classification (Kampen et al., 2012).Olhager (2003) provides a list of general characteristics that could be of influence on the classification and Soman, Van Donk and Gaalman (2004) argue that a set of particular food processing characteristics could also be of influence, which make the classification even more complex. Additionally, some studies, propose a model or tool that, given the characteristics, determines the classification (Aktan & Akyuz, 2017; Hemmati & Rabbani, 2010; Van Donk, 2001; Van Kampen et al., 2012). However, in recent years limited papers have been devoted to how companies decide which characteristics to consider in their classification. To have a better understanding of how companies make a classification within various settings more case studies are needed to explore the influencing factors in the decision of a classification.

2.2 Classification over time

(10)

9 classification on the other hand (Van Kampen & Van Donk, 2014b). However, it is unclear whether these are the only two reasons to revise a classification and how this trade-off is affected by the different market, product and process characteristics of a company.

Since the classification is one of the main strategic decisions in combined MTO-MTS production systems, deciding on how often companies revise their classification is also of great importance. Van Kampen & Van Donk, (2014b) identified that at least five characteristics affect the length of the classification interval, as Figure 2.2 illustrates. A short classification interval requires a lot of workload in communication and negotiation the changed ordering possibilities as the classification is often adapted (Van Kampen & Van Donk, 2014b). Therefore, both companies and customers prefer that the classification is not revised too often (Bacchetti & Saccani, 2012; Van Kampen & Van Donk, 2014b). However, a short interval does give the company the ability to adapt better to changing characteristics, which increases the competitive advantage of a company. While this paper identified five main characteristics which influence the reclassification interval, they lack a structure in how companies select an appropriate reclassification interval.

Figure 2.2; five characteristics that affect the length of the classification interval (Van Kampen & Van Donk, 2014)

Based on these characteristics Van Kampen et al. (2014b) argues that the classification over time is less of an isolated process, as is often described in the literature. For example, the customer wishes can, to some extent, negotiate the classification and reclassification (Van Kampen & Van Donk, 2014). For instance, if a specific customer suddenly demands very short lead times, reclassification from MTO to MTS might be needed. While this increases the risks of absolute of stock, it could be more important to avoid the risk of a cost-penalty of not delivering on time. Hence, further studies are needed to confirm the dynamics of the classification process to see how companies approach classification over time.

high demand predictability (incl. seasonality) low

low Product perishability high

high effort to adapt a classifcation low

low increased competivieness or reduced risks high

Long Product life cycle short

(11)

10 2.3 The food processing industry characteristics

The food processing companies must deal with a set of specific characteristics when making the classification. To deal with these characteristics production systems in food processing companies are often characterized as a two-stage process: 1) raw material processing and 2) the packaging of a final product (Van Kampen & Van Donk, 2014a). Food processing companies are therefore capable of product-customization while facing downstream pressure to be flexible in delivery lead time. Van Donk, (2001) proposed a list with plant, product and production process characteristics that characterize the food processing industry.

Later, some other papers made a distinction between more general market, supply, product and process-related characteristics, that can influence the classification (Hemmati & Rabbani, 2010; Olhager, 2003; Romsdal et al., 2011). As a starting point for this research, we used an integrated version of the FPI characteristics by Van Donk, Hemmati & Rabbani and the influencing factors identified by Van Kampen & Van Donk, (2014b). This integrated version is given in Table 2.3, we have categorized them into the product, process, and market-related characteristics.

Table 2.1: Food processing characteristics (Hemmati & Rabbani, 2010; Van Donk, 2001 and Van Kampen & Van Donk, 2014b).

Product Market Process

Product perishability Demand predictability Production lead time (Van Donk, 2001) (Hemmati & Rabbani, 2010) (Van Donk, 2001) Product variability Product customization Process flexibility

(Van Donk, 2001) (Van Donk, 2001) (Hemmati & Rabbani, 2010) Product Life Cycle Required delivery reliability Holding costs (Van Kampen & Van Donk, 2014b) (Hemmati & Rabbani, 2010) (Hemmati & Rabbani, 2010)

Raw material variability Required delivery time Production volumes (Van Donk, 2001) (Hemmati & Rabbani, 2010) (Van Donk, 2001)

Payment reliability

(Van Kampen & Van Donk,

2014b)

(12)

11 2.4 Research framework

Based on the insights from the previous sections a research framework has been developed and is presented in Figure 4. The framework indicates that the classification over time has been examined, in a combined MTO-MTS production system and it’s influences of food processing characteristics. Based on the previous section the classification over time is divided into three questions that are investigated: 1) the classification decision making which focuses on when companies change a classification, 2) the reclassification objectives, which relates to how companies change their classification and 3) the reclassification interval, which relates to how often companies change their classification. Based on the literature background it is expected that companies with a particular set of characteristics such as a short product life cycle and high product perishability, as described in Figure 3, have a shorter reclassification interval and a more dynamic approach to the classification over time.

(13)

12 3

Methodology

Given the exploratory aim of this study, we opted for a multiple case study, based upon interviews as we are interested in the what and how of the observed phenomena and have only a limited conceptualisation of the relationship between the phenomena of interest. In this section, we explain the research method, case selection, and data collection and analysis practices of this study.

3.1 Research method

This study aims to investigate the affecting factors of classification over time within combined production systems. An explorative approach is appropriate as there is little known on the process of revising the classification. Therefore, an in-depth analysis of the specific contextual factors of the food processing industry is required to understand how and when companies decide to revise their classification. Given the exploratory aim, we studied multiple cases of a combined production system, to compare the cases to determine similarities and differences between the companies. Additionally, it allows studying the phenomena and the context of the phenomena to deduce both cause and effect (Leonard-Barton, 1990). This provides us with a deeper understanding of the processes and the chance to understand how companies make certain decisions related to the classification over time. To gain an understanding of the classification over time, the field-based approach of these cases is vital since it offers a chance to understand how it functions in its natural settings (Dehoratius & Rabinovich, 2011). Mainly primary data is used to address and resolve issues specific to this research topic and to guarantee accurate and up-to-date information.

3.2 Case selection

(14)

13 Literal replication has been applied by only selecting food processing companies, with a combined production system. Theoretical replication, however, has been applied by selecting cases with different sizes, sectors products and having a different capacity distribution between MTO and MTS products (Karlsson, 2016, Flick et al., 2004).

Based on the local business-to-business directory and the network of the researchers, a list of food processing companies was compiled. Before adding them to the list, the companies were assessed on the mix of criteria mentioned above. When a company had more than 15 employees, it was considered sufficiently professional to be included in this study. We assessed the type and size of the company, by studying company website, LinkedIn page and product assortment. The initial contact with the companies was made by telephone to determine whether the company was willing to participate in the research. An additional purpose of these initial telephone contacts was to explain the outline of the research project and to ask whether the company engaged in both MTO and MTS production. After the initial contact, additional e-mails were sent to further elaborate on what was expected from the company and researchers.

3.3 Case description

The structure provided us with an opportunity to study four different cases of combined production systems within the food processing industry. Table 3.3 provides an overview of the cases with sample characteristics. Each case represents one site of a food processing company, over a wide variety of product sectors. We can verify that all interviewed companies are operating in the food processing industry and use a combined MTO and MTS production, which was a criterion for this study.

(15)

14 this study that were outlined in chapter 3.1. Meaning this study is well placed to investigate the classification over time in different kind of food processing companies.

Table 3.1: Sample characteristics of the cases

Company

#Years of

Existence %MTS %MTO #Employees #Products

Competitive advantage

A 30 40 60 300 <100 SKU

Product quality and know how

B 405 33 66 400 +200 SKU

Product quality and taste

C 35 85 15 140 +900 SKU Product variety

D 137 90 10 250 <100SKU Product quality

3.4 Data collection

Semi-structured interviews are the main tool for data collection. Therefore, the effectiveness of the multiple case research is primarily dependent on the quality of interviews (Karlsson, 2016). To increase the richness of data we allowed some freedom through open-ended questions while still following a certain structure (Flick et al., 2004). The interview follows an interview protocol that is based on the hierarchical planning model of Soman et al. (2004), as it covers the broad context decision making in combined MTO-MTS production. As this study followed the recommendation of Van Kampen and Van Donk (2014b) in assessing the classification over time specific questions on classification over time and the reclassification interval were added based on their interview protocol (Van Kampen & Van Donk, 2014b). The protocol was divided into four main sections: 1) general company information, 2) production planning, 3) classification and 4) reclassification. The third section aimed to define the classification decision-making process and when companies change the product classifications. The reclassification sections aimed to determine how and how often the company revises their classification and why. The interview protocol was used for all the cases, to improve the analysability and generalizability of the findings. The interview protocol is presented in Appendix A.

(16)

15 the interviewee was asked if he/she was the right person to answer questions on this topic. Table 3.2 provides an overview of how the information was collected for each case.

Table 3.2: Interview details

Cases #interviewees Function

Interview

length Data gathered

A 1

Continuous improvement

manager 130 minutes Skype interview

B 1

Tactical planner & Reporting

specialist 130 minutes Skype interview

C 1

Head Supply Chain

Management 110 minutes Skype interview

D 2

Production planner and head

business office 55 minutes Skype interview

3.5 Data analysis

(17)

Table 3.3: Example coding tree. A: represents case company A, B: represents case company B, C: represents case company C and D: represents case company D.

(18)

Using the codes presented in Table 3.5, we were able to effectively compare how companies approach product classification over time based on companies FPI characteristics. Following the recommendations of Van Kampen and Van Donk (2014b), a cross-case analysis was carried out with these four cases to understand how different production settings influence the classification over time. Based on the cross-case analysis we compared how different the FPI characteristics of the companies influence the classification over time. We summarized the main findings on classification over time per each case in excel by using quotes from the coding tree. The detailed first-order coding for each case is presented in Appendix C.

To enable cross cases analyses we ranked the FPI characteristics for each case. Scores rank from Very Low (1) to Very High (5) and are based on the coded data. Looking at for instance product perishability, company C mentioned to have no perishable products: “C: for the

majority, our products consists out of sugar, glucose and gelatine which have a long shelf live so we are quite safe, and the product will be good for 1 to 2 years”. Whereas company A

mentioned having high product perishability “ Cool-fresh products have a relatively short shelf

life, you can think of 10/20 days of which ⅓ can be spent at our facility”. Which indicates the

differences between the two levels of perishability, which are given a score Very Low (1) in company C in comparison to Very High (5) in company A. Another example is product variety, which was measured by the number of SKU’s: a low product variety means 0-99 SKU’s, a low/medium product variety means 100-199 SKU’S, a medium product variety means 200-299 SKU’s, a medium/high product variety means 300 -399 SKU’S and a high product variety means more than 400 SKU’S.

While the scale of perishability and product variety are very objective as it is expressed in a number, the scales for other characteristics such as process flexibility, were more difficult to create objectively, due to difference in interpretations. By defining these clearly and objectively, we could make clear comparisons between cases. For example, low process flexibility (1) was assigned to company D, “There is one production line for one product, x.

This yields for both the mixing and packaging department”, as production lines are specific to

one product and high process flexibility (5) was assigned to company A, “A product can be

produced on multiple production lines”. To ensure the objectivity of these scales the scales are

developed in cooperation with two other researchers who performed research on the same case companies. Some additional examples and the meaning of the scales are given in Appendix B. Lastly, we included the percentage of MTO and MTS products of each case. The %MTO/MTS relate to the percentage of total capacity being allocated to each of the production strategies.

(19)

18

4 Findings

The following findings on the classification over time have been found based on the four processing cases which participated in this study. First, an overview of the different FPI characteristics per case is given in Section 4.1. Then the most important decisions companies make regarding the classification over time are presented in Section 4.2. Based on these findings a cross-case analysis is done in Section 4.3 of this study to determine the important characteristics of influence on the classification over time.

4.1 Overview of characteristics

Company A (40%MTS and 60% MTO): Company A is a clear example of a two-stage

production system with a semi-finished product being produced to stock, whereas the packaging capacity was distributed more evenly between MTO/MTS: “ roughly 90% of all the

articles we have in intermediate stock. From there 40% is MTS and 60% is MTO, so the customer order decoupling point is at the storage of semi-finished products”. The products are

highly perishable and as the market of company A is relatively new, it is hard to predict demand. Due to the stock of semi-finished products, company A has an extremely short production to delivery lead-time ratio (P/D ratio) for both MTO and MTS products. “if I get an

order today I can package the product and deliver it tomorrow, so the delivery lead time fits within the order-lead time”. The p/d ratio can be described as the relation between production

and delivery lead times. Additionally, the life cycles of products are short as the recipes are improved and changed regularly.

Company B (66% MTS and 33% MTO): The products of company B have low product

perishability and medium product variety. The demand is highly predictable and stable due to long customer relationships and low customer turnover. Company B has very limited customization options due to large production volumes and their production lead times are long as some products need to mature. Company B aims for stocking products for a maximum of 8 weeks as the warehouse capacity is very limited. Lastly, we see that the payment reliability of customers is very high as company B is not financially liable for stock related to MTS products. “All responsibilities that we take by stocking products in name of the customers are basically

for the customer”.

Company C (85% MTS and 15% MTO): The products of company C are not perishable as

(20)

19 customizable as company C gets its competitive advantage from their wide variety of products. Some of the products of company C are subject to high seasonal demands: “ for example during

Halloween we see that the demand of certain products increases significantly”. Company C

can predict demand well as they invested in a forecasting tool which supports them in predicting demand trends. Lastly, due to the large variety of products and customers, company C has some customers who are unreliable in their payments.

Company D (90% MTS and 10%MTO): Company D has only two product types which are

highly perishability as they have a shelf life of 5 weeks. The demand of the products of company D is highly predictable “Orders of the past years are a representative image of what

will be ordered again” and the demands of some products are subject to seasonality “We see that during the spring break, certain product sell very good due to all the events and activities that happen in the spring.” Production volumes are high and production lead time is long as

products need to be baked and cooled before they can be packaged. Lastly the production processes are not flexible as lines are specific to one product type only.

Based on these findings we provide an overview of the FPI characteristics introduced in Table 4.1 and their scores in relation to other companies in the sample. As mentioned in Section 3.5, scores are allocated through discussion with two fellow researchers. Additional examples of how the scores are given can be found in Appendix 2. Table 4.1 will be used for the cross-case analysis in Section 4.3 to determine their influence on the classification over time.

Table 4.1: Overview of FPI characteristics

FPI characteristics Company A Company B Company C Company D

%MTO/MTS 60%/40% 33%/66% 15%/85% 10%/90%

Product

Product perishability 5 3 1 4

Product variability 1 3 5 1

Product Life Cycle 1 4 2 5

Raw material variability 1 4 1 4

Market

Demand predictability 1 5 3 4

Product customization 4 2 5 3

Required delivery reliability 5 5 5 5

Required delivery time 4 5 4 3

Payment reliability 3 5 1 4

Process

Production lead time 3 5 3 4

Process flexibility 5 2 3 2

Holding costs 4 3 5 3

(21)

20 4.2 Results on classification over time

This section provides the findings on when, how and how often each company decides to revise or change product classifications.

Company A: The classification of products is an ongoing process within company A which

needs to be reviewed frequently. The product classifications are reviewed and discussed with a supply chain steering committee every month and reclassifications are made without the involvement of the customers. Company A decides on a reclassification interval based on the life cycle stage of a product. In the introduction, growth and declining stage of a product company A review the classification more than once per week, “with a new article I will review

the classification more frequently than once a week, because we have no data to base the classification on”, while in the maturity stage they review it less frequently.

To determine a product classification, company A continuously reviews the demand predictability and the order frequency of a product in relation to the product perishability. Products with a stable and high order frequency are classified as MTS and products with an unpredictable and low order frequency are classified as MTO. However, when the level of either one of these characteristics changes over time, company A considers adjusting a product classification. The product classification is changed from MTO to MTS to increase competitive advantage by gaining the ability to group orders and reduce changeovers. “ So that I can choose

when I want to produce and how much I want to produce of an article. So not today a bit and tomorrow a bit, but to group all tree, this saves two changeovers”. Subsequently, a product

classification is changed from MTS to MTO to reduce the risk of stock obsolescence, for example, due to unpredictable demand. Additionally, company A frequently reviews the balance between MTO and MTS products to determine whether a reclassification is needed “If

I have too much MTO articles versus MTS articles. So that I cannot play with my production capacity as easily, it becomes very random for example, based on that I am willing to change a classification”. Company A needs a healthy share of MTS products to be able to efficiently

utilize the production capacity during demand fluctuations of products.

Company B: Company B uses more of an ad-hoc approach to the classification over time and

reviews the product classification less than once per year. A classification is adapted when either the customer or the company feels the need to change a product classification. “These

(22)

21

therefore there are less reasons to review this decision very often.”. Company B decides on a

product classification together with a customer: “We basically decide together with the

customer whether the product is MTO or MTS because most implications of this decision are for the customer”. Products are classified in a quite straightforward manner, smaller customers

who order less frequently are MTO and bigger customers with a more frequent and stable demand are MTS.

Company B changes a product classification from MTO to MTS to increase delivery performance for a customer: “often the new customers, which are often new markets for us,

start with an MTO strategy because they begin small. As they become bigger, there is an increasing pressure to reduce the lead time and we will have a conversation to make it an MTS product”. Additionally, a product classification is changed from MTS to MTO to optimize

warehouse utilization, as company A has limited warehouse space. “for a customer we

reclassified from MTS to MTO, because we kept stalking them to pick up their inventory as we produced a stock of 6 weeks.”. The order reliability of a customer is therefore important and a

key determinant to change a product classification for company B.

Company C: Company C describes the product classification as an ongoing process which

needs to be reviewed based on the life cycle of the product. Company C reviews the products classification weekly by checking the forecast with the actual demands to determine whether the current product classifications are optimal. The frequency of reviewing the classification is based on the life cycle phase of a product: “We track the phase of all our articles, especially

at the starting phase of products and the stopping phase of a product, then we will review more frequently whether our classification should be changed”. Within this decision they also

consider the customer agreement and the customer relationship: “With more uncertain or less

mature customers, we review more frequently what the influence of our production strategy is”.

(23)

22 to reduce stock obsolete or to deal with the payment reliability of a customer “yes, we keep

track of the customers who pay unreliably and then we will not produce the product to stock until the customer has paid his bills”. Additionally, company C changes a classification to deal

with seasonal demand of products. If seasonal patterns are predictable, company C is willing to change the product classification to reduce the risk of shortage in delivery during extreme demand peaks. To demonstrate this, company C gave the following example regarding Halloween products, which are normally MTO products, but the month before Halloween they will produce it to stock: “while it remains an MTO article, planning decides to produce the

product to stock”.

Case D: Company D describes the product classification over time as an ad-hoc decision and

is reviewed twice a year. The classification moments are highly formalized and only occur when there is a change in the product portfolio of the customers. A classification is done in a straightforward manner, products with a high order frequency and high volumes are MTS and products with a low frequency and low volume are MTO. Company D only changes a product classification when the order frequency or volume changes significantly for a product, due to for example a customer turnover: “ If for example a product that we export to Germany which

also has been sold in The Netherlands but has been removed from the portfolio of the Dutch customers…. then we decide to reclassify from MTS to MTO, because the German customer orders only three months per year”. However, this does not happen frequently as customer

relationships are long and there is low level of customer or product turnover.

Lastly, an overview of the findings on classification over time per each case is provided in Table 4.2. These findings will be used for the cross-case analysis in Section 4.3 to determine whether the differences between companies on classification over time can be explained by their food processing characteristics.

Table 4.2: Overview of findings on classification over time

Company A Company B Company C Company D

Reclassification Determinant

Risk of Stock Obsolescence MTO-MTS balance Product Life Cycle

Customer agreements Order reliability Demand predictability Demand seasonality Payment reliability Product Life Cycle

Customer agreement

Interval Monthly Yearly Weekly 6 months

(24)

23 4.3 Cross case analysis between cases

In this section, a cross-case analysis is executed based on the findings of the previous two paragraphs. Table 4.3 presents an overview that is used for the cross-case analysis, which reveals the different market, process and product characteristics between the cases and the finding on classification over time. This section aims to determine whether the differences practices in classification over time can be explained by the FPI characteristics of a company. As the influence of some of these FPI characteristics is strongly intertwined, we group certain characteristics to efficiently compare their effect on the classification over time between the different cases.

Table 4.3: Overview of results on classification over time and important FPI characteristics.

Company A Company B Company C Company D

FP I c hara ct eri st

ics High perishability

Short product LC Low product variety Low perishability Low perishability High product variety/customization Med-high perishability Long product LC Low demand predictability High financial liability

High demand predictability Low financial liability High demand predictability Seasonal demand High demand predictability Seasonal demand Short lead time Long lead times Long lead time Long lead time

Reclas. Det.

Stock obsolescence MTO-MTS balance Product Life Cycle

Customer agreements Order reliability Demand predictability Demand seasonality Payment reliability Product Life Cycle

Customer agreement

Interval Monthly (LC) Yearly (LC) Weekly (LC) 6 months

DM Continuous Ad-hoc Ongoing Ad-hoc

Product characteristics

(25)

24 explains why company D sees no added value in reviewing the product classifications frequently while company A does.

Secondly, the different levels of product variety and product customization between the companies influence how they approach classification over time. It appears that companies with a high product variety or customization options, more frequently review the classification than companies with low product variety. Company C has a high product variety and customization levels which increases the number of design switches and the number of customer relationships. This increases the risk of excess stock that is specific to one product, to reduce this stock they frequently review the product classifications. Company B also has medium-high product variety, but only review the product classification on an ad-hoc basis. This can be explained by the that the risk of product-specific stock, that is less problematic for company B, as they are financially not responsible for stock related to MTS products, but the customers are, which reduces this risk. This explains why company B does not frequently review the product classifications and company C does.

Another product characteristic that appears to be of influence is the difference in product life cycles between products of the companies. The products of company A and C have relatively short life cycles due to for example updates in recipes or design switches. Therefore, they continuously review the product classifications during introduction, growth or decline stages of products to reduce the risk of stock obsolete or increase competitive advantage. On the contrary, the products of company B and D have a relatively long-life cycle as their product assortment does not change often, due to stable and mature customer relationships. The added value for reviewing the product classification frequently is therefore low for companies B and D, as their products are not subject to change.

Market characteristics

(26)

25 to be able to do a reclassification to increase competitive advantage. In this case, demand predictability can also enable the opportunities of a short reclassification interval. However, if demand is highly unpredictable, as in company A, the findings show clearly that companies frequently review a classification, as there is a high risk involved that products go obsolete.

Additionally, the results show that the product classification over time is influenced by seasonal demands. Both the products of companies C and D are subject to seasonal demand, but in both cases its influence on the classification over time is different. Company C changes its product classification from MTO to MTS to deal with these demand peaks, while company D do not. By comparing the other characteristics of both companies, this can be explained by the difference in product perishability. The products of company D have a limited shelf life of 5 weeks, while the product of company C is not perishable. Company C is, therefore, able to stock products for a long period, as products can be stocked for longer than 1 year. While for company D, there is a high risk that products will expire before an order is received, they, therefore, must wait for an order before they will produce a product. This explains that there is an added value for reviewing the classification frequently for company C while there is not for company D.

Process characteristics

(27)

26

5. Discussion

This paper aimed to investigate how the characteristics of the food processing industry affect the MTO-MTS classification over time. The results show that general factors as well as specific food processing characteristics matter for how companies approach classification over time. In this chapter, the results on classification over time will be discussed with the literature.

5.1 The classification over time.

In investigating the product classifications over time, it has been shown that companies approach the product classification over time differently. The differences in market, product and process characteristics call for different approaches toward product classifications over time. In contrast to what most classification literature states, it has been shown that a product classification is not a static decision but is more dynamic. Companies that deal with a high level of uncertainty, risk or changes frequently review and change a product classification to either deal with the risk of obsolescence or to increase competitive advantage. Companies with a more stable, predictable environment make the classification once and only review or change it on an ad-hoc basis when either the company or the customer feels the need for reclassification. This is in line with the findings of Van Kampen & Van Donk, (2014b), which argue that the classification of a product is less of an isolated process, as is often reported in the literature over time, but is to some extend negotiable.

The results suggest that companies change a product classification from MTO to MTS to widen ordering possibilities for a customer and to increase the competitive advantage of a company. A product reclassification from MTS to MTO is done to reduce the risk of stock obsolescence. On the other hand, we see that a reclassification requires effort to communicate new ordering settings to the customer. Therefore, a product reclassification is a trade-off between reducing risks or increasing competitive advantage and the effort it takes to change a product classification. This finding is in line with and a further specification of Van Kampen & Van Donk, (2014b) who also observed this trade-off. This paper shows that specific food industry related characteristics influence this trade-off significantly.

5.2 The affecting factors on the classification over time

(28)

27 five characteristics they proposed will be discussed. The five characteristics and their influence are illustrated in Figure 5.1.

Figure 5.1; five characteristics that affect the length of the classification interval (Van Kampen & Van Donk, 2014)

The first characteristic that is proposed by Van Kampen & Van Donk, (2014b) is the level of demand predictability. The cross-case analysis in this study confirms that demand predictability influences the classification over time. However, it shows contradicting findings on how it influences the reclassification interval, as high demand predictability also enables the benefits of a short reclassification interval, as is done by Company C. Additionally, the results show that the level of demand predictability within a company is strongly influenced by the length of the customer relationship, as it determines the quality of historical data that is available. Therefore, the customer turnover rate should be identified as a critical influence on a classification over time.

Secondly, Van Kampen & Van Donk, (2014b) argue that the effort it takes to change a classification is important in how frequent a company reviews its product classifications. An important finding of this study is that the workload of reclassification seems to be related to the p/d ratio of a company. Company A has such a short p/d ratio due to the delayed customisation that they are capable of changing product classifications without having to change the ordering agreements. Therefore, a short p/d ratio, due to postponed customization decreases the workload of a classification and makes the organisation able to adapt planning procedures quickly. These planning and organisational changes are also suggested by Van Kampen & Van Donk, (2014a) and Trentin, (2011). Therefore, the p/d ratio should be identified as a critical influence on a classification over time.

The third characteristic presented by Van Kampen & Van Donk, (2014b) is the level of risk of obsolescence or competitive advantage. The cross-case analysis of this study confirms that companies with a high risk of obsolescence more frequently review or change product

high demand predictability (incl. seasonality) low

low Product perishability high

high effort to adapt a classifcation low

low increased competivieness or reduced risks high

Long Product life cycle short

(29)

28 classifications. Subsequently, this paper shows that the level of competitive advantage is influenced by the balance between MTO and MTS products. This paper suggests that companies frequently review or change product classifications to maintain a healthy balance between MTO and MTS products to increase capacity utilization. This finding is to some extent in line with the results of Beemsterboer et al. (2016), which argue that a hybrid policy benefits most when it has a healthy balance between MTO and MTS products.

Additionally, Van Kampen & Van Donk, (2014b) argue that product perishability influence product classification over time. This study confirms this finding, as companies with high product perishability are subject to high risk and need to be able to adapt quickly. However, we also see that solely product perishability is not a reason to review a classification frequently, as it is not a risk if demand fluctuations are known in advance. We, therefore, argue that it should be considered together with the predictability of demand as is also stated by Van Kampen & Van Donk, (2014b).

The last characteristic that is proposed by Van Kampen & Van Donk, (2014b) is the influence of the product life cycle on a product classification over time. Our cross-case analysis confirms that the product life cycle strongly influences the product classification over time. One of the findings of this study is that the length of a product life cycle is influenced by the level of product variety and product customization. This increases the number of customers relationships and design switches which require a company to more frequently review and change product classifications. The product variety/customization should be considered as a critical factor in determining a classification over time.

Based on the results of this paper we propose that at least three additional characteristics should be added to the five characteristics proposed by Van Kampen & Van Donk, (2014b). Based on the findings we propose that: the customer turnover rate, p/d ratio and the product variety/customization influence the reclassification interval as is illustrated in Figure 2. Further study is needed to confirm these characteristics and to determine how these are influenced in various production settings.

Figure 5.2: New affecting characteristics on the product classification over time.

High p/d ratio Low

Low Customer turnover rate High

Low Product variety/customization High

(30)

29 5.3. How companies decide on a reclassification interval

While literature confirms the need for adaptions of product classifications, it did not provide a structure for selecting an appropriate reclassification interval (Van Kampen & Van Donk, 2014b). Our findings show that companies adapt their reclassification interval on the product life cycle stages. These findings are to some extent in line with the results of Aitken, Childerhouse and Towill (2003), which argues that classification should be carefully matched throughout the product life cycle. To structure these findings, Figure 5.3 provides a generic structure that captures the frequency of which classifications are reviewed along the product life cycle.

Figure 5.3: Generic structure to determine an approach for classification over time

(31)

30

6. Conclusions

The aim of this paper was to gain an understanding of how companies deal with classification over time within a combined MTO-MTS production system in the food production industry. Despite limited attention the literature pays to the dynamics of classification, our paper shows that companies do frequently review and change their product classifications if the underlying characteristics change over time. Based on the results two main approaches for classification over time can be determined. When companies are subject to a high degree of uncertainty, change or risks they review and change product classifications frequently, while companies who operate in a mature and stable environment only change product classifications on an ad-hoc basis when either the company or the customer feels the need to change the classification. The reclassification is a trade-off between the implications of changing and communicating a reclassification on the one hand and competitive advantage and reduced risk of stock obsolete on the other. This study showed that many industry-specific characteristics influence this trade-off.

This study explores the dynamics of classification within combined production systems and adds as such to our understanding of which factors are of influence on a classification over time. Rather than focusing on classification as a single moment in time, we focus on the dynamics of the classification to offer explorative insights into how companies approach classification over time. This research provides empirical evidence that the dynamics of product classification is practically and academically relevant and cannot be neglected. As will be discussed in the next section of this paper, given the exploratory nature and limited sample, it does not provide final answers but should be used a steppingstone for academics to further explore classification over time.

(32)

31

7. Limitations and directions for future research

As with all studies, this paper has some limitations. Looking at the data collection of this study we acknowledge the findings are based solely based on one interview per case company with mostly only one interviewee represented. This could decrease the reliability of the data of the interviews, as the results could not be validated by other interviewees within the same company. However, as most of the interviewees held senior management positions which make this issue less of a concern. Additionally, due to COVID-19, face-to-face interviews and factory tours were not possible, as all interviews were needed to be done digitally, which could have provided extra information. We tried to mitigate this limitation by requesting further clarifications on certain topics during follow-up questions in e-mails.

A second limitation of this study could be the sample size, as the results are based upon only four cases within the food industry. This makes part of the findings specific to the company characteristics, as some of the influences of the industry-specific characteristics such for example product perishability, could not be compared with in other cases. Even though this study provides general insights in the classification over time further research is recommended to further explore and validate the influences of industry-specific characteristics on the classification over time.

Additionally, this paper focused mainly on the influence of industry-specific characteristics while company characteristics such as company age could also be of influence on the classification over time. For example, the younger companies within this paper (less than 30 years of existence), frequently review and adapt the classification while the older companies (more than 100 years of existence), only change a product classification on an ad-hoc basis. Therefore, it might be interesting to investigate how a company’s life cycle stage, size or company culture influences the classification over time.

(33)

32 In general, following the findings of Van Kampen & Van Donk, (2014b), it is remarkable that so little attention has been paid to the classification over time within combined production systems. Future research should aim to incorporate the dynamic of a product classification and further explore the suggested relationships form this study. As indicated, we suggest investigating classification over time along with a wider sample, which might also include other industries, to see if the same mechanism and influences will be detected and to look at additional factors. As mentioned above, such studies should preferably take a longitudinal design to be able to capture if and how product classifications change over time and how these changes are related to industry-specific characteristics.

(34)

33

References

Aitken, J., Childerhouse, P., & Towill, D. (2003). The impact of product life cycle on supply chain strategy. International Journal of Production Economics, 85(2), 127–140.

Akkerman, R., & Van Donk, D. P. (2007). Product prioritization in a two-stage food production system with intermediate storage. International Journal of Production Economics, 108(1– 2), 43–53.

Aktan, H. E., & Akyuz, G. (2017). Positioning the decoupling point along a supply chain: A case study. International Journal of Productivity and Quality Management, 22(3), 309– 339.

Bacchetti, A., & Saccani, N. (2012). Spare parts classification and demand forecasting for stock control: Investigating the gap between research and practice. Omega, 40(6), 722–737. Beemsterboer, B., Land, M., & Teunter, R. (2016). Hybrid MTO-MTS production planning:

An explorative study. European Journal of Operational Research, 248(2), 453–461.

Dehoratius, N., & Rabinovich, E. (2011). Field research in operations and supply chain management. Journal of Operations Management, 29(5), 371–375.

Dora, M., Kumar, M., & Gellynck, X. (2016). Determinants and barriers to lean implementation in food-processing SMEs - A multiple case analysis. Production Planning

and Control, 27(1), 1–23.

Flick, U., von Kardorff, E., & Steinke, I. (2004). Wolff in Flick et a . July, 178–183.

Hemmati, S., & Rabbani, M. (2010). Make-to-order / make-to-stock partitioning decision using

the analytic network process. 801–813.

Karlsson, C. (2016). Research Methods for Operations Management (Second Edi).

Köber, J., & Heinecke, G. (2012). Hybrid production strategy between make-to-order and make-to-stock - A case study at a manufacturer of agricultural machinery with volatile and seasonal demand. Procedia CIRP, 3(1), 453–458.

Leonard-Barton, D. (1990). A Dual Methodology for Case Studies: Synergistic Use of a Longitudinal Single Site with Replicated Multiple Sites. Organization Science, 1(3), 248– 266.

Nallusamy, S., Balaji, R., & Sundar, S. (2017). Proposed model for inventory review policy through ABC analysis in an automotive manufacturing industry. International Journal of

Engineering Research in Africa, 29(May), 165–174.

Olhager, J. (2003). Strategic positioning of the order penetration point. International Journal

of Production Economics, 85(3), 319–329.

Perona, M., Saccani, N., & Zanoni, S. (2009). Combining make-to-order and make-to-stock inventory policies: An empirical application to a manufacturing SME. Production

(35)

34 Romsdal, A., Arica, E., Strandhagen, J. O., & Dreyer, H. C. (2013). Tactical and operational issues in a hybrid MTO-MTS production environment: The case of food production. IFIP

Advances in Information and Communication Technology, 398(PART 2), 614–621.

Romsdal, A., Strandhagen, J., & Dreyer, H. (2014). Can Differentiated Production Planning and Control enable both Responsiveness and Efficiency in Food Production?

International European Forum on System Dynamics and Innovation in Food Networks,

324–333.

Soman, C. A., Van Donk, D. P., & Gaalman, G. (2004). Combined to-order and make-to-stock in a food production system SOM-theme A: Primary processes within firms. Int.

J. Production Economics, 90, 223–235.

Syntetos, A. A., Keyes, M., & Babai, M. Z. (2009). Demand categorisation in a European spare parts logistics network. International Journal of Operations and Production

Management, 29(3), 292–316.

Tanco, M., Santos, J., Rodriguez, J. L., & Reich, J. (2013). Applying lean techniques to nougat

fabrication : a seasonal case study. 1639–1654.

Trentin, A. (2011). Third-party logistics providers offering form postponement services : value

propositions and organisational approaches. 7543.

Van Donk, D. P. (2001). MTS or MTO : The DP in the food processing industry. International

Journal of Production Economics, 69(3), 297–306.

Van Kampen, T. J., Akkerman, R., & Van Donk, D. P. (2012). SKU classification: A literature review and conceptual framework. International Journal of Operations and Production

Management, 32(7), 850–876.

Van Kampen, T. J., & Van Donk, D. P. (2014a). Coping with product variety in the food processing industry: The effect of form postponement. International Journal of

Production Research, 52(2), 353–367.

Van Kampen, T. J., & Van Donk, D. P. (2014b). When is it time to revise your SKU classification: Setting and resetting the decoupling point in a dairy company. Production

Planning and Control, 25(16), 1338–1350.

Zhang, Z. G., Kim, I., Springer, M., Cai, G., & Yu, Y. (2013). Dynamic pooling of make-to-stock and make-to-order operations. International Journal of Production Economics,

(36)

35

Appendix A: Interview protocol

In this appendix, the used Interview protocol is presented

Opening

1. Introduction of interviewers and interviewee 2. Confidentiality assurance

3. Permission to audiotape

General

1. Interviewee job-title

2. Could you describe the company?

a. Characteristics such as; size, market share and locations b. Number of employees (plant and organization)

c. Turnover plant in terms of money and volume d. Location in entire supply chain

3. Could you give a rough description of the manufacturing process from raw material to final products?

4. What makes your plant more successful than your competitors? a. What is the driving force in your production planning? 5. At what point do your products become customer specific? 6. Do you produce private label products?

7. Do you sub-divide products in MTO, MTS or Mix-to-order (ATO)? a. Approximately what percentage of products is produced to order?

Product planning

1. Do you subdivide products in product families or groups?

2. Which steps do you take to create your production planning per product group? a. What is the timeline

b. Who are involved?

c. How do you deal with capacity constraints? d. What are your process constraints?

(37)

36 3. What criteria or guidelines do you use to determine which product to allocate capacity to

first?

4. Are there certain production lines that are specific to one product or are they flexible? 5. What is your planning horizon (strategic, tactical, operational) and how do you translate

your long-term planning to a detailed scheduling?

Product Classification / positioning the CODP.

1. What are the criteria in deciding which products are classified as MTO or MTS a. What product related characteristics do you use when making this decision?

i. Holding costs (suggestion) ii. Perishability (suggestion)

iii. Modularity of design (suggestion) iv. Product customization (suggestion)

b. What market related characteristics do you use when making this decision?

i. Demand predictability (suggestion) ii. Delivery lead time (suggestion) iii. Demand volatility (suggestion)

iv. Payment reliability (suggestion)

v. Order frequency or volume (suggestion)

c. What process related characteristics do you use when making this decision?

i. Production lead time (suggestion) ii. Production flexibility (suggestion) iii. Production setups (suggestion)

2. Do these market, product and process characteristics change over time? a. Do certain changes repeat every year?

i. Seasonal demand (suggestion) ii. Seasonal raw materials (suggestion)

b. Is there a certain change-pattern that you can describe?

i. Length of seasons (suggestion)

ii. Predictability of changes (suggestion)

(38)

37 4. Can you describe how you decide on the CODP? How are these market, process and

product characteristics used in positioning the CODP? What methods are used?

i. Using certain factors (suggestion) ii. Qualitative framework (suggestion) iii. Quantitative methods (suggestion)

5. How do certain customer-specific needs used in positioning the CODP? Are customer specific needs used for positioning the CODP? Or is this mainly determined by the product characteristics.

i. Attractiveness of customers/ strategic influences (suggestion) ii. Customer service level (suggestion)

6. How does your process of positioning the CODP differences from the varies from the regular production industry? What industry-specific characteristics needs to be taken into?

i. Product perishability (suggestion)

Reclassification of products

1. When is an MTO-MTS product classifcation reviewed or changed? a. Time driven?

b. Event driven?

c. Is there a limit to how often you can change a product classification?

d. What would you describe as an appropriate interval for reviewing/ changing the MTO-MTS classifications?

2. What is the decision-making processes of a reclassification and how is this organized? a. How are involved?

b. What kind of method or structure is used for a reclassification? c. How would you describe the level of structure of this decision? 3. What are the objectives of changing a product classification?

a. From MTO to MTS b. From MTS to MTO

Understanding the characteristics affecting the classification over time within combined MTO-MTS production systems.