MSc Thesis - University of Tilburg - Tilburg University [PDF]

MSc Thesis Business Intelligence systems to mitigate information load with store managers

Document

:

MSc Thesis

Author

:

M.J. van Strien BICT

Supervisors

:

dr. A-F Rutkowski J. Pluyter MSc dr. B.A. van de Walle

Organisation

:

Tilburg University

Date

:

13-12-2010

MSc Thesis Business Intelligence systems to mitigate information load with store managers Thesis submitted for the MSc degree

Document

:

Version

Author

MSc Thesis Final

:

M.J. van Strien s697471 [email protected]

Supervisors

:

dr. A-F Rutkowski J. Pluyter MSc

Exam committee

dr. A-F Rutkowski dr. B.A. van de Walle

Study

:

Tilburg University Tilburg School of Economics and Management Master Information Management

Date

:

13-12-2010

Abstract Technological advances have created a more global, interconnected and complex business environment, while also increasingly enabling managers to generate and collect data. Due to the emerging ubiquity of information systems (IS‟s), today‟s managers are facing an information-rich environment when making decisions. Although this cannot ensure that correct decisions will be made, human information processing theory predicts that the availability of information positively affects a manager‟s ability to recognize the need for a decision, develop decision alternatives, and select a course of action from these alternatives. However, because an individual‟s ability to process information is limited, this effect is up to the point at which information overload occurs. From the point of information overload, decision makers will apply a less accurate heuristic decision strategy and decision quality is suggested to decrease rapidly. This research investigates how business intelligence (BI) systems can mitigate information load with store managers in order to increase their decision quality. It proposes that information overload is a result of inexperienced decision makers, decision making under time pressure, and receiving a large amount of information. In order to investigate the information overload concept with store managers, a caste study research was conducted. This research evidences that store managers do not experience information overload, and consequently the quality of their decisions does not decrease from information overload. Although they acknowledge that the information load is increasing, experience - both in the using IS‟s and domain knowledge - enables them to chunk the information and prevent them from being overloaded with information. Also the use of IS‟s decreases the amount of information that a store manager is required to process for decision purposes. Particularly the use of BI technology for order advices allows store managers to perform interactive analysis on data that was gathered data from various dispersed sources. Contextualization of the information allows store managers to adapt the information to their preferences and requirements of the task. This reduces the effort a decision maker invests in acquiring the information. This is key in preventing store managers from being overloaded.

Management samenvatting Technologische

ontwikkelingen

hebben

een

meer

globale,

verbonden

en

complexe

bedrijfsomgeving, terwijl ze managers ook meer en meer de mogelijkheid bieden om gegevens te creëren en te verzamelen. Door de alom aanwezigheid van informatiesystemen (IS) hebben hedendaagse managers te maken met een informatierijke omgeving bij het maken van beslissingen. Hoewel dit niet kan garanderen dat managers goede beslissingen maken, voorspelt informatieverwerkingstheorie dat de beschikbaarheid van informatie een positieve impact heeft op het vermogen om een beslissingsbehoefte te herkennen, alternatieven te ontwikkelen en één van deze alternatieven te selecteren. Maar aangezien het menselijke vermogen om informatie te verwerken beperkt is, is dit effect geldig tot op het punt dat information overload optreedt. Vanaf dit punt past de beslissingsmaker een minder accurate beslissingsstrategie toe waardor de beslissingskwaliteit snel verslechtert Deze scriptie onderzoekt hoe business intelligence (BI) systemen information load bij supermarktmanagers kunnen verminderen teneinde de beslissingskwaliteit te verbeteren. He stelt voor dat information overload een resultaat is van onervaren beslissingmakers, beslissingen onder tijdsdruk en een grote hoeveelheid informatie. Teneinde het bestaan van information overload bij supermarktmanagers te onderzoeken is er een case study uitgevoerd. Hier wordt aangetoond dat supermarktmanagers geen information overload ervaren en dat de beslissingskwaliteit niet verslechterd door information overload. Alhoewel de supermarktmanagers erkennen dat de informatielast toeneemt stelt ervaring – zowel in het gebruik van IS en domeinkennis – ze in staat om informatie te filteren en ze zodoende te verschermen van overbelasting. Daarnaast vermindert het gebruik van IS de hoeveelheid informatie die een supermarktmanager moet verwerken voor de beslissing. Vooral het gebruik van BI technologie voor besteladviezen laat de supermarktmanagers toe om interactieve analyses uit te voeren op data uit verspreide informatiebronnen. Contextualisatie van de informatie staat de supermarktmanagers toe om informatie aan te passen aan hun preferenties en eisen van de taak. Dit vermindert de inspanning die een beslissingsmaker in de acquisitie van informatie moet investeren. Dit is essentieel in het voorkomen van overbelasting met informatie.

Preface Ironically, the process of writing this thesis has caused information overload frequently – or maybe even continuously. Wading through the information-rich and dispersed knowledge-bases of scientific research to research one of the most intriguing subjects of the past time was truly pleasurable. But, as science predicts, having an information overload has proven to be frustrating sometimes.

However I am sincerely happy to present the end-result of my first intimate experience with scientific research. Fortunately I did have the access to some well-designed decision supporting systems in the form of my supervisory board. My thanks go out to all the people that have accompanied me towards a satisfactory end-result, either through supervision or support.

Particularly I want to thank dr. Anne Rutkowski and Jon Pluyter in guiding me through the challenges of scientific research methodology, the interesting field of behavioural decision making research, and most of all their commitment and time to clear my mind from being overloaded. Secondly I want to thank Eelco Rouw for supervising me during the start-up and research design. His viewpoint has broadened my view on the concept of information overload, often causing overload from his creativity. I also had the pleasure to have my own supervising board of consisting of Ernst & Young colleagues. Bas van de Raadt, Knut Grahlman and Roel Drost have provided me with scientific knowledge, practical experience and critical remarks that have shaped this research. Last, but certainly not least, I want to thank the respondents that have taken the time to answer my questions. Without their answers that provided a basis for the evidence, I would not have been able to conduct this research.

I hope you enjoy reading this thesis as much as I enjoyed writing it.

Maarten van Strien December 2010

Table of contents

List of figures

A

List of tables

A

1. 1.1 1.2 1.3 1.4

Introduction Problem statement Theoretical background Relevance of research Reading guide

1 3 4 5 6

2.

Research methodology

7

3. 3.1 3.2

Conceptual model Dependent variable Independent variables 3.2.1 Decision-maker experience 3.2.2 Time pressure 3.2.3 Information amount 3.3 Mediating variable 3.4 Exogenous factors

11 12 12 12 13 14 14 15

4. 4.1

Support in managerial decision-making Organizational decision-making 4.1.1 Decision styles 4.1.2 Decision pyramid 4.2 Decision-making process 4.2.1 Bounded rationality 4.2.2 Heuristic decision strategy 4.3 Information-seeking 4.4 Computerized decision support

17 17 18 18 19 20 21 22 23

5. 5.1

25 25 26 27

Business Intelligence as decision support systems Business Intelligence 5.1.1 Defining BI 5.1.2 Data warehouse

5.1.3 BI Tooling 5.1.4 Real time Business Intelligence 5.2 Decision effort

29 30 31

6. 6.1

Information processing and overload Information processing 6.1.1 Individual differences 6.1.2 Processing strategies 6.2 Information processing ability 6.2.1 Information processing requirement 6.2.2 Human processing capacity 6.3 Information overload 6.3.1 Causes 6.3.2 Decision quality 6.3.3 Mitigation

33 33 34 35 35 36 37 38 39 41 42

7. 7.1 7.2

Analysis Case description Store managers 7.2.1 Independent variables 7.2.2 Mediating variable 7.2.3 Dependent variables 7.3 System designers

43 43 44 46 54 58 61

8. 8.1

Conclusion & discussion Conclusions 8.1.1 Information processing capacity 8.1.2 Information processing requirements 8.1.3 Overall conclusion 8.2 Discussion & limitations 8.2.1 Research validity 8.2.2 Disability to test propositions 8.2.3 Improper measurement for decision quality 8.2.4 Search costs of BI 8.3 Recommendations for future research

65 65 66 67 68 69 69 70 71 71 72

References

A

Appendix I: Research strategy

I

Appendix II: Measurement validity

XIV

Appendix III: Interview protocol store managers

XIX

Appendix IV: Interview protocol system designers Semi-structured Questionnaire Measure validity

XXVII XXVII XXVIII XXXI

Appendix V: Quantitative evidence Store managers Systems designers

XXXIII XXXIII XLIV

Appendix VI: Content analysis

XLVI

Appendix VII: Quotes systems designers

LVIII

Appendix VIII: Empirical Model

LXIII

List of figures Figure 2.1 Research model ............................................................................................................... 7 Figure 3.1 General conceptual model ............................................................................................ 12 Figure 3.2 Exogenous factors ......................................................................................................... 16 Figure 5.1 General data warehouse architecture (Turban et al., 2007) .......................................... 28 Figure 6.1 Relationship between information processing ability and information load ................ 36 Figure 6.2 Mapping of information overload causes to the variables ............................................ 41 Figure 7 Convergence and non-convergence; from Yin (1989) .................................................... IX Figure 8 Empirical model ......................................................................................................... LXIII

List of tables Table 4.1 Levels of decision-making and characteristic attributes (Harrison, 1987) .................... 19 Table 4.2 Phases in the general decision making process .............................................................. 20 Table 5.1 Definitions of BI ............................................................................................................ 26 Table 5.2 Data warehouse characteristics (Turban, Aronson et al. 2007) ..................................... 27 Table 6.1 Definitions of information overload (Eppler and Mengis, 2004) .................................. 38 Table 6.2 Causes of information overload (Epler and Mengis 2004) ............................................ 39 Table 7.1 Categorization for the content analysis .......................................................................... 45 Table 7.2 Descriptive statistics for experience and time pressure ................................................. 46 Table 7.3 Descriptive analysis for information load and time pressure ......................................... 49 Table 7.4 Descriptive statistics for information amount (1/2) ....................................................... 50 Table 7.5 Descriptive statistics for information amount (2/2) ....................................................... 50 Table 7.6 Descriptive statistics for Russel and Mehrabian's scale ................................................. 55 Table 7.7 Descriptive statistics for Rutkowski and Saunder‟s scale .............................................. 55 Table 7.8 Descriptive statistics for dependent variables ................................................................ 58 Table 15 Criteria for judging research design quality; (Yin 1989; p.40) ........................................ V Table 16 Respondent matrix ........................................................................................................ VII Table 17 Scale for received information amount .......................................................................XVII

A

Table 18 Scale for time indication of information load .............................................................XVII Table 19 General information load measurement scale .............................................................. XIX Table 20 Aspects of information overload in the system design ............................................. XXXI Table 21 Questionnaire data for independent variables, including frequency table ............. XXXIII Table 22 Descriptive statistics for independent variables ..................................................... XXXIV Table 23 Scale explanation for independent variables ........................................................... XXXV Table 24 Questionnaire data for information amount; perceived quantity ........................... XXXVI Table 25 Questionnaire data for information amount; perceived frequency ....................... XXXVII Table 26 Descriptive statistics for information amount; perceived quantity ..................... XXXVIII Table 27 Descriptive statistics for information amount; perceived frequency .................. XXXVIII Table 28 Questionnaire data for information load measure .................................................. XXXIX Table 29 Descriptive statistics for information load scale ............................................................ XL Tabel 30 Descriptive statistics for information load variable ....................................................... XL Table 31 Questionnaire data for information load variable; including frequency table ............. XLI Table 32 Questionnaire data for dependent variables; including frequency table ..................... XLII Table 33 Descriptive statistics for dependent variables ............................................................ XLIII Table 34 Questionnaire data for systems designers ................................................................. XLIV Table 35 Mapping the quotes with categories.......................................................................... XLVI Table 36 Abbreviation for the categories of the content analysis ........................................... XLVII Table 37 Content analysis quotes............................................................................................ XLVII Table 38 Quotes from interview with C1000 Retail designer .................................................. LVIII Table 39 Quotes from interview with Plus Retail desinger .......................................................... LX

B

C

BI systems to mitigate information load with store managers

1. Introduction In the last decades, the environment in which organizations operate is becoming more and more complex. Developments in Information Technology (IT) - and more particularly the Internet have led to increasingly more global, complex and connected business environments (Huber, 2003; Shim, Warkentin, Courtney, Power, Sharda and Carlsson, 2002). Hence organizations will interact with a greater diversity of cultural, political, social, economic and ecological environments. Mitroff and Linstone (1993) argue that managers of organizations facing such an environment need a radical change in thinking when facing decision making. Turban, Aronson, Liang and Sharda (2007) present a business pressures-responses-support model to address the various factors (both threats and opportunities) in an organization‟s environment. This model illustrates how Business Intelligence (BI) systems can help address these factors by providing the ability to conduct appropriate analysis in managerial decision-making. BI‟s major objective is to provide computerized decision aid by enabling interactive access to data from various sources. BI architecture and tooling enables manipulation of data in order to provide managers with the means to conduct interactive analysis (i.e. business analytics). The manipulation process is based on the transformation of data into information. Computerized decision support methodology recognizes the need for data in order to solve problems and provide decision aid (Turban et al., 2007).

Interactive analysis - based on historical and current data, situations and performances - allows decision-makers to get valuable insights into business‟ situations. Through interactive analysis, BI enables managers to be more informed, and make better decisions (Zaman, 2005). So, by providing computerized decision support when managers are facing decision problems that require a large amount information to be processed, BI intends to improve decision quality (Shim et al., 2002).

Pg. 1

MSc thesis M.J. van Strien

Introduction

Hence, when considering a more holistic view on organizational decision making, BI systems should also include data about a business‟ environment in order to provide optimal aid. Cheung and Babin (2006) support this as they suggest that delivery of accurate decision support from internal and external data sources is a vital requirement for decision making. Additionally, Courtney (2001) suggests that decision support researchers should embrace a more comprehensive view of organizational decision making. Herewith the author proclaims the development of Information Systems (IS‟s) for decision aid that are capable of handling “softer” information and have a broader scope than traditional mathematical models. Courtney (2001) refers to this approach as a new decision aid paradigm. In order to provide an optimal support for decision-making in today‟s complex business environments, BI design should adopt the new decision aid paradigm, and include both internal and external data sources. Cheung and Babin (2006) note that data warehousing has been adopted as a key technology in decision aid as it enables to consolidate data from multiple sources into one multidimensional repository (i.e. data warehouse). Data warehouses have emerged to meet managers‟ need for integrated and high quality information that is presented timely and convenient (March and Hevner, 2007).

Due to emerging ubiquity of web-connection and advances in data warehousing technology, today‟s managers face richer information environments (Lurie, 2004), and have easier access to BI systems (Shim et al., 2002). This has resulted in a situation in which individuals have access to more information than ever. Although there are obviously some benefits from more and easier accessible information, other technological developments (e.g. web 2.0 and social media (Bawden and Robinson, 2009) produce, manipulate and disseminate information much faster than the human brain can process. Researchers across various disciplines have found that the performance of decision makers (i.e. the decision-quality) is positively related with the amount of information an individual receives - up to a certain point. The information that is received beyond this point will not be incorporated into the decision making process (because supply exceeds capacity) and will result in information overload (Rutkowski and Saunders, 2010; Eppler and Mengis, 2004). Occurrence of information overload will confuse the decision maker, affect

Pg. 2

BI systems to mitigate information load with store managers

his or her ability to set priorities and make prior information harder to recall – resulting in rapid declining of the decision quality (Eppler and Mengis, 2004).

1.1

Problem statement

Technological advances have enabled IS‟s to provide an increasing amount of information to a wider target of individuals, placing managers in an information-rich environment when considering decisions-making. Edmunds & Morris (2000) note that this leads to a paradoxical situation; “although there is an abundance of information available, it is often difficult to obtain useful, relevant information when it is needed”. Koniger & Janowitz (1995) argue that this contradicting situation arises because of the high information load individuals receive is making it more difficult to extract relevant information. Hence decision-makers perceive that they do not receive enough information.

So, on the one hand the increasing information availability is beneficial as researchers are calling for a more comprehensive, holistic and integrated approach on BI design in order to better support managerial decision making in an organization‟s complex business environment. Courtney (2001) argues that in this more comprehensive view on decision making, managers should also incorporate external (e.g. economic, social and environmental) data into their decisions making processes. A more holistic approach to BI design should provide managers with more information and should theoretically improve strategic decision making quality. On the other hand however, researchers from various disciplines claim that abundance of information instead of providing better decision aid – threatens to diminish a manager‟s control over the situation (Edmunds and Morris, 2000), ultimately decreasing decision quality. Williams, Dennis, Stam and Aronson (2007) show that decision makers – when using a computerized decision support - may develop a better understanding of the decision problem and therefore feel more confident and satisfied. However, as they note, the assumption that use of computerized decision support improves decision quality is “tenuous at best”. Reviews of the

Pg. 3

MSc thesis M.J. van Strien

Introduction

literature indicate that the influence of computerized decision support on decision performance has been mixed; in some studies performance improved, in others there was no effect, and in still others decision quality worsened when computerized support was employed (Todd and Benbasat, 1992). Concluding, BI provides decision makers with the tools to improve decision quality (Turban et al., 2007), but like any other tool, “their impact depends on how well they are used” (Königer and Janowitz, 1995).

1.2

Theoretical background

Human information processing theory predicts that information processing performance will increase when information load increases. However behind the point of information overload processing performance will decrease rapidly, up to the point of information overload. A number of studies have tested the relationships between information load, information processing and decision quality (e.g. Simnett, 1996; 1990; Iselin, 1988) but their results have been inconsistent (Hwang and Lin, 1999). Eppler and Mengis‟ (2004) literature study suggests that information overload will decrease a manager‟s control over information processing and result in declined quality of the decisions.

Various researchers address specific countermeasures to alleviate the effects of information overload in organizational decision environments (Eppler and Mengis, 2004). Additionally Brueggemann, Reader and Brannan (2008) take a more abstract approach in order to mitigate information overload as they call for a more context-based approach to the design of computerized decision support. This approach encompasses incorporation of the decision maker‟s and the decision-task‟s context (e.g. nature of the task, goals, roles, and experience) into the BI system. This multi-disciplinary approach integrates psychological, sociological and human factors into a highly technical field. The authors evidence a positive effect on a decision maker‟s capacity to handle information more effectively and efficiently.

Pg. 4

BI systems to mitigate information load with store managers

Previous research on computerized decision support has focused on technologies that assist managers in making decisions. Many of these technologies can be considered static tools since the data presentation and analysis are predetermined by the designers of the technology and unlikely to change after implementation (Datta and Thomas, 1999). Static refers to the fact that system designers predetermine the way a decision-maker can view or analyze data (Tremblay, Fuller, Berndt and Studnicki, 2007).

Unlike earlier technologies,

BI tools provide decision makers with a dynamic information

environment which allows decision makers to customize the selection, aggregation and presentation of information (Tremblay et al., 2007), theoretically enabling decision makers to adapt information to fit into the required context. March and Hevner (2007) note that BI enables decision makers to integrate that data with respect to a specific business process – as which they refer to as „contextualization‟ of the data.

1.3

Relevance of research

To make a quality decision, decision-makers must have quality information pertinent to the decision at hand (Park, 2006). But, living in an “information society” means also that managers are bombarded with information whether or not they actively seek it. Since, human rationality is limited, a certain subset considered to be relevant is selected from the mass of all potentially useful (Klausegger, Sinkovics and Zou, 2007). Several studies have shown that information and system quality affect an individual‟s performance. Thus, improving system quality should lead to enhanced decision quality (Park, 2006). The designers of IS‟s often do not consider the capabilities and limitations of the decision maker. As a result, the IS‟s with vast information processing abilities tend to overload the human user. This obviously interferes with proper functioning of the decision maker. IS‟s can be improved by understanding the behavioural processes by which humans process information and make decisions (Benbasat and Taylor, 1982). Understanding how people search through and combine

Pg. 5

MSc thesis M.J. van Strien

Introduction

information before making decisions is an important concern in the design IS‟s that support decision making (Svenson, 1979). Additionally, it is necessary to understand the differences in the needs of managers in a given organization because the quality of the decision they make is directly tied to the information they receive (Davis, 2005). From the point of view of information suppliers, an understanding of the effect of information dimension on information load and decisions quality is critical (Hwang and Lin, 1999; Casey, 1982). Because of the moderating effect of information load on decision quality – information suppliers and decision makers should be interested in the information load (Schroder, Driver and Streufert, 1967).

The implication for information suppliers is that more is not necessarily better in the case of information amount. Hwang and Lin (1999) evidence that even a moderate increase in information dimension can hinder decision quality. This is considered detrimental since managers have a tendency to want as much information as possible. Hence, presenting the right amount of information may be a challenging, but yet critical, task for information suppliers. Also, the point on which information overload occurs is not clear. Hwang and Lin (1999) argue that the optimal number of information dimensions for each task should be determined empirically. Former research results provide a good reference point for this specific amount. It however has often been too detached from specific contexts, and information suppliers should not extrapolate the values to other tasks.

1.4

Reading guide

This chapter has discussed the theoretical background, resulting a relevant problem statement. The following chapter will present the research methodology which we have applied for the present research. Thereafter the conceptual framework, the basis for this research, will be presented. The next three chapters will address the theoretical basis that was found to be necessary to conduct the research. Hereby we focus on managerial decision making, how BI systems can support decision making, and how information load affects in individual‟s information processing ability. Concluding we will present the analysis of our case study research and the conclusions and discussions that were drawn from the research. Pg. 6

BI systems to mitigate information load with store managers

2. Research methodology In all academic research, researchers need a strategy that supports them in the development and conduction of the research (Maimbo and Pervan, 2005). Following Yin (1989), we refer to this as the research design – which represents a logical set of statements. Davis (2005; p.134) claims that “the formal specification of a research design is an integral part of the research”. For matters of overview, the research design is divided into three components: model construction, data collection (case study), and analysis (see Figure 2.1).

Figure 2.1 Research model

Following Kumar (Kumar, 2005) we argue that the most suitable research method is contingent upon the research purpose. Therefore the objective of this research will first be elaborated. The objective is defined as “to present guidelines on how Business Intelligence (BI) systems can be employed in order to alleviate information overload”. The research will be conducted within the context of managerial decision-making activities of retail organizations‟ store managers. Since information overload is theorized to decrease decision quality the present research intends to provide guidelines that allow BI systems to increase store managers‟ decision-making quality.

When decomposing the research purpose, the research will: (1) investigate the concept of information overload at store manager‟s, and (2) its relationship with the quality of their decisions,

Pg. 7

MSc thesis M.J. van Strien

Research methodology

(3) investigate how BI systems can be deployed to alleviate information overload, and support store managers to incorporate all relevant information into their decisions, and (4) present recommendations on how information systems (IS‟s) designers might employ BI systems to fit decision makers‟ information-processing capacities. Whether a study‟s purpose is exploratory, descriptive, or hypothesis testing, depends on the stage to which knowledge about the research topic has advanced (Sekaran and Bougie, 2010). The concept of information overload already has a broad theoretical basis (Eppler and Mengis, 2004), hence we classify the present study as descriptive (i.e. explanatory) for the specific context.

To be able to achieve the research purpose, one structured research questions as well as sub research questions are defined. The research questions bring clarity (i.e. a well-defined focus) to the research, and supports to focus research efforts (Eisenhardt, 1989). Drawing on the research purpose, we define the research question as follows: “How can BI systems decrease information overload, in order to support increasing store managers’ decisions quality?”

Additionally we define six sub-questions: 1. How do managers make decisions? 2. When does information overload occur? 3. How can information overload be mitigated? 4. Do store managers experience information overload? 5. Does the occurrence of information overload decrease decision quality? 6. How can BI systems help store managers to tailor the information to their specific needs?

Questions 1, 2 and 3 will be addressed theoretically, while questions 4 and 5 will be addressed empirically. A combination of theoretical and empirical findings will be used to answer question 6.

Pg. 8

BI systems to mitigate information load with store managers

The following directional hypotheses are proposed to answer the empirical research questions. The direction of these propositions is integrated into the conceptual model which is depicted in Figure 3.1.

P1: Information overload will decrease decision quality P1a: Decision maker experience will decrease information load P1b: Decision-making under time pressure will increase information load P1c: High information load will increase information load P2: Decreased decision quality will decrease business performance

To ensure the purposefulness (i.e. to add value to existing scientific work) we need to draw our conclusions on evidence from real-life data (Kumar, 2005). Information processing behaviour from a decision-making perspective, and more specific information overload, is hard to investigate independent from a particular context (Eppler and Mengis, 2004; Rouse and Rouse, 1984). Eppler and Mengis (2003; p.35) argue that prior research has often been “too detached from the specific overload contexts”. The authors argue that future information overload research should encompass an extensive contextual investigation, rather than yet another survey or experiment. We argue that our research purpose requires a context-specific approach; hence we adopt a CSR strategy to investigate the information overload phenomenon in the context of store managers.

We acknowledge some prerequisites which our case study should meet in order to address the research purpose; (1) measurability, (2) decision complexity, and (3) decision autonomy. These conditions have directed us towards selecting a retail chain with a limited franchise strategy: C1000 retail. In this way, the various supermarkets are comparable, while the entrepreneurial freedom allows for certain decision autonomy.

Measurability of the decision quality is found to be complex (Keller and Staelin, 1987). For matters of scoping we have chosen to focus on decisions regarding inventory management. This is considered a structured decision process (i.e. the decisions maker is aware of the decision

Pg. 9

MSc thesis M.J. van Strien

Research methodology

activities and it has a clear goal system (Iselin, 1988; Gorry and Scott-Morton, 1979) and this goal system allows performance to be measured objectively. Decision complexity is needed to investigate an information overload „sensitive‟ decision, but could only be predetermined theoretically. Preliminary investigation concludes that inventory management – although being structured - encompasses some complex decisions. Obvious it is an interplay between actual inventory levels and sales forecasts, on product level. But it also relates to information regarding sales targets, financial performance and decisions regarding pricing/promotions, store lay-out and service levels (e.g. stock-out probability). Moreover, it is closely bound to time (Levy and Weitz, 2007).

Concerning decision autonomy, Trusov, Bodapati, and Cooper (2006) argue that store managers receive emerging support from information systems that provide aid on the order quantity and time. The store manager however remains responsible for information management activities, and has full decision autonomy (van Strien, 2010). Additionally van Donselaar, van Woensel, Broekmeulen and Fransoo (2006) witness that automated order systems (AOS‟s) automatically place orders without any intervention needed. However these AOS‟s contain user-determined business-rules. Moreover, Waller (1999) argues the vendor-managed inventory (VMI) strategy being a successful inventory management strategy in grocery retailing. VMI centralizes decision autonomy (i.e. at the vendor, supplier, or distribution centre), hence it diminishes the store manager‟s autonomy. The case we investigated (C1000 Retail) uses order advice systems (OAS‟s) to support the store managers. OAS‟s, different from AOS‟s, do not automatically place orders. Their advice still has to be revised by the store manager. Also C1000 Retail has presented plans to implement a VMI strategy, but this will take some time before it is operational. Hence we conclude that our case satisfies all the prerequisites.

Pg. 10

BI systems to mitigate information load with store managers

3. Conceptual model The conceptual model shows how we theorize the interrelationships between the variables that are considered integral to the situation being investigated (Sekaran and Bougie, 2010).The foundations for this framework lie in literature survey (Maimbo and Pervan, 2005). The framework is be depicted schematically in Figure 3.1. Additionally, textual support will be used to (1) clarify theoretical interrelatedness between variables, (2) state the theorized nature and direction of relationships, and (3) give an explanation of the expected relationships (Sekaran and Bougie, 2010). The conceptual model forms the basis upon which the research is conducted, is used to derive propositions, and provides a framework to determine which data is analysed (Maimbo and Pervan, 2005).

In constructing the model, we first need to determine the independent and dependent variables. As discussed, we investigate a store manager‟s decision quality. We propose “information load” as a mediating role between a decision context (decision-tasks, and support) and decision quality. In the decision context we propose three variables that determine the information load: (1) decision-making experience, (2) time pressure, and (3) information amount. To investigate the role of information load on decision quality, the above variables are the independent variables, while decision quality is considered the dependent variable. Since decision quality is complex to measure, we propose that business performance is dependent upon the decision quality. Concluding, we propose that information load affects the decision quality, hence we consider information load as the mediating variable.

The conceptual model is based on a broad literature study, however, the model is supported by one phrase in Eppler and Mengis' (2003) literature review. The authors relate increasing information load with (1) less reoccurring routine (i.e. an individual's experience), (2) greater time pressure, and (3) high information amount. The authors conclude that the combination of these factors can lead to information overload.

Pg. 11

MSc thesis M.J. van Strien

Conceptual model

Figure 3.1 General conceptual model

3.1

Dependent variable

The aim of the research is to understand and describe/explain variability of the dependent variable; decision quality (Sekaran and Bougie, 2010). The correctness of the decision was often addressed in previous studies. Researchers have referred to different terms; e.g. decisions effectiveness (Keller and Staelin, 1987), decision performance (Paquette and Kida, 1988), decision quality (Williams et al., 2007; Raghunathan, 1999), and decision outcomes (Olson and Parayitam, 2007). Although these terms are literarily different, we argue that - after closer examination - they all intend to measure the correctness of the decision. For the present study, we refer to decision quality. Decision quality was found difficult to measure, therefore we searched for an effective measure to determine decision quality. We argue that business performance is dependent upon the manager‟s decision quality. Hence we propose that decision quality can be measured by business performance. This measure is elaborated in chapter 2.

3.2

Independent variables

3.2.1

Decision-maker experience

Eppler and Mengis (2004) argue that personal traits (e.g. attitude, qualification, and experience) are an important factor in occurrence of information overload. While early studies (e.g. Newell and Simon, 1972) simply state that an individual‟s capacity to process information is limited, Pg. 12

BI systems to mitigate information load with store managers

more recent studies argue that information overload is a “function of individual differences in information processing” (Rutkowski and Saunders, 2010; p.96). The latter view includes specific limiting personal factors; e.g., level of experience, attitude/motivation, and personal skill level (Eppler and Mengis, 2004). More specific, Russel and Mehrabian (1974) note that the information load of more improbable events (i.e. when a decision maker is less experienced) is greater. They find that a more experienced decision maker will experience a decreased information load as he or she is familiar with the received information. Researchers have suggested both positive and negative effects of a decision-maker‟s experience. Experts may be better at using and selecting relevant information, whereas novices may be more sensitive to novel information (Fisher, Chengalur-Smith and Ballou, 2003). Taylor (1975) claims that little evidence was found that older decision-makers – which are likely to be more experienced - are less capable of processing large amounts of information. We propose that decision-maker experience is an important determinant for information overload, and – based on prior findings - we suggest that more (less) experience will decrease (increase) information load.

3.2.2

Time pressure

Time pressure is an important factor in determining whether people have control over their information or not (Hahn, Lawson and Lee, 1992; Schick and Gordon Susan, 1990). Hahn, Lawson et al. (1992) argue that in the absence of the requirement to do the task in a hurried fashion (i.e. high time pressure), individuals will have control over their own processing systems. Contrasting, when people experience a pressure in the available time to execute the (information processing) task, they may not be able to continue adequate control over the processing of all available and relevant information. The authors argue that this may lead to information overload.

Traditional information overload theories take an information-processing approach to conceive the concept as a workload bottleneck; where there is too much to do in the time available (Woods, Patterson and Roth, 2002). Similar, Eppler and Mengis (2004) observe that the

Pg. 13

MSc thesis M.J. van Strien

Conceptual model

phenomenon has been explained by comparing the individual‟s information-processing capacity (IPC) with the information-processing requirements (IPR) (see chapter 0). This implies that “information overload can be explained via the following formula: information processing requirements > information processing capacity” (Eppler and Mengis, 2004; p.326). The terms “requirements” and “capacity” in this definition are determined in terms of available time. If the capacity of an individual allows a smaller than necessary amount of information to be processed in the available time slot, then information overload is the consequence. Hence, we propose that high (low) time pressure increases (decreases) information load.

3.2.3

Information amount

Schroder, Driver and Streufert‟s (1967) model relates information amount with human information processing. The model predicts that increases in information amount will increase information processing. However, if information amount keeps increasing – till it exceeds the individual‟s capacity – information processing will decrease. Hahn, Lawson and Lee (1992) argue that information amount is the most obvious variable to explain occurrence of information overload. Schneider (1987b) stresses the fact that not only the amount of information affects information load, but also specific characteristics of information (e.g. uncertainty, ambiguity, novelty and complexity). Moreover, Eppler and Mengis (2004) note that information amount is determined by four information factors; (1) quantity, (2) frequency, (3) intensity, and (4) quality (Schneider, 1987b). Eppler and Mengis (2004) state that the higher the information amount, the greater the probability of information overload. We propose that a higher (lower) perceived information amount will increase (decrease)information load.

3.3

Mediating variable

The mediating or intervening variable surfaces between the dependent and independent variable, and is proposed to explain the relationship between the independent and dependent variables

Pg. 14

BI systems to mitigate information load with store managers

(Sekaran and Bougie, 2010). We theorize that information load performs such a function between the decision context and decision quality.

In previous research, various researchers (e.g. Chewning and Harell, 1990; Iselin, 1988) have addressed the relationship between information overload and decision quality. However these studies have demonstrated inconsistent results (Hahn et al., 1992). For example, Simnett (1996) failed to report any significant changes in decision quality that could be related to information overload. In contrast, Shields (1983) found that the decision quality showed an inverted U-curve as information load changed. This means that increasing information load increases decision quality, up to a certain point where the decision maker‟s IPC is reached. Problems may occur when that limit is reached (Hwang and Lin, 1999). Various researchers refer to this phenomenon as information overload (Rutkowski and Saunders, 2010; Eppler and Mengis, 2004; Hwang and Lin, 1999). Occurrence of information overload will confuse the decision maker, diminish his/her control over the situation, and affect his or her ability to set priorities and make prior information harder to recall – resulting in rapid declining of the decision quality (Eppler and Mengis, 2004; Edmunds and Morris, 2000). Hwang and Lin (1999; p.213) argue that these contradicting findings “may be attributed to failure to rule out the effect of confounding variables”. We propose that occurrence of information overload will have a declining mediating effect on decision quality, i.e. it will decrease decision quality.

3.4

Exogenous factors

Besides information overload, other factors may also influence decision quality. Failing to control these factors, might lead to drawing improper conclusions (Hwang and Lin, 1999). We refer to these factors as exogenous. When designing the experiment, we have to control exogenous factors that might influence the relationship between the decision context and decision quality. The following factors are considered exogenous; information quality (Ge, 2009), decision models (Janis and Mann, 1977), decision autonomy, and decision maker quality (Raghunathan, 1999). These influencing factors are depicted in Figure 3.2

Pg. 15

MSc thesis M.J. van Strien

Conceptual model

Figure 3.2 Exogenous factors

According to Ge (2009) a researcher can deal with exogenous factors using two approaches; (1) considering the factors as independent variables, or (2) fixing them to one status. For scoping purposes we choose a controlling approach, but when possible we incorporate some of the influencing factors into the proposed independent variables. The exogenous factors are discussed below.

Information quality is found to directly influence inventory management decision quality (Ge, 2009), but also indirectly affects information load (Eppler and Mengis, 2004). Hence we include the subjective view on information quality in the information load variable (see appendix II). Decision models describe the model a decision maker applies in its decision making tasks. Janis and Mann (1977) argue that, based on (organizational) values and culture, decision makers may apply different ways of choosing between conflicting decision options. In order to investigate similar decision models, we select similar organizations. Raghunathan (1999) refers to decision-maker quality as the quality of the decision-making process, i.e. the processes that are undertaken, and more important how. The decision maker quality consists of a people and a process-component. The people component will be included in the decision maker experience variable. We argue that similar organizations have similar goals, and similar decision making processes. By selecting similar organizations (see chapter 2), we control both the decision autonomy and decision maker quality factors.

Pg. 16

BI systems to mitigate information load with store managers

4. Support in managerial decision-making Decision-making is an essential managerial action in any modern organization. In contrast to continually changing processes, environment, individuals and organizations, the necessity of decision-making seems to be the only constant in managerial action (Davis, 2005). Turban, Aronson, Liang and Sharda (2007) argue that, to be able to research a manager‟s decision-making activities, the researcher should understand the theories behind decision-making.. Additionally, understanding of the decision making process is necessary to investigate how various decisionmaking tools are incorporated in their decision processes (Martinsons and Davison, 2007). In order to obtain a theoretic basis, this chapter will provide some of the underlying theories and models of decision making.

4.1

Organizational decision-making

Simon (1977) defines decision-making as a process of choosing among two or more alternative courses of action for the purpose of attaining a goal. Managerial decision-making is synchronous with the whole process of management as for example controlling, planning, and organizing. Organizations - and more specific managers - usually determine multiple goals and make plans to achieve this desired state (Turban et al., 2007). This state is often different from the current situation or may become different in the future. A decision problem arises if a manager develops a conscious idea of a desirable state, and is faced with the problem of choosing between different courses of action to reach the target situation. A decision problem is the starting point for a conscious decision-making process (Grünig and Kühn, 2009). What is important is the fact that with the purpose of reaching the desired state, managers have to act rationally in dynamic and turbulent settings (Iselin, 1996, 1988). They, as decision makers, are expected to implement a high level of decision making quality, which depends on a variety of factors. On the one hand, the decision quality is dependent on the managers' knowledge, experience, ability to grasp analyze and understand a problem. On the other hand it depends on availability of information, decision support, and their use of decision making techniques and methods (Harrison, 1999).

Pg. 17

MSc thesis M.J. van Strien

4.1.1

Support in managerial decision-making

Decision styles

Depending on their style (i.e. preferences and experiences), different managers use different means for decision making (Davis, 2005). The author finds four distinct means for making decisions: research, intuition, authority, and experience. Research (either scientific or by consultants) provides the decision maker with relevant and useful information with which to make decisions. Research is specifically designed to address managerial problems with a systematic and controlled framework and aims to ensure the best possible information for decision making. The problem for research is that, when managers use intuition (e.g. instincts, and feelings) in decision making, they avoid hard systematic thinking and analysis of data, and thereby rely more on intuitive judgment (Tipuric and Prester, 2003). Also, decision makers may use different decision styles at different times to reach a decision in a particular situation (Davis, 2005). Consequently, the authors states that managerial decision making is hard to describe. Intuition

Authority

Decision-making process

Experience

Research

Figure 4.1 Major means of obtaining information for decision making (Davis, 2005)

4.1.2

Decision pyramid

Another implication in managerial decision making is that different levels of management use different approaches to the decision making process. Hence there is a heterogeneous information need within an organization (Davis, 2005). The author differentiates strategic, tactical and technical levels of decision-making to distinguish different decision structures in an organization.

Pg. 18

BI systems to mitigate information load with store managers

Multiple decision levels imply the existence of a decision making pyramid, which in case raises the question which decisions should be made by executive level, and which decision should be delegated to lower levels (Simon, 1997).

Table 4.1 outlines the three major decision levels and their related attributes. Although the classification based on these levels does not provide a clear-cut distinction, it does serve to highlight the important differences in managers‟ information needs in organizations. Table 4.1 Levels of decision-making and characteristic attributes (Harrison, 1987)

Decision-making level

Relative programmability

Information need

Strategic

Low

Heavy reliance on external information

Tactical

Limited programmability

Mixture of internal and external

Technical

High

Heavy reliance on internal information

Strategic decisions are those that fall within the sight of top management. Hence they are considered to be „important‟ to the organization, either because of the scope of their impact, or their long-term focus (Harrison, 1987). Generally speaking, strategic decisions affect the general direction of the organization, and are dealing with corporate level issues, e.g. diversification, product or market development, and divestiture. Tactical and technical decisions are those concerning planning, controlling and organizing activities that are derived from strategic decision outcomes.

4.2

Decision-making process

The problem of decision making has become a common meeting ground for psychological and economic research, diverging in the behavioural decision making research field . Attempts to develop a general model of decision making have proved frustrating. Most of the models proposed have received some empirical support, but none seems completely satisfactory as an explanatory model (Payne, 1976). The most accepted model was developed by Simon (1977) ,

Pg. 19

MSc thesis M.J. van Strien

Support in managerial decision-making

and describes an abstract, general decision making process that consists of four phases: (1) intelligence, (2) design, (3) choice, and (4) implementation (see Table 4.2).

Table 4.2 Phases in the general decision making process

Phase Intelligence Design Choice Implementation

Involved actions searching for conditions that call for decisions inventing, developing, and analyzing possible alternative courses of actions selecting a course of action from among those available adapting the selected course of action into the decision situation

Decision-making in organizations is often pictured as a coherent and rational process in which alternative interests and perspectives are considered in an orderly manner until the optimal alternative is selected. However, real decision behaviour and processes in organizations rarely fit such a description (Staw and Shapira, 1997). Simon (1955) refers to the term bounded rationality to present a more realistic view on the decision making process.

4.2.1

Bounded rationality

The notion of bounded rationality has become an important recognition in previous research and discussions about an alternative approach to traditional (i.e. rational) decision processes (Dequech, 2001). The rational model is based on a set of assumptions that prescribe how a decision should be made, rather than describing how a decision is made. While the bounded rationality framework views individual decisions makers as attempting to make rational decisions, it acknowledges that decisions makers often lack important information on the definition of the problem, and are guided by an aspiration level (Simon, 1957). Bounded rationality concludes that human inference is systematically biased and error sensitive. This suggests that laws of human inference – how humans make decisions – are “quick-and-dirty heuristics and not the laws of probability” (Gigerenzer and Goldstein, 1996; p.2). These psychological limitations – or boundaries – drop the actual human rationality striving that is implied by rational models (Kahneman, Slovic and Tversky, 1982). Boundaries in rationality affect the decision making process at (1) forming a set of decision alternatives, (2) the

Pg. 20

BI systems to mitigate information load with store managers

relationships that determine the pay-offs for each alternative, and (3) the preference-orderings among pay-offs (Turban et al., 2007; Simon, 1955). Time and cost constraints limit the quantity and quality of available information. Consequently managers apply a heuristic decision strategy and make a variety of systematically and predictable decision errors (Turban et al., 2007; Bazerman, 2006).

4.2.2

Heuristic decision strategy

The decision strategy is considered a method (i.e. a sequence of operations) that a decision maker uses to search through the problem space. Much research on decision making strategies focuses on how individuals process information in order to provide meaning in decision making. We distinct two types of decision strategies; heuristic and analytic strategies (Grünig and Kühn, 2009; Turban et al., 2007; Payne, Bettman and Johnson, 1993). Empirical studies on decision making indicate that decision makers are highly adaptive in selecting strategies (Payne et al., 1993; Johnson and Payne, 1985). Thereby decision makers typically trade off the amount of effort to be spent in making a decision (i.e. the use of resources1) and the benefits they expect from these efforts (Todd and Benbasat, 1992). Heuristic strategies are applied when any of the formal requirements of analytic procedures are not fulfilled (Grünig and Kühn, 2009). When applying a heuristic decision strategy, managers edit and simplify situations, thereby possibly ignoring important information and focusing on other information. They recognize patterns in situations and apply rules of thumb that are believed to be appropriate for the particular situation. Put different, they simplify complex phenomena, and treat this as an equivalent to the more complex reality they represent (Bazerman, 2006; Staw and Shapira, 1997). This makes a heuristic decision strategy less effortful in use, but also leads to a large group of decision making errors – making the strategy less accurate regarding the decision outcome (Bazerman, 2006). Effective information supply, e.g. through

1

Shapira and Staw (1997) have classified recourse-consuming groups: (1) observation; the gathering of information, (2)

memory; the storing of information, and (3) computation; the manipulation of information

Pg. 21

MSc thesis M.J. van Strien

Support in managerial decision-making

computerized decision aid, can support a decision maker in order to induce the use of a more accurate strategy (Todd and Benbasat, 2000).

4.3

Information-seeking

Humans seldom seek for information in itself, but instead it is part of the decision-making process (Rouse and Rouse, 1984). Decision making in management involves utilisation of information, hence managers need information efficiently and in a timely manner (Turban et al., 2007). In Shannon and Weaver's (1949) information theory, the term information has come to mean something which reduces uncertainty at the receiver. Additionally Rouse and Rouse (1984) emphasize the relevance of information. Thus, in order to be valuable, information involves relevance and reduction of uncertainty. This implies that the value of information is subjective to the one seeking the information (Rouse and Rouse, 1984; Zunde, 1978).

Although good information of the appropriate type and quality cannot ensure that the correct decision will be made, its availability increases the probability of better decisions (Davis, 2005). Therefore it is crucial that organizations systematically acquire the environmental information that is needed for decision making (Shim et al., 2002). Due to contemporary economic conditions and developments (e.g. globalization, compliance, IT-developments) managers must wade through an ever-increasing amount of information that is often confusing or conflicting.This suggests that research must gain a better understanding of the process of obtaining and using information (Olszak and Ziemba, 2006).

Information-seeking behaviour are the patterns of behaviour that individuals employ when they (1) perceive information needs, and (2) retrieve the information (Wilson, 1999). Choo (2006) suggests that an individual executes the patterns because he or she requires information to reach the desired end state – and a gap in is recognized in this information need. Once the information needs have been identified, the decision maker purposefully searches for information that can change that state of understanding (i.e. information seeking). Information seeking is the process of identifying and choosing among alternative information sources and contains a procurement

Pg. 22

BI systems to mitigate information load with store managers

decision, in which the decision maker decides whether to use information or not (Grünig and Kühn, 2005; Rouse and Rouse, 1984). It is a dynamic process because the methods and criteria whereby information is selected or rejected often vary in time and depend on intermediate results (Rouse and Rouse, 1984).

The more that is invested in information procurement, the greater the probability that good options will be found and improve decision-quality. Together, information seeking and processing (see chapter 0) is part of the larger human and social activity through which information becomes useful to the decision maker. Consequently, the processes of information seeking and use are contingent upon the changing conditions of the individual‟s context of information use. Therefore information search and use are dynamic processes that often appear disorderly (Davis, 2005).

4.4

Computerized decision support

Various information systems (IS‟s) have been developed in order to support managerial decisionmaking (Martinsons and Davison, 2007). Successfully supporting managerial decision-making is critically dependent upon the availability of integrated, high quality information that is organized and presented in a timely and easily understood manner (March and Hevner, 2007). Computerized decision support methodology recognizes this need, and intends to provide the manager with information, but also to overcome a manager‟s cognitive limits by quickly accessing and processing vast amounts of information (Turban et al., 2007). The effect of IS‟s is to replace or reduce the effort expended in information acquisition, management, and processing. From this perspective a well-designed decision supporting system (DSS) should reduce the effort required to follow one or more decision strategies. The decision maker then has the option of (1) investing the effort saved into finding a better quality solution, or (2) making the decision at a lower level of cognitive cost (Todd and Benbasat, 1992).

Pg. 23

MSc thesis M.J. van Strien

Support in managerial decision-making

Whereas decision makers are selective in their decision style, different styles require different types of support. For a computerized system to successfully support the decision maker, it should fit the decision style, and hence should be adaptable to different users (Turban et al., 2007). The fit determines the degree in which the decision maker incorporates the aid‟s recommendations into the decision process (Mascha and Smedley, 2007). Davis, Bagozzi and Warshaw (1989) find heterogeneity in the extent that decision makers rely on a DSS‟ recommendations.

Hong and Vogel (1991) argue that the decision maker should decide which search strategies or decision processes to use, possibly with the help of DSS‟s. Contrasting, , Cook (1993) argues that the organization initiating the DSS should specify the approach to be used. He notes that this is especially the case in decisions that require a certain extent of consistency over various decision makers (e.g. capital budgeting, or capital loan decisions). The author indicates that these decisions require a DSS that provides some structure to encourage desirable search strategies or decisions processes (i.e. decisions channelling), or discourage undesirable decision processes (i.e. decision restrictiveness), while the individual decision maker may be more interested in search strategies that reduce the cognitive effort – but still lead to a decisions that is acceptable (Todd and Benbasat, 1991).

Pg. 24

BI systems to mitigate information load with store managers

5. Business Intelligence as decision support systems From the previous chapter we conclude that information technology (IT) can support decisionmakers by providing valuable information, and reducing the decision-maker‟s cognitive effort. More than ever, decision-making and managing an organization requires information (Olszak and Ziemba, 2006; Gangadharan and Swami, 2004). On the one hand, organisations that are interested in improving quality of decision making should incline towards the development of IT infrastructure that will represent a holistic approach to business; e.g. operations, customers, suppliers. While on the other hand dispersion of information assets results in inefficiency in the current information management models. Business Intelligence (BI) meets the requirements that arise from this situation(Olszak and Ziemba, 2006).

5.1

Business Intelligence

According to Turban, Aronson et al. (2007), BI‟s major objective is to provide computerized decision aid by enabling interactive access to data from various sources. The objective is enabled through BI architecture (i.e. data warehousing) and tools that enable the manipulation of data to conduct interactive analysis (i.e. business analytics). The manipulation process is based on the transformation of data into information. This implies that BI is a data-driven decision support system (DSS) which combines data gathering, data storage, and knowledge management in order to provide input to the decision process (Golfarelli, Rizzi and Cella, 2004).

BI is the result of a series of technological innovations(Negash and Gray, 2008; Shim et al., 2002), and organizational developments (e.g. metric-driven management and a more holistic view on decision making) have advanced BI technologies and use (March and Hevner, 2007; Golfarelli et al., 2004). Golfarelli et al. (2004; p.1) note that (academic) research has managed to transform “a bundle of native techniques” into a well-founded approach to information extraction and processing. This has significantly altered how organizations apply IT in order to provide managerial decision aid.

Pg. 25

MSc thesis M.J. van Strien

5.1.1

Business Intelligence as decision support systems

Defining BI

In the vast amount of BI literature many authors have established a definition for BI. There however is a lack of a clear academic definition as most definition are extracted from popular (scientific) literature. In order to gain an abstract overview of the literature that is available, some of the most adopted definitions of BI are provided in Table 5.1. Table 5.1 Definitions of BI

Definition

Source

BI involves collecting, managing, mining and analyzing the data generated by an enterprise,

(Thierauf, 2001)

resulting in information and knowledge with strategic value for those who possess it BI refers to a managerial philosophy and a tool used to help organizations manage and

(Lönnqvist

and

refine business information with the objective of making more effective decisions

Pirttimäki, 2006)

BI refers to inferences and knowledge discovered by applying algorithmic analysis to

(March and Hevner,

acquired information – in order to aid the purposeful execution of business processes.

2007)

BI are the systems that combine data gathering data storage with analysis to evaluate

(Negash and Gray,

complex corporate and competitive information for presentation to planners and decision

2008)

maker, with the objective of improving the timeliness and the quality of the input to the decision process

The literature study shows a discussion on the scope of BI; some argue that BI is primarily decision making, while others apply a broader business performance management perspective. Since the aim of this research on decision making, BI is adopted as a decision supporting paradigm (i.e. management philosophy and tool set (Gilad and Gilad, 1986)) for this research. Also, following recent trends in BI (Azvine, Cui, Nauck and Majeed, 2006; Watson, Wixom, Hoffer, Anderson-Lehman and Reynolds, 2006; Azvine, Cui and Nauck, 2005) this research applies a more real-time perspective which enables decision support for tactical and technical organizational levels (Golfarelli et al., 2004). Consequently BI is defined as: “the systems that combine data gathering and data storage with information analysis, which allows decision makers to integrate (i.e. select, aggregate and present) the relevant information into a specific decision task”.

Pg. 26

BI systems to mitigate information load with store managers

Based on a literature review and the definitions provided in table Table 5.1, two major aspects in the BI approach are distinguished; (1) BI architecture (also referred to as data warehousing), and (2) BI tooling. BI architecture is capable of collecting and integrating data that mirrors the organizational environment. Additionally BI tooling (i.e. the user-interface) allows managers to integrate this environmental data into their decision-making processes. Hence the total of BI systems allows users to be better informed about their decision environment. 5.1.2

Data warehouse

Due to the increasing rate of data generation, and gathering, it has become complex to maintain and use the data and information. To keep up with these advances, many organizations need to create data warehouses (Turban et al., 2007). Conceptually a data warehouse serves as an integrated repository of current and historical data which are of potential interest to managers throughout the organization (Moss and Atre, 2003) The data are extracted from different operational IS‟s (e.g. online transaction processing systems, batch systems) and acquired from external sources (March and Hevner, 2007; Turban et al., 2007).

Because of this integration, data warehouses contain a wealth of data which is usually structured in order to be available immediately for analytical processing (i.e. BI tooling). Turban and Aronson (2007) therefore argue that a data warehouse is produced to support decision making. This purpose can be achieved through the acquisition, integration transformation, of internal and external data (March and Hevner, 2007; Moss and Atre, 2003). To provide decision support, a a data warehouse threoretically is a subject-oriented, integrated, time-variant, non-volatile collection of data from both internal and external data sources (Inmon, 2009; Turban et al., 2007). These characteristics are addressed in Table 5.2. Table 5.2 Data warehouse characteristics (Turban, Aronson et al. 2007)

Characteristic

Meaning

Subject oriented

Data are organized by detailed subject (e.g. sales, customers, or marketing) containing only information relevant for decision aid. Subject orientation enables managers to not only determine how their business is performing, but also why

Pg. 27

MSc thesis M.J. van Strien

Integrated

Business Intelligence as decision support systems

A data warehouse is presumed to be totally integrated. In order to do so data warehousing must place data from various sources into a consistent format. To do so, they have to overcome several syntactic and semantic conflicts and discrepancies.

Time variant

Data warehouses detect trends, deviations, and long-term relationships. Hence time is an important dimension that all data warehouses must support. Real-time data warehouses provide current and historical data.

Non-volatile

Data in the data warehouse cannot be changed or updated. This enables the data warehouse to perform exclusively for data access.

Additionally, a data warehouse architecture may contain one on more data marts (see Figure 5.1). Whereas a data warehouse combines databases across an entire organization, a data mart is usually smaller and focuses on a particular subject or department (e.g. marketing, operations, or customers). A data mart is considered a subset of a data warehouse which can either be dependent or independent on the data warehouse (Turban et al., 2007).

Figure 5.1 General data warehouse architecture (Turban et al., 2007)

Data warehousing Turban and Aronson (2007) differentiate a data warehouse from data warehousing, which literally is the entire process that facilitates data warehouses. This process consists of the development, management, operational methods, and practices that define how these data are collected, integrated, interpreted, managed and used (March and Hevner, 2007). The activities for data warehousing are categorized in three processes: extract, transform, and load (ETL). The data Pg. 28

BI systems to mitigate information load with store managers

are extracted from various data sources, then the data is transformed (e.g. cleansed, and customized), and loaded into the data warehouse (Vassiliadis, Simitsis and Skiadopoulos, 2002). The ETL functions in a data warehouse are considered the most time-consuming portion of the development lifecycle. Often operational systems are not designed to be integrated and data extracts must be performed manually or on a schedule determined by the operational systems. Also, data acquired from external sources is rarely in a form conducive to integration (March and Hevner, 2007).

5.1.3

BI Tooling

BI tooling intends “to provide decision makers at various levels in organisations with timely, relevant and easy-to-use information”. These tools are generally used for business analysis (BA), which is “the ability to analyse business information in order to support and improve management decision making across a broad range of business activities” (Elbashir, Collier and Davern, 2008; p.135, 136). BA involves tools like online analytical processing (OLAP), data mining, data visualization, and multidimensional presentation (Turban et al., 2007).

Different from traditional, static computerized decision support, the new breed of BI tools is malleable. This allows the decision maker to present exactly the information that is relevant for the decision context, and in the way that is best for the decision-maker. Traditional computerized decisions support is considered rather static since the data presentation and analysis are predetermined by the designer of the tools. BI tools allow decision makers to work with vast amounts of data (stored in the data warehouse) and give them the ability to modify how and which data are presented. Also they enable decision makers to select the methods used to analyse the data (Tremblay et al., 2007). BI decision-support initiatives also call for new technology to be considered, additional tasks to be performed, roles and responsibilities to be shifted, and analysis and decision-support applications be delivered quickly while maintaining acceptable quality (Moss and Atre, 2003).

Pg. 29

MSc thesis M.J. van Strien

Business Intelligence as decision support systems

Concluding, through interactive interfaces, BI tools allow decision makers to perform dynamic queries on a large (enterprise-wide) data repository. We argue that BI tools facilitate a more interactive information acquisition, which enables the decision maker to present the information relevant to the decision task, and his or her own information processing preferences. Hence, BI systems theoretically support decision makers in acquiring the information they need to make the decision. By allowing interactive access through BI tools, BI facilitates an effort reduction in the acquisition process. The next paragraph will address information acquisition from a BI perspective.

5.1.4

Real time Business Intelligence

As shown above, BI allows decision makers to analyse data in order to perform BA; e.g. predict market trends of products and services. However, traditional BI systems are generally designed to periodically gather data, hence restricting the users in analysing the present situation (Turban et al., 2007). Azvine et al. (2006) argue that it is becoming clear that current competitive conditions require such data analysis to be carried out in real-time. With ever-increasing competition and rapidly changing customer needs and technologies, enterprise decision makers are no longer satisfied with scheduled analytics reports, pre-configured key performance indicators or fixed dashboards. Golfarelli et al. (2004) emphasize that technological advances allow BI systems to collect, store and disseminate real-time. This allows decision makers to perform actions in response to the analysis results in real-time.

Real-time BI requires the data warehouse to gather data in real-time, rather than to create periodical snapshots of the business environment. Despite the benefits of real-time BI, developing a real-time data warehouse can create issues as for example architecture, database design storage, scalability and maintainability (Turban et al., 2007). Golfarelli and Rizzi (2004) argue that a realtime approach to BI requires data warehouses to be constantly fed and made available at the right time, at the proper decision, and in the best form. The authors claim that this approach distinguishes from the „traditional‟ periodic data warehousing approach in users, delivery time, information coarseness, and user interface. Pg. 30

BI systems to mitigate information load with store managers

5.2

Decision effort

Decision making in management has always involved utilization of different information assets (Olszak and Ziemba, 2006). But when are managers considered to be better informed, and what is the impact regarding decision quality? According to cost-benefit theory, which has been applied extensively to choice tasks, decision makers make a trade-off between the effort to make a decision (Todd and Benbasat, 2000; Stigler, 1961) and the expected quality of the decision outcome (Vessey, 1994). Cost-benefit theory is used to theoretically explain how BI systems affect how decision makers acquire information.

Todd and Benbasat (2000; p.92) have examined the moderating effect of effort in the relationship between decision aid usage and decision quality. In their research, the authors find that “in order to induce the use of a superior decision strategy and, as a consequence, improve decision quality, a decision aid must make that superior strategy at least as easy to employ as any simpler but less accurate heuristic”, otherwise, a decision aid may only influence decision-making efficiency. Hence, decision aid should facilitate effort reduction to increase information processing performance and induce a more accurate decision making strategy (Paquette and Kida, 1988). So, from a cost-benefit perspective, being better informed means that decision aid allows decision makers to employ a more accurate decision strategy; i.e. process more relevant information into their decision.

Information processing performance is contingent upon various characteristics of the decision task (e.g.. information load (Schroder et al., 1967), individual processing abilities (Henry, 1980; Newell and Simon, 1972), and task complexity (Payne, 1976)). In order to allow an accurate strategy, the information should fit the task demands. Task demands are plans, algorithms, or systematic procedures used to integrate the information needed to make a decision. A demand can be exogenous (i.e., a company operating procedure) or endogenous (i.e., habit or experience in solving similar tasks) to the decision maker (Jarvenpaa, 1989).

Pg. 31

MSc thesis M.J. van Strien

Business Intelligence as decision support systems

Additional to the effort reduction by use of DSS‟s, BI tools provide decision makers with a dynamic information environment which allows decision makers to customize the selection, aggregation and presentation of information (Tremblay et al., 2007), theoretically enabling decision makers to adapt information to fit into the required task. March and Hevner (2007) refer to this as contextualization of the information. Theoretically, contextualization should facilitate decision makers to reduce effort, which can either lead to (1) increased decision making efficiency, or (2) increased information processing, and decision effectiveness.

Pg. 32

BI systems to mitigate information load with store managers

6. Information processing and overload Decision makers are often required to evaluate and integrate several information cues simultaneously. Studies have suggested that the information processing capacity of human decision makers is limited and that problems may arise when that limit is reached (Newell and Simon, 1972; Miller, 1956). Henry (1980) underpins the importance of individual‟s ability to assimilate, retain, and integrate information in order to make (complex) judgments in a decision making task. This is referred to as an individual‟s information processing ability or performance. This chapter will address the effect of increasing information load on an individual‟s information processing ability. Drawing on information processing theories it addresses how decision quality theoretically changes when information load is affected. For purposes of scope, we emphasize high information load. Based on Schroder et al.‟s (1967) human information processing model (see paragraph 6.2) a relationship between information load and an individual‟s level of information processing is theorized. This suggests that reduced information processing strategies tend to ignore much of the (relevant) information available to a decision maker, implying reduced decision quality (Paquette and Kida, 1988). Consequently the relationship between an individual‟s information processing ability and the decision-making quality is addressed.

6.1

Information processing

The descriptive research approach to individual information processing is concerned with understanding how people process information and make decisions (Driver and Mock, 1975). Within the descriptive approach, there are multiple views on how humans process information. The generalists adopt a simple view, and posit that all humans are more or less similar in processing information (Starbuck, 1983). In contrast, the unique school (e.g. Newell and Simon, 1972) suggests that each individual processes information uniquely. “The basic problem with the generalist view is pragmatic; “whereas generalization may be sufficient for science, it often fails

Pg. 33

MSc thesis M.J. van Strien

Information processing and overload

in organizations” (Driver and Mock, 1975; p. 494). Additionally, Henry (1980; p.42) claims that it is “widely accepted” that individual‟s information processing ability differs. Hence, the present research adopts the “unique” view on information processing. This paragraph will focus on how individuals process information, while the subsequent paragraph will address how information load affects an individual‟s processing ability.

6.1.1

Individual differences

The information-processing approach taken by Miller (1956) had two components; (1) the general concept of human as information-processing system, and (2) a specific mathematical theory of information that allowed for accurate measurement of the capacity of the informationprocessing system. With this approach Miller proposed a concept of the individual as an information-processing channel with limited and measurable capacity. This concept has been enormously influential in developing the field that became known as cognitive psychology (Baddeley, 1994), and acknowledges that information processing ability is dependent upon, amongst others, an individual‟s processing abilities (Henry, 1980; Newell and Simon, 1972),

Besides differences in processing abilities, individuals also use different processing strategies (Newell and Simon, 1972). Consequently, a considerable amount of research on decision making has investigated the underlying processing strategies that decision makers employ in choice contexts (e.g. Payne et al., 1993; Paquette and Kida, 1988; Svenson, 1979; Payne, 1976). Paquette & Kida (1988) view decision-makers as problem solvers who have the ability to choose from many coexisting information processing strategies. Hence the authors consider decision behaviour flexible and adaptable, meaning that different decision strategies can be used by the same decision maker, depending upon the properties of the task.

Pg. 34

BI systems to mitigate information load with store managers

6.1.2

Processing strategies

Paquette & Kida (1988) identified a number of information processing strategies such as additive compensatory, additive difference, and elimination by aspects. Some of these strategies are cognitively complex, requiring the decision maker to consider large amounts of data. Others are reduced processing strategies, which require a limited information search and simpler evaluation processes. Contingent utilization of heuristic strategies will keep the information processing demands of the situation within the bounds of their limited capacity (Paquette and Kida, 1988). The use of heuristics in decision making is systematically related to certain characteristics of the decision situation (i.e. the demands of the task), and their own preferences (Payne, 1976). Research suggests that decision makers rely on simplifying information search strategies as information load increases. This may lead to the use of heuristics for evaluating and combining information (Svenson, 1979).

6.2

Information processing ability

Drawing from Miller‟s (1956) work on cognitive processing limitations, Schroder et al. (1967) proposed a nonlinear relationship between the information load and an individual‟s information processing ability. The relationship between information load and information processing is often depicted as an inverted U curve (Paul and Nazareth, 2010; Hwang and Lin, 1999). As shown in Figure 6.1, information processing increases when the decision maker is experiencing information underload, and information processing decreases when he or she is experiencing overload. This model predicts that increases in information load will increase information processing initially. However, if information load keeps increasing – till it exceeds the individual’s capacity – decision maker‟s information processing will decrease.

Research has shown that increases in information load affects human information processing (Payne, 1976). Since decision quality is a result of human information processing ability, this suggests that information load may have an effect on decision quality (Iselin, 1988). Because of this moderating effect on decision quality – decision makers should be interested in the Pg. 35

MSc thesis M.J. van Strien

Information processing and overload

information load. The effects of information load on decision quality may vary with context (Schroder et al., 1967), This paragraph will discuss theoretical exploration of information load in relation to an individual‟s information processing ability. In order to do so, the approach of individuals as information processing systems is applied. This systematic approach involves an individual‟s information processing capacities (IPC‟s), and information processing requirements (IPR‟s).

Figure 6.1 Relationship between information processing ability and information load

6.2.1

Information processing requirement

IPR is the load of information that a decision maker is required to integrate into the decision task (Eppler and Mengis, 2004). In more traditional research (e.g. Streufert, 1973) information load has commonly been assessed in terms of number of information inputs or cues provided to a decision maker (Iselin, 1988). This approach is too narrow, and consequently information load should be approached more broadly. Various researchers (e.g. Paul and Nazareth, 2010; Hwang and Lin, 1999; Iselin, 1988) have identified two major sub-dimensions for information load: (1)

Pg. 36

BI systems to mitigate information load with store managers

information complexity, and (2) information novelty. Complexity refers to the number of different elements or features of information, which can be the result of increased information diversity. Increased information complexity significantly decreases the accuracy in information processing (Paul and Nazareth, 2010; Henry, 1980). Novelty involves the unexpected, surprising, new, or unfamiliar aspects of the information element (Huang, 2000). An increase in information load will also lead to increased IPR. Additional to information load, research has shown that task and process characteristics (e.g. non-routine tasks, task interdependencies, task interruptions, and time pressure) also affect IPR (Paul and Nazareth, 2010).

6.2.2

Human processing capacity

Miller (1956) conjectured that there is an upper limit on our capacity to process information on simultaneously interacting elements. This limit is seven plus or minus two elements. Baddeley (Baddeley, 1994) argues that Miller‟s notion has been influential for future research, and various researchers (e.g. Saaty and Ozdemir, 2003; Baddeley, 1994) have proven the validity of this number after all these years. This concludes that human IPC is limited. Henry (1980) finds that information processing ability is correlated to an individual‟s IPC and that individuals with different IPC‟s indicate a wide variety in processing ability. Schroder et al‟s (1967) model predicts that the level of information processing is the maximum when the information load is moderate; i.e. nether too high or too low. This is referred to as the “optimal point” of information load (Paul and Nazareth, 2010). As mentioned above, individuals have different IPC‟s which reflect different optimum input ranges. Based on the idea of different optimal input levels, Driver and Steufert (1969) reason why a curvilinear relationship is not always observed between system input and output. Furthermore, the authors argue, as input moves away from the optimum, each system responds with a characteristic temporary loss in integrative complexity.

Pg. 37

MSc thesis M.J. van Strien

6.3

Information processing and overload

Information overload

In ordinary language the term information overload is often used to address the notion of receiving too much information. In research literature however, the term has resulted in various constructs, synonyms, and related terms; e.g. cognitive overload, sensory overload, communication overload, knowledge overload, and information fatigue syndrome, but all refer to the situation where IPR exceeds IPC (Eppler and Mengis, 2004). Information overload does not simply occur because of a decision maker‟s limited capacity (i.e. short-term memory) to process and manipulate an excessive amount of data. Information processing theory shows that information overload rather is a function of differences in individual‟s IPC (Rutkowski and Saunders, 2010). Eppler and Mengis (2004) find that research has defined a wide variety of definitions for information overload, see Table 6.1. The authors however conclude that, although literary different, all definitions state that a decision maker will experience information overload when the IPR exceed the individual‟s IPC (Eppler and Mengis, 2004). This paragraph will briefly address the causes, the effect on decision quality, and solutions for information overload. Table 6.1 Definitions of information overload (Eppler and Mengis, 2004)

Definition

Reference

The decision maker is considered to have experienced information overload at the (Chewning

and

point where the amount of information actually integrated into the decision begins Harell, to decline. Beyond this point, the individual‟s decisions reflect a lesser utilization of

Schroder

the available information.

1967)

1990; et

al.,

Information overload occurs when the volume of the information supply exceeds the (Jacoby, 1984) limited human information processing capacity. Dysfunctional effects such as stress and confusion are the result. Information overload occurs when the information-processing requirements

(Tushman

and

(information needed to complete a task) exceed the information-processing capacity Nadler, 1978) (the quantity of information one can integrate into the decision-making process). Information overload occurs when the information-processing requirements exceed (Keller and Staelin, the information-processing capacity. Not only is the amount of information 1987)

Pg. 38

BI systems to mitigate information load with store managers

(quantitative aspect) that has to be integrated crucial but also the characteristics (qualitative aspect) of information Information overload occurs when the decision maker estimates he or she has to

(Iselin, 1993)

handle more information than he or she can efficiently use.

The complex problem of information overload has grown with advances in technological capabilities. The authors state that each round of technological advances (e.g. artificial intelligence, processing capacities, connectivity) has advanced our ability to collect, transmit and transform data, leading to production of a huge amount of data. However, the authors claim, our ability to extract a meaning (i.e. to interpret and transform to information) from this unprecedented pile of data has advanced much more slowly (Woods et al., 2002). Information access has increased to the point that it has raised concerns that we now suffer from a harmful condition of information overload (Himma, 2007).

6.3.1

Causes

The causes of information overload, are primarily taken from Eppler and Mengis‟ (2004) literature study. The authors have distinguished five categories for information overload causes: (1) personal characteristics, (2) information characteristics, (3) task and process parameters, (4) organizational design, and (5) information technology. The authors argue that information overload is not caused by one of these factors alone, but rather through a mix of all as all factors may affect IPC and IPR. Based on their literature review, the authors present a list of specific causes of information overload. This list – divided into the categories – is presented in Table 6.2.

Table 6.2 Causes of information overload (Epler and Mengis 2004)

Category Personal factors

Information characteristics

Cause Limitations in individual IPC Decision scope and resulting documentation needs Motivation, attitude, satisfaction Personal traits (experience, skills, idealogy, age) Personal situation Number of information rises

Pg. 39

MSc thesis M.J. van Strien

Task and process parameters

Organizational design

IT

Information processing and overload

Uncertainty of information Diversity of information and number of alternatives increase Ambiguity of information Novelty of information Complexity of information Intensity of information Dimensions of information increase Information quality Overabundance of irrelevance information Less routine tasks Complexity of tasks and task interdependencies Time pressure Task interruptions Too many, too detailed standard procedures Simultaneous input of information into the process Innovations evolve rapidly Interdisciplinary work Collaborative work Centralization or disintermediation Accumulation of information Group heterogeneity New information and communication technologies Push systems E-mails Intranet, extranet, internet Rise in number Various distribution channels and dispersed information sources Vast storage capacity of the systems

Figure 6.2 depicts how the categories of information overload causes – except organizational design - can be mapped with the independent variables of the present research. This suggests that organizational design is missing in the conceptual model.

Pg. 40

BI systems to mitigate information load with store managers

Figure 6.2 Mapping of information overload causes to the variables

6.3.2

Decision quality

Hwang and Lin (1999) have suggested an inverted U-curve to describe the relationship between information load and decision quality. This implies that information processing and decision quality correlate positively with the amount of information that the individual receives. This is however up to a certain point. If information load keeps increasing and finally exceeds the IPC of the decision maker, information processing will cease being increased. Instead, the decision maker will decrease information processing as they experience a phenomenon termed information overload (Hwang and Lin, 1999). If any information is received beyond this point, the decision maker will not be able to integrate this information into the decision processes, and decision quality will rapidly decline (Eppler and Mengis, 2004; Chewning and Harell, 1990). The occurrence of information overload will disable the decision maker to exercise executive control over the information processing system (Hahn et al., 1992). Concerning the quality of the decision, this will confuse the decision maker, make prior information harder to recall, and affect the ability to set priorities - resulting in rapid declining of the decision quality (Eppler and Mengis, 2004; Edmunds and Morris, 2000; Schick and Gordon Susan, 1990). Additionally, Payne (1976) finds that information overload increases the decision maker‟s confidence.

Pg. 41

MSc thesis M.J. van Strien

6.3.3

Information processing and overload

Mitigation

Various researchers address specific countermeasures to alleviate the effects of information overload in organizational decision environments (Eppler and Mengis, 2004). From their literature research, the authors present specific countermeasures related to the specific causes of information overload (see Table 6.2)The solutions that have been proposed in previous research can be divided into managerial and technical solutions (Bawden, Holtham and Courtney, 1999).

However, Brueggemann et al. (2008), apply a more abstract approach in order to address solutions to the information overload problem. The authors call for a more context-based approach to the design of computerized decision support. This approach encompasses incorporation of the decision maker‟s and the decision-task‟s context (e.g. nature of the task, goals, roles, and experience) into the presentation of the information. The authors evidence a positive effect on a decision maker‟s capacity to handle information more effectively and efficiently when it is presented in the context of the decision maker and task demands. This aligns with Woods et al.’s (2002) notion that increasing information pertinence (i.e. contextual) allows individuals to better process the information.

Pg. 42

BI systems to mitigate information load with store managers

7. Analysis This chapter will present the analysis of the evidence that have been collected from the interviews, questionnaires, and observations. We divide the analysis into two types of subjects; i.e. store managers as being the decision makers, and system designers as being responsible for the information systems that support store managers. Results of these distinct analyses will be combined in the conclusion. For matters of understanding we first present a brief case description. Thereafter we will present the results of the analysis.

7.1

Case description

Store managers of C1000 supermarkets have a relative autonomy to be entrepreneurial and make their own decisions2. They are part of the C1000 franchise, and therefore are serviced by the C1000 Retail chain (formerly known as Schuitema Retail) and bound to some of C1000 Retail‟s central agreements. C1000's Retail's headquarter is responsible for supporting approximately 370 stores in the Netherlands, this is done in various aspects like procurement, store management, marketing, and Information Technology (IT).

Store managers' relative autonomy means that they are able to run their own store, and find the right balance between entrepreneurship and advantages of synergies in the franchise formula. The playfield of decision making autonomy for C1000 store managers varies from tactical to operational decisions. This allows them to shape their own store and store management by deciding upon (1) local (price) promotions next to the national brochure, (2) manage their own inventory by being fully responsible for ordering decisions, (3) their store lay-out, (4) their (human) resource management, and (5) resource planning. However, C1000 Retail offers the 2 Last year, C1000 Retail has started a project (STORE) that intends to transform C1000 from a wholesale to a retail organization, meaning that store's are more and more serviced from a centralized perspective. This also implies a shift from decentralized decision making (e.g. ordering decisions) towards more centrally managed supermarkets (e.g. vendor-managed inventory). However we acknowledge that this shift might affect a store manager's autonomy and exposure to information overload, we did not incorporate this development into our research, hence data collection

Pg. 43

MSc thesis M.J. van Strien

Analysis

store managers several methods to support them in taking full advantage of the franchise formula. Amongst these offerings are an advisor for store and shelving lay-out, logistics, marketing, finance and logistics, and an information system that advises store managers in their ordering decisions.

Concerning the IT, retail chains are known to be impacted by IT developments, as technological advances enable them to be more efficient and effective in forecasting, procurement, and customer service (e.g. self-scanning services). This important role is acknowledged both by C1000 Retail, and the store managers we interviewed. Some even agreed that the application of IT is one of the major competitive advantages in efficiency and customer service. Because of this, the information systems (IS‟s) have become an important part in a store manager‟s job. Through the years, the IS landscape (e.g. financial, digital invoices, C1000net/intranet, resource planning, and order management systems) has exploded; not necessarily the amount of systems, but certainly the amount of functionalities. The interviews have showed that store managers need to cope with a lot of information and documents for internal management, but also increasingly deal with external factors (e.g. regulations, competitors, customers, and news). The majority of this information can be described as qualitative (e.g. messages and brochures) and semi-structured, rather than quantitative (e.g. financial and sales data). To properly deal with this information, the managers have established an organizational design, which divides various activities and decision responsibilities. All interviewed store managers had set up different divisions (e.g. fruit and vegetables, sustainable products, bakery, and butchery) that were made responsible for different sections in the store. The division head is responsible for staffing and inventory management on the short term.

7.2

Store managers

This paragraph will discuss the data that have been collected from the interviews with seven store managers. The interview protocol (see Appendix III) aimed at collecting contextual data first,

Pg. 44

BI systems to mitigate information load with store managers

focussing on decision maker characteristics, decision making processes and the decision environment. Thereafter the semi-structured interview addressed measurement of the variables. Table 7.1 shows how the questions are mapped with the variables. Contextual evidence was collected qualitatively, while evidence for the variables was collected both quantitative and qualitative.

The evidence was analysed using descriptive statistics and content analysis (see Appendix V and VI). The categorization that was applied for the content analysis is presented in Table 7.1. For the categories that have been determined a priori, we use both quantitative and qualitative data for the analysis.

Table 7.1 Categorization for the content analysis

Category

Subcategory

Measurement

Development

Question

Qualitative, quantitative

A priori

4

Domain knowledge

Qualitative, quantitative

A priori

3, 4

Technology use

Qualitative

Emergent

Trust

Qualitative

Emergent

Qualitative, quantitative

A priori

Interruptance

Qualitative

Emergent

Structuredness

Qualitative, quantitative

A priori

8, 9, 11

Qualitative, quantitative

A priori

12, 13

Complexity

Qualitative, quantitative

A priori

14

Novelty

Qualitative, quantitative

A priori

14

Pertinence

Qualitative

Emergent

14

Reliability

Qualitative, quantitative

A priori

14

Relevance

Qualitative

Emergent

14, 15

Decision responsibility

Qualitative

Emergent

1

Information load (moderating)

Qualitative, quantitative

A priori

14, 16

Decision quality (dependent)

Qualitative, quantitative

A priori

17, 18, 24

Inventory performance (dependent)

Qualitative, quantitative

A priori

19, 20, 21, 22,

Experience

Time Pressure

Information Amount

6, 7

23

Pg. 45

MSc thesis M.J. van Strien

Analysis

To allow triangualitation we first address descriptive statistics, and thereafter check whether there is congruence with the content analysis. The independent, dependent and mediating variables are analysed separately. These analyses will come together in the conclusion.

7.2.1 Independent variables When analyzing the descriptive statistics (see Table 7.2, 7.3, and 7.4), the first thing that draws our attention is the small variability for the independent variables, this is the case for both the quantitative and qualitative evidence. Only perceived time pressure, time availability and external steering are found to have a relative high variability. This shows that dependent variables are more or less the same at all subjects. This disables us to explain what variables affect information load, but allows us to clearly describe and explain the case. Moreover the descriptive statistics show that the subjects (1) feel very autonomous, but perceive steering effort from C1000 Retail, (2) are very experienced, (3) have time pressure when making decisions, (4) find the decisions structured and not complex, and (5) the perceived information amount is fairly low.

Table 7.2 Descriptive statistics for experience and time pressure

Autonomy

Experience

Time pressure

Information Novelty

Goal

Subjective

Subjective

Time avail.

Objective

Subjective

Steering

1=fully disagree 5=fully agree

Autonomy

Complexity

Median

5

4

5

5

2

2

3

4

4

Mode

5

4

5

5

2

2

3

4

4

Minimum

5

2

4

4

2

2

2

4

2

Maximum

5

4

5

5

4

5

3

4

4

Count

7

7

7

7

7

7

7

7

7

Experience Pg. 46

BI systems to mitigate information load with store managers

Overall we find that experience and domain knowledge help the store manager to gather and filter the relevant information and play a very important role in not overloaded b information. Additionally, experience is found to be a precondition to get acquainted, and cope with the IS‟s.

The content analysis shows that all subjects are moderate to very experienced. This is not necessarily for managing the store, but all managers have broad domain knowledge. “De zaak is van mijn vader geweest, dus ik ben er eigenlijk ingerold" (s4) “Ik ben pas sinds dit jaar ondernemer. Ik ben begonnen als stagiair en heb hiervoor wel 8 jaar in de supermarkt gewerkt”(s5) “Ik ben begonnen in de supermarkt als bijbaantje en ben er zo verder ingerold. Ik heb nu ongeveer 25 jaar ervaring met het werken in een supermarkt” (s6) "Ik werk al vanaf mijn achttiende in de supermarkt. Ik ben begonnen als vakkenvuller en daar is mijn interesse door gegroeid”(s7)

A familiar scenario is that they started working stocking shelves, and advanced to store manager. This obviously gives the subjects a lot of domain knowledge, and explains why the subjects are experienced. All the managers and designers have addressed experience as one of the most important factors that prevents a store manager from being overloaded with information. “Het ligt niet aan C1000 dat ik alle informatie goed weet te delegeren, dat is een stukje ervaring dat je opdoet” (s1) “Alleen door mijn ervaring weet ik informatie goed te filteren. Dan kan ik goed bepalen of ik het moeten weten of dat dit naar iemand anders moet” (s4) “Mensen kunnen dit alleen behappen door hun ervaring” (designer2)

Moreover, the content analysis shows that technology use and trust affect the relationship between experience and information load. Because IS‟s have become so important in their job, the data shows that store managers need to be committed to using technology to be able to efficiently gather and process the information. “De systemen zijn toch wel pittig. Je moet er mee om leren gaan. Je ziet nu nog vaak dat ze blijven bestellen op de manier waarop ze deden” (s2) “Als je niet begaan bent met techniek, wordt het lastig” (s1)

Pg. 47

MSc thesis M.J. van Strien

Analysis

Additionally trust (based on prior experiences) is important to utilize the computing effort decreasing advantages of IS‟s. The interviews show that the majority of subjects needed time to get acquainted with IS‟s, and all subjects indicate that they now feel acquainted with the systems. “Je moet er mee om leren gaan. Je ziet nu nog vaak dat ze blijven bestellen op de manier waarop ze deden” (s3) “Ik heb het vertrouwen in de systemen ontwikkeld omdat ik er zeker de meerwaarde van inzie” (s4) “Ik heb in het begin slapeloze nachten gehad omdat ik niet aan die systemen kon wennen. Ik vertrouwde er nog niet op. Inmiddels heb ik dat vertrouwen wel” (s4)

“Het bestelsysteem is heel duidelijk. Ik maak 95% van de tijd gebruik van de adviezen” (s7)

The observation however shows that some younger store managers operate the IS‟s more intuitively. Moreover, information quality and contradicting information seem to be important in decreasing trust in IS‟s

Time Pressure Although the decision activities are becoming more complex because of regulations, customers and competitors, all subjects agree that the decision process is structured. The goal of the decision (especially for the ordering decision) is always known, and the information queues that are used to make the decision are often the same. Although the subjects agree on decision complexity and structuredness, the quantitative evidence shows a variability in perceived time pressure. “We hebben nu 20 uur om te bestellen, met ons systeem is dat geen problem”(s1) “Het is wel eens gebeurd dat ik ergens geen tijd voor had, terwijl ik er dieper in moest duiken”(s2) “Er is helemaal geen tijdsdruk bij het maken van bestelbeslissingen. Ik heb meestal meer dan een dag om de beslissing te maken, het is een kwestie van je zaakjes goed op orde hebben”(s4) “Het is vaak lastig om alle informatie goed op orde te krijgen voordat we daadwerkelijk de beslissing moeten maken”(s6) “Ik ervaar bij het maken van de beslissing zeker wel eens een tijdsdruk. Dat komt vooral omdat het veel tijd kost om overal van op de hoogte te zijn” (s7)

The majority feels (moderately) pressured when making decisions. They indicate that they experience problems in collecting and processing all the information before they have to make a

Pg. 48

BI systems to mitigate information load with store managers

decision. There is a remarkable variability in the indicated time available. Even though all subjects have the same supply-frequency from vendors, the indicated time available varies from less than four hours to more than a day. We could not find a clear reason for this variability, but based on subjects 6 and 7‟s quotes we presume that this can be a result of high information load. This implies a two-way relationship between time pressure and information load. We apply descriptive analysis (see Table 7.3) to explore the relationship between emotional load and perceived time pressure. This analysis shows that five out of seven subjects fit the proposition that high (low) time pressure will lead to high (low) emotional load. The other way around, this is the same for the proposition that high (low) emotional load results in high (low) perceived time pressure. Table 7.3 Descriptive analysis for information load and time pressure

High perceived time pressure

Low perceived time pressure

s3, s4, s5, s7 (4)

s1, s2, s6 (3)

High emotional load

Low emotional load

s3, s7 (2)

s4, s5 (2)

High emotional load

Low emotional load s1, s2, s6 (3)

From the content analysis we also find that task interruptions increase time pressure and decrease a store manager‟s capacity to process information properly. “Ik heb het wel eens dat als er veel belangrijke dingen zijn waarvoor ik gestoord wordt, er minder tijd over blijft voor de dagelijkse communicatie” (s2) “Het is een hectisch beroep, er komen altijd dingen tussendoor” (s7)

Information Amount Descriptive statistics shows that store managers are generally satisfied with the amount of information they receive. The evidence states that both store managers are satisfied with both the frequency and the amount of the information they receive. Also there is a fairly small variability in the data for the different types of information. According to the system designer, all subjects

Pg. 49

MSc thesis M.J. van Strien

Analysis

receive the same amount of information in the same frequency. Our data shows that there also is a small variability in the (subjectively) perceived information amount.

Product prices

Actual Sales

Financial performance

Steering reports

Benchmarks

Advice IT-systems

Suypplier information

Promotions/sales

Customers

Colleagues/Competitors

(market) research

Local

National

Median Mode Count

Sales forecasts

1=too few 5=too much (amount)

Stock level

Table 7.4 Descriptive statistics for information amount (1/2)

2 2 7

3 3 7

4 4 7

3 3 7

4 4 7

4 4 7

3 3 7

3 3 7

2 2 7

3 3 7

3 3 7

3 3 7

2 2 7

3 3 7

3 3 7

Product prices

Actual Sales

Financial performance

Steering reports

Benchmarks

Advice IT-systems

Suypplier information

Promotions/sales

Customers

Colleagues/Competitors

(market) research

Local

National

Median Mode Count

Sales forecasts

1=too little 3= too often (frequency)

Stock level

Table 7.5 Descriptive statistics for information amount (2/2)

2 2 7

2 2 7

3 3 7

2 2 7

2 3 7

2 2 7

3 2 6

2 2 7

2 2 7

2 2 7

1 1 7

2 2 7

1 1 7

2 2 7

2 2 7

However we find a larger variability in financial, steering and benchmark reports. This was supported during the interviews. “Zestig tot zeventig procent van de informatie is belangrijk voor mijn bedrijfsvoering, de rest lig ik niet wakker van” (s2) “De financiële sturing vind ik niet altijd nodig, ik weet goed hoe het gaat en daarbij hoef ik geen cijfertjes vanuit C1000 Retail”(s4)

Pg. 50

BI systems to mitigate information load with store managers

The subjects that felt that they receive too much financial reports and benchmarks, indicated that they did not see the added value of this information. They were typically the managers who based their decisions on experience and intuition, rather than on information. This implies that the relevance and perceived usefulness of information is an important determinant for the perceived information load.

In general the information is considered to be very important, both for the management of the stores and operations. “Sommigen informatie is essentieel voor je bedrijfsvoering” (s1) “De meeste berichten zijn wel relevant. Ik denk bijna nooit dat ik een bericht niet had hoeven lezen” (s2)

To gather the information, the store manager has multiple sources (e.g. C1000net, e-mail, conversations). This makes the information dispersed, especially as external information of becoming more important. “Het meeste vanuit C1000 gaat via C1000net, daarnaast ook veel via email” (s1) “Ik moet zeggen dat het steeds lastiger wordt om goed met die informatie om te gaan. Ze proberen dat wel te stroomlijnen met C1000net, maar soms is dat bijna niet meer te volgen. Dan moet je echt gaan zoeken naar nieuws en moeten we steeds vaker gaan filteren” (s3) “Vooral de externe informatie speelt een belangrijke rol in mijn werk. Deze wordt niet altijd via het systeem geleverd” (s5) “Als je kijkt naar informatie systemen dan hebben we een aantal losstaande systemen ter beschikking. Bijvoorbeeld C1000net, het bestelsysteem en het facturensysteem” (s5) “Eigenlijk ben ik best tevreden over de ICT-systemen, alleen ik verlies er soms het overzicht door” (s7) “Daarnaast moet ik ook vaak zoeken om alles bij elkaar te krijgen, ik vind dat C1000net een betere structuur kan krijgen” (s7)

The interviews have primarily focused on the information amount received from the intranet and the Order Advising System (AOS). To present a clear overview of the perceived information amount the systems are addressed separately: Regarding C1000net we find that the structure, presentation and navigation is not very intuitive to the store managers. This complexity makes it hard to find the right information. Also there is a lot Pg. 51

MSc thesis M.J. van Strien

Analysis

of information in C1000net, and to remain the overview the messages can be categorized into the store sections. This is a good measure to prevent overload, but the structure does not always correspond with store sections, resulting in irrelevant messages in an individual‟s queue. The effect of irrelevant information is strengthened by lack of an ability to order or filter the messages. Additionally subjects are not satisfied with the information pertinence. Some feel that C1000net makes is easier for C1000 to communicate e.g. news or mistakes, but do not feel that the information is always pertinent to them, and rather have to search for additional information instead. “C1000net wordt de Schuitema als communicatiemiddel ingezet om het voor hen makkelijker te maken. “Terwijl er nog niet helemaal wordt gekeken naar hoe het voor de ondernemer makkelijker wordt gemaakt” (s2) “Je kunt zelf bepalen of je de informatie uit C1000net wilt zien”(s2) “De ondernemer is het lastig, die heeft een gigantische lijst met berichten en hij moet overal achteraan gaan bellen als hij denkt gaat dit wel goed” (s2) “Er is geen goed onderscheid tussen ondernemersberichten en medewerkersberichten. Hierdoor krijg ik de neiging om alles te gaan lezen. Daardoor kunnen medewerkers bij belangrijke documenten”(s2) “Mijn bezwaar in wat zij doen is, ze gooien informatie over de schutting heen. En dat vind ik niet goed ” (s2) “Ik zou er meer voor zijn dat je bijvoorbeeld inlogt als X en dat dat systeem dan weet welk bericht voor haar is” (s2) “Er moet iemand zijn die, voordat het die trechter ingegooid wordt, bepaald dit is wel of niet interessant” (s3) “Het intranet is eigenlijk alleen een berichtendoorgeefluik. Ik mis het vaak dat ik iets met het bericht kan doen, bijvoorbeeld weggooien, of juist doorsturen naar de verantwoordelijke”(s4) “Er komt veel informatie op me af. Ik heb het gevoel dat de meeste informatie via C1000net komt”(s6) “De informatie is zeker veel. Daarnaast moet ik ook vaak zoeken om alles bij elkaar te krijgen, ik vind dat C1000net een betere structuur kan krijgen” (s7)

The AOS produces a lot of (structured) information, but more important, decreases the required processing efforts by „automatically‟ combining inventory levels, sales forecasts and the store manager‟s preferences regarding its inventory. The AOS produces an average of 1500 orderlines per day which are presented to the store manager as the order advice – waiting for his approval before sent to the distribution centre. Because the AOS does not incorporate external factors (e.g. Pg. 52

BI systems to mitigate information load with store managers

weather and events) into the advice, 10% of the advices (aprox. 30 orderlines per day) need to be checked manually. This results in an extra information load, but is not considered a problem. The AOS lets users create an attention group to filter the order lines that require attention; e.g. ice cream when weather forecast is sunny. “Aandachtsgroepen kun je zelf aanmaken en daarin kun je aangeven wat je belangrijk vindt dat er gecontroleerd wordt” (s1) “De basis van het besteladvies is goed. 90% van de gevallen is het advies goed, is 10% behoeft wat meer aandacht. De managers moeten hierop focussen. Hiervoor zijn aandachtsgroepen gemaakt in het systeem”(s2) “De systemen leveren ons veel informatie, maar halen dan ook veel van het reken- en denkwerk weg” (s4) “Het systeem houdt geen rekening bijvoorbeeld het weer voor verkoopvoorspelling. Dit moeten we dan zelf dus nog doen”(s4) “Voor de forecast houden de systemen geen rekening met de omstandigheden zoals weer en evenementen. Dit vind ik wel jammer, want nu moeten we dat handmatig gaan verwerken” (s5)

Decision responsibility Although we determined this variable emergent, we have collected quantitative data about the decision autonomy. All subjects indicate to be fully autonomous for the decision. There is some variability in perceived steering efforts from C1000, but this does not impact this variable. From the content analysis we find that some subjects feel more responsible for the decision than others. "Er zijn vijf afdelingsmanagers die verantwoordelijk zijn voor hun eigen toko" (s1) "Ze zijn zelf verantwoordelijk voor de informatie" (s1) “Ik kan zo drie weken wegblijven en dan gaat alles goed” (s2) "De dagelijkse dingen voor bestelling doe ik niet, maar ik controleer wel wat ik zie in de winkel. Aandachtspunten uit de winkel zijn aanleiding om in de computer te kijken" (s2) “Ik overleg wel veel met ze, maar ze mogen zelf de beslissingen maken”(s3) "De afdelingen kennen hun eigen verantwoordelijkheid voor het reilen en zeilen op de vloer. Zij zijn verantwoordelijk voor het goed uitvoeren van de processen” (s4) "Klopt, de afdelingsmanagers mogen ook hun eigen beslissingen maken. Natuurlijk overleggen we hier wel over”(s4)

Pg. 53

MSc thesis M.J. van Strien

Analysis

“Ik zorg wel dat wat van belang is binnen het bedrijf, wat van belang is, dat ik dat wel zie ergens. Ik wil graag van alles op de hoogte blijven, dat is mijn werk” (s5) “Mijn visie is dat je veel verantwoordelijkheid bij het personeel neer moet leggen”(s5) “Als ik het zelf niet weet, kan ik nooit weten of het überhaupt in de winkel goed gaat komen” (s6) “Als ik zelf de informatie niet zie, kan ik ook niet bepalen of het in de winkel ergens wel goed terecht komt. Daarbij hoef ik het niet helemaal uit te pluizen, maar wil ik wel weten waar het over gaat en vooral weten of ik dat in de winkel terug zie”(s6) “Ik geef de afdelingsmanagers wel taken, maar wil dat ze toch alles met mij bespreken” (s7) “Ik vind wel dat afdelingsmanagers een eigen invloed mogen hebben op de beslissingen, maar ik blijf toch eindverantwoordelijke” (s7)

All the stores are divided into sections, and decision activities are delegated to section managers. This implies decentralized decision responsibility. However we find that some store managers tend to stay involved in these decision activities, rather than fully trusting the section managers. We find that managers who want to stay more involved have a higher information need, and thus tend to collect more information. Opposing, the other store managers place full decision responsibility at the section managers, hence do not need to keep involved with certain information – decreasing their own information load. Decision responsibility varies from „wanting to absorb 90% information in order to check the operations‟ (s5) to „being able to leave the store for several weeks and not feel worried about the decisions‟.

7.2.2

Mediating variable

Information load was measured using two measures (see Appendix II). We analyse the results of these measures separately. Ten (out of 16) scales of Russel and Mehrabian‟s (1974) extended measure are rated with a median of four, indicating that the subjects do not experience a high information load (see Table 7.6). When viewing the three constructs of information load (complexity, novelty, and reliability) separately, analysis shows that complexity is rated lower than novelty and reliability. However, we find only one scale that has a median of lower than three small scale. This indicates that the subjects do receive a fairly large amount of information.

Pg. 54

BI systems to mitigate information load with store managers

Table 7.6 Descriptive statistics for Russel and Mehrabian's scale

Information load

Precise

Reliable

Complete

Usual

Familiar

Important

Reliability

Meaningful

Common

Pertinent

Consonant

Consistent

Varied

Good form

Patterned

Simple

1=fully disagree 5=fully agree

Novelty

Small scale

Complexity

Median

3

3

2

3

4

4

3

4

4

4

3

4

4

4

4

4

Mode

3

4

2

3

4

4

4

4

4

4

3

4

4

4

4

4

Minimum

2

2

1

2

2

2

2

3

3

2

2

3

3

3

3

2

Maximum

4

4

3

4

4

4

5

5

5

5

4

5

4

5

4

4

Count

7

7

7

7

7

7

7

7

7

7

7

7

7

7

7

7

When looking at Rutkowski and Saunder‟s (2010) scale, these quantitative measures show that the subjects overall are not overloaded with information (see Table 7.7). However, the evidence for information load shows a higher variability than the independent variables. Interestingly there is a relatively large variability in the perceived information load; especially for information complexity, irritation, and ability to handle, process, and cope with the information.

Table 7.7 Descriptive statistics for Rutkowski and Saunder’s scale

Median Mode Minimum Maximum Count

2 2 1 4 7

2 2 2 4 7

Cannot cope with Overwhelmed by the effort Irritated Emotionally pressured Confused

Cannot handle

1=fully disagree 5=fully agree

Cannot process

Cognitive load Emotional load

2 2 1 4 7

2 2 2 4 7

2 2 1 5 7

2 2 2 4 7

2 2 2 3 7

Pg. 55

MSc thesis M.J. van Strien

Analysis

Overall the quantitative evidence shows that the subjects are satisfied with the information load they perceive, and do not feel overloaded. We use content analysis for triangulation, and to further explain the findings. This shows that the subjects indeed are satisfied with the information they receive. The subjects especially indicate that they are satisfied with the performance of the Order Advising System (AOS). Although the system produces a lot of (order) information, it also decreases the store managers computing and information processing effort. The ability to create attention areas enables the store managers to filter the relevant information for their decision making activities, and therefore decreasing the information load. “Inhoudelijk kan ik daar best mee omgaan. Het kan wel eens zijn dat je veel dingen krijgt en dat je daar veel tijd mee kwijt bent” (s1) “Ik vind de hoeveelheid informatie die via C1000net wordt verspreid fors. Het is en onverwacht en vrij veel”(s2) “Ondanks de hoeveelheid kan ik toch goed filteren” (s2) “Ik maak daarmee de stapel van informatie voor mezelf erg hoog, maar doe dat wel bewust”(s3) “Ik moet zeggen dat het steeds lastiger wordt om goed met die informatie om te gaan. Ze proberen dat wel te stroomlijnen met C1000net, maar soms is dat bijna niet meer te volgen. Dan moet je echt gaan zoeken naar nieuws en moeten we steeds vaker gaan filteren”(s3) “De systemen leveren ons veel informatie, maar halen dan ook veel van het reken- en denkwerk weg”(s4) “Ik ben over het algemeen tevreden met de informatie die we krijgen, dus ik heb niet veel moeite om hiermee om te gaan” (s4) “Ik voel me regelmatig onder druk gezet door de hoeveelheid informatie die ik moet verwerken”(s6) “Dat vind ik frustrerend, want mijn medewerkers weten niet wanneer ze iets kunnen verwachten en weten niet wie er actie op heeft ondernomen”(s2) “Als je dit ziet dan denk je 'hoe kan dit nou?', alleen de informatievoorziening is gegroeid in de loop van de jaren. Het is niet dat de informatie in 1x op ons af is gekomen” (s1) “Alleen door mijn ervaring weet ik informatie goed te filteren. Dan kan ik goed bepalen of ik het moeten weten of dat dit naar iemand anders moet”(s4)

The subjects also indicate to be fairly satisfied with the intranet (C1000Net), but the analysis draws our attention towards some remarks on this system. The subjects do indicate that the portal

Pg. 56

BI systems to mitigate information load with store managers

reduces the number of dispersed information sources, however some are not fully satisfied with the relevance and pertinence of the information. “De indeling naar afdeling (in C1000net) zou nog een stap gedetailleerder moeten. De indeling is ook niet in elke winkel hetzelfde, maar in C1000net wel” (s2) “De opbouw van C1000net is niet intuïtief. Het is ouderwets met hoofdstukken”(s2) “Als je kijkt naar de inhoud ben ik wel tevreden. Maar als je kijkt naar het proces, daar heb ik wel wat opmerkingen over”(s3) “C1000net is niet zo ingericht dat het aansluit bij onze afdelingen. Dit zorgt soms dat er onbedoeld berichten bij iemand komen”(s7)

First, the structure of the portal allows the information provider to bundle the messages for the (group of) receiver(s) – potentially reducing the number of irrelevant messages that are directed towards a decision maker. However, some subjects indicate that the structure of the portal does not align with the structure of the sections in their store. Due to the static structures, they feel that they (and their employees) receive too much irrelevant information.

Concerning the pertinence of the information, some subjects feel that the information providers use the portal just for messaging, not to really help them with the application of the information in the message. The subjects indicate that it is not always clear what is meant with the message, or sometimes they need to search for additional information in other systems – obviously increasing their information load. Additionally some subjects indicate that the information is not interactive. "Het afvinken van berichten zou ik heel fijn vinden" (s2) “Het intranet is eigenlijk alleen een berichtendoorgeefluik. Ik mis het vaak dat ik iets met het bericht kan doen, bijvoorbeeld weggooien, of juist doorsturen naar de verantwoordelijke” (s7)

They feel that the lack to interact with the message (e.g. mark as read, publish a reply) forces them to use other, additional communication channels to make the information pertinent.

Pg. 57

MSc thesis M.J. van Strien

7.2.3

Analysis

Dependent variables

Descriptive statistics for the dependent variables show that perceived decision quality is satisfactory, but actual inventory performance is bad. The subjects indicate not to have problems with deciding when to order, but do experience difficulties in determining the quantity. The prior can be explained because the decision is found to be structured; hence the decision (i.e. moment of ordering) is often the same. On the other hand, determining the order quantity is experienced more complex. Table 7.8 Descriptive statistics for dependent variables

Financial impact

Stock costs

Waste

Understock

Overstock

Subjective

Quantity

1=fully disagree 5=fully agree

Frequency

Decision Quality Business Performance

Median

4

3

4

2

2

2

2

2

Mode

4

2

4

2

2

2

2

2

Minimum

2

2

3

1

1

2

2

2

Maximum

5

5

4

4

4

3

4

5

Count

7

7

7

7

7

7

7

7

“Ik ben ervan overtuigd dat het voorraadbeheer hier heel goed gebeurt” (s1) “Nee, ik heb absoluut nooit moeite om te bepalen hoeveel we moeten bestellen” “Natuurlijk maak ik ook wel eens fouten, maar over het algemeen ben ik tevreden over mijn beslissingen” (s2) “Ik ben tevreden met mijn beslissingen, ik hoor er niemand over klagen” (s3) “Het is niet moeilijk om te bepalen wanneer er besteld moet worden, dit is elke dag hetzelfde” (s3) “Ik bestel meestal wat extra artikelen zodat de klanten geen lege schappen hebben” (s4) “Ook al doe ik dit nog niet zo lang, ik ben ervan overtuigd dat ik dit goed doe” (s5) “C1000 bepaalt wanneer we moeten bestellen, dus daar kunnen wij weinig over beslissen” (s6) “De meeste beslissingen zijn goed, het ligt meer aan andere dingen waardoor de schappen leeg zijn” (s7)

Pg. 58

BI systems to mitigate information load with store managers

“Het is vaak wel lastig om te bepalen hoeveel er besteld moet worden, je moet met allerlei dingen rekening houden” (s7)

Content analysis shows that determining the order quantity is dependent on a variety of factors; e.g. inventory level, sales forecast (based on historical data, and on intuition/experience), promotions (cross sectional), trends, customer requirements, vendor status, and desired service level. Although the advice from the AOS is considered right for at least 90percent, the subjects indicate the importance of determining the right order quantity for optimal inventory management. The subjects indicate that approximately 10 percent of the order advices have to be reviewed manually. “Wij moeten de bestelling nog helemaal aanmaken en beoordelen. Wij moeten zelf rekening houden met hiaten door bijvoorbeeld weersverwachting” (s1) “De basis van het besteladvies is goed. 90% van de gevallen is het advies goed, is 10% behoeft wat meer aandacht. De managers moeten hierop focussen. Hiervoor zijn aandachtsgroepen gemaakt in het systeem”(s2) “In het weekend is er meer fluctuatie. Omdat we dan geen real-time hebben, kan het gebeuren dat bwe 50 tot 100 producten niet bestellen" (s3) “Het maken van de beslissingen ervaar ik niet als lastig. Wel is het zo dat 10% van de artikelen afhankelijk is van de situatie. Hierbij moeten we rekening houden met het weer enzo” (s4) “Het systeem houdt geen rekening bijvoorbeeld het weer voor verkoopvoorspelling. Dit moeten we dan zelf dus nog doen” (s4) “Ik vind het vooral moeilijk om te beopalen hoeveel ik moet bestellen”(s5) “Ik ben niet ontevreden over de bestellingen die ik maak, maar ik denk wel dat het beter kan. Je moet op een gegeven moment tijd afwegen tegen het resultaat” (s5) “Voor de forecast houden de systemen geen rekening met de omstandigheden zoals weer en evenementen. Dit vind ik wel jammer, want nu moeten we dat handmatig gaan verwerken” (s6) “Nee, daar kan ik heel kort over zijn; ik vind dat het bestellen hier prima gebeurt. Ik - en de medewerkers ook niet - heb nooit moeite om te bepalen wanneer en hoeveel er besteld moet worden”(s7)

This is because the OAS does not include some external factors (e.g. weather forecast, events, etc) into the advices. The articles that are dependent on these external factors can change over time; e.g. winter, sport events, and promotions (trends). Manually checking and adjusting these

Pg. 59

MSc thesis M.J. van Strien

Analysis

order advices is based on multiple information sources and intuition (i.e. experience and craftsmanship)

Actual inventory performance is rated with a score of two. This means that inventory performance is below satisfactory level, and the inventory is often overstocked or understocked – resulting in waste, empty shelves and decreased service level. The content analysis shows that unsatisfactory inventory performance can be due to various factors other than information overload. "Ik zal de laatste zijn om te zeggen dat het hier perfect gaat“ (s1) "Natuurlijk worden er wel eens dingen weggegooid, omdat dan de THT datum verlopen is" (s1) “Het komt ook regelmatig voor dat er producten uit voorraad zijn, of dat er juist te veel in het magazijn liggen. Dit kost altijd geld”(s5) “Je kunt best wel opschrijven dat gemiddeld 1% van de omzet weg wordt gegooid omdat producten te oud zijn. Met voedingsmiddelen is dit nou eenmaal snel het geval” (s6) “Helaas komt het dan toch wel eens voor dat er te veel of te weinig producten op voorraad zijn. Hierin spelen andere factoren ook een belangrijke rol”(s7)

To prevent from drawing incorrect inferences for the unsatisfactory inventory management, we asked the subjects to rank the reasons for this performance. The results show that information quality is the most important factor, followed by delivery failure, information overload and deliberate choice. “Ik heb liever teveel producten in het magazijn zodat de klant zo min mogelijk lege schappen ziet. Het is een bewuste keuze om soms wat meer of eerder te bestellen” (s4) “Vaak komt het dat een besteling niet goed geleverd wordt. Ik denk dat dit zelfs dagelijks gebeurt. Hierdoor is het voor ons lastig om de voorraad goed te beheren” (s5) “Wat ik alleen jammer vind is dat de informatie soms niet klopt. Dan denk ik vooral aan voorraadniveaus waardoor bestellingen verkeerd kunnen zijn” (s5) “De meeste verkeerde bestellingen komen doordat de informatie niet klopt” (s6) “De informatie uit de systemen blijkt niet altijd te kloppen. Dit kan resulteren in foute bestellingen” (s7)

Pg. 60

BI systems to mitigate information load with store managers

7.3

System designers

For matters of comparison, system designers from different retail chains (C1000 Retail and Plus Retail) were interviewed. This allows to make a brief comparison between their philosophy about the role of information management and supply through information technology (IT). However no store managers from the comparing case (retail chain Y) was interviewed. Thus it is not possible to validate this philosophy on (1) perception of the managers, and (2) performance of the information systems (IS‟s) supporting this philosophy. For this reason the analysis is started by addressing the similarities in the philosophy and thereafter emphasize the differences between the two organizations. First of all it was found that both retail chains emphasize the importance of IT in the information supply for store managers. Although retail chain Y has started developing the systems somewhat later, they both claim that the IS‟s fully support the store managers in their operations and support both the retail organization and the store manager to work more efficient and effective. They also both acknowledge that IS‟s intend to decrease information load caused by dispersed information sources. Both organizations support the managers with an intranet, financial systems, and OAS‟s. “Voor mij is IT een middel om de processen goed te laten lopen. An sich helpt IT met het effectiever en efficiënter uit laten voeren van processen Het is de informatie die je eroverheen stuurt dat het hem doet. IT kan hierbij helpen” (d1) “Het belangrijkste van de IT vind ik dat het werkt, de beschikbaarheid. IT is een randvoorwaarde, dat werkt gewoon” (d1) “Uiteindelijk zijn alle processen door ons systeem ondersteund” (d1) “IT ondersteunt de ondernemers door ze zoveel mogelijk werk uit handen te nemen. G&A is hier een goed voorbeeld van, dit zorgt dat er automatisch een besteladvies wordt gegenereerd” (d2) “Je moet wel omdat IT essentieel is geworden voor de bedrijfsvoering. Er gaat hier bijna niets meer zonder dat IT er bij komt kijken” (d2)

Also both organizations claim to give the store managers a relatively large autonomy in making decisions. The IS‟s align with this organizational strategy as they allow a decentralized control.

Pg. 61

MSc thesis M.J. van Strien

Analysis

“De systemen hebben gevolgd wat de policy was. En dat was dat ondernemers de vrijheid hebben, die kun je alleen adviseren” (d1) “Ik denk dat ondernemers bij ons nog autonomer zijn dan bij C1000. Plus is een cooperatieve organisatie, de ondernemers zijn aandeelhouders van de organisatie” (d2)

Because of this, they also place responsibility of inventory information quality at the store managers. Also due to the decentralized control, both systems designers indicate that they have to make trade-offs between the controllability and functionality of the system. This trade-off is especially important during the implementation of the systems. Both designers indicate that the system is adjusted to the store-characteristics during implementation, however it is not adjusted to store manager-characteristics. “De informatie die je voor het bestellen nodig hebt, de voorraad dus, kan niet uit de winkel. Maar vervolgens, dus wat hij moet gaan bestellen, dat kunnen we beter centraal”(d1) “We zijn duidelijk geworden in dingen die kunnen, moeten en mogen. Daarmee willen we veel meer dingen centraal oppakken” (d1) “Wij kennen voor onze ondernemers drie hoofdsystemen.: personeel & efficiency; goederen & assortiment, en finance. Daarmee ondersteunen wij alle winkels” (d2)

We find that the philosophy behind the OAS is the same. Both organizations aim to reduce computing and processing effort, but allow the store managers to override the system‟s advice. Also the internet portal is considered to be one of the most important systems as the messaging between the central organization and the store managers is very important. “C1000net is zoveel mogelijk ingericht op bundeling van de informatie. Daardoor is het wel gebundeld, maar wel erg veel” (d1) “Er staan veel berichten op waarbij er vanuit C1000 correcties worden gedaan richting de ondernemers” (d1) “Wij hebben PlusWeb. Daarover gaan heel veel berichten heen-en-weer, eigenlijk veel te veel” (d2)

Both organizations indicate that these channels are used for a vast amount of information and that a lot of information is sent to the store managers.

Pg. 62

BI systems to mitigate information load with store managers

“Je hebt dagberichten en weekberichten. Dagberichten zijn meer de urgente correcties en operationele dingen. Heel eerlijk gezegd zijn die dagberichten meestal dat wij centraal iets fout hebben gedaan. Ik heb niet eens tijd om ze allemaal te bekijken, het zijn er zo veel” (d1) “Maar ik geloof wel dat er altijd een overload is. Dat kan op hele kleine dingen zijn, maar er wordt zoveel informatie gepusht via PlusWeb” (d2)

Business Intelligence (BI) systems are not used to support store managers, but rather to support decision making at the retail chain. Information from the stores (primarily post-data from sales systems) are used to fill the data warehouses and allow strategic and tactical decision makers to analyse chain performance. The quantitative evidence (see Appendix V) also shows that both philosophies are fairly similar. There are however difference in the way the designers see the dissemination of the information. C1000 Retail prefers a pull-philosophy and indicates that not all information has to be disseminated through the IS‟s. Opposing, organization Y prefers a push-mechanism and dissemination of all the information through the IS‟s. When analysing the qualitative evidence two minor details at organization Y are found. Although being minor, they particularly address the significant remarks on information load were found at the store managers of C1000 Retail; inventory information quality, and information pertinence and relevance. Plus Retail addresses the information quality by using smart automated alerts. If, for instance, the system expects to sell eights products of an article, but it recognizes that only two have been sold today, the shelve might be empty while this is not shown in the IS. When this occurs, the system sends an alert that requests an employee to compare the IS with the actual inventory level. “Het komt niet zo heel vaak voor dat informatie over de voorraad in ons systeem niet klopt. We hebben een aantal alerts daarvoor ingericht. Bijvoorbeeld als we verwachten dat we vandaag 8 van dat artikel verkopen, en we merken dat er maar 3 verkocht worden. Dan kan het goed zijn dat het schap leeg is. Daardoor wordt een telopdracht aangemaakt” (d2) “Er worden per dag zo'n 50 tot 100 gestuurde telopdracht voor een supermarkt gegenereerd. Daarmee worden ook diefstal, derving en breuken gemonitord. Dat zorgt voor een goede voorraadprestatie. We merken dat de prestatie hiervoor wel is verbeterd” (d2)

Pg. 63

MSc thesis M.J. van Strien

Analysis

The latter remark (information relevance and pertinence) is addressed by letting an editor check the messages before they are published on the portal. The designer indicates that this (full-time) editor reads every message and determines whether the included information is clear, complete, and if it is addressed to the right (group of) recipient(s). “Wij hebben PlusWeb. Daarover gaan heel veel berichten heen-en-weer, eigenlijk veel te veel Daarom hebben we er nu een redacteur opgezet. Hierdoor kunnen mensen niet meer zomaar berichten erop plaatsen, maar de redacteur vraagt altijd wat bedoel je ermee. En dan nog gaat het niet altijd even goed. Soms goed er een bericht dat ze de promotie zelf in het systeem moeten zetten. Dan zou het wel handig zijn als er gelijk bij zou staan hoe ze dat doen” (d2)

Pg. 64

BI systems to mitigate information load with store managers

8. Conclusion & discussion In this chapter, our conclusions relating to each aspect of our work is presented. In addition, the limitations of our finished work and the extensions for future research are discussed.

8.1

Conclusions

This thesis investigates the relationship between information overload and decision quality. It is evidenced that store managers are not experiencing information overload when performing managerial decision making activities. The lack of variability in the variables disables to formally test whether the independent variables explain information overload. Also, due to the inexistence of information overload it is failed to investigate whether information overload decreases decision quality.

Additionally the research investigated a relationship between decision quality and business performance. Although the research shows that actual inventory (management) performance is dissatisfactory, this finding cannot be contributed to wrong decision making performance caused by information overload. Rather it was found that information quality and delivery failure are the most important contributors for decreased inventory management performance. Information overload was indicated to be the third contributor to poor inventory performance.

Failing to formally draw inferences from the propositions compels to apply a broader approach to answer the research question. So instead, the case study is used to explain why store managers do not explicitly suffer from information overload. When referring back to the problem statement, it is found that a majority of the subjects acknowledge the developments of increasing competition and growing information amount from internal and external sources. This creates a situation where - while information processing requirements (IPR) keep increasing - decision quality has become increasingly important to retain a competitive position. Drawing on human information processing theory, a better insight into the investigated situation is provided. This is

Pg. 65

MSc thesis M.J. van Strien

Conclusion & discussion

done by aiming at how the variables affect store managers‟ information processing capacity (IPC), and IPR. This allows to explain how the variables affect information load.

8.1.1

Information processing capacity

On the one hand computerized information systems (IS‟s) are found to positively affect a store manager‟s IPC as they reduce the effort that a store manager needs to put into searching, gathering and processing the information. This is particularly the case for the Order Advice System (OAS) and the internet portal, as they both cluster and filter the information from various dispersed sources, and process the information for decision making use. Due to the use of these systems, the store manager is no longer required to acquire information from dispersed systems, and – especially for the OAS – does not need to process and calculate the data for decision making use. On the other hand, time pressure is found to have a decreasing effect on the store manager‟s IPC. Although store managers generally do not experience time pressure, a relative large variability in the perceived amount of time available for inventory decisions was found. It was found that time constraints sometimes prohibit the store manager to acquire and incorporate all the relevant information into the decision making activities. Failure to integrate the relevant information into decision activities might lead to the use of a less accurate heuristic decision strategy. Application of a heuristic decision strategy disables the store manager to properly evaluate the decision alternatives; i.e. how much to order. This aligns with the finding that store managers consider the order quantity the most complex decision. The relative large variability in the perceived time available is remarkable because all subjects should have the similar (absolute) time available. This suggests that the relationship between time pressure and information load is duplex: high time pressure increases information load, and high information load increases perceived time pressure.

Moreover, task interruptions have a frequent decreasing effect on IPC. The participant observations evidence that a supermarket is a hectic decision environment, and that store Pg. 66

BI systems to mitigate information load with store managers

managers are interrupted frequently. Due to these interruptions the store manager experiences both increased time pressure and dispersed decision focus.

8.1.2

Information processing requirements

Paradoxical to its increasing effect on the IPC, the application of IS‟s was also found to increase IPR. As Woods, Patterson et al. (2002) argue, technological developments have advanced an organisation‟s ability to collect and transmit information, leading to production of a vast amount of information. Due to the application of IS‟s, store managers receive more information. Particularly, the internet portal produces a huge amount of information for the store managers. Although the system designers indicate to be aware of this fact - and have initiated mitigating measures (e.g. clusters to filter irrelevant information) - the system increasingly requires the store manager to process information. Also, the structure of the clusters in the portal rarely aligns with the clustering in the stores. For this reason store managers indicate to that the portal provides irrelevant information frequently. Hence the increase in information load is dedicated to increasing information complexity (particularly caused by information amount), rather than information novelty.

Although information overload was not found, there is a remarkable variability in the managers‟ ability to process the amount of information. This suggests that the increasing amount of information increasingly becomes a threat for store managers‟ ability to properly process the information needed to make a decision. This however is indicated as a future anxiety by some subjects, and will not be included as a formal conclusion.

Store managers agree that the increase in information has reached a point at which they face the concern of being overloaded. Information amount has however increased steadily over the past years. This allowed the store managers to get acquainted with both the information, and the IS‟s providing the information. This experience enables them to reduce their search effort by effectively chunking the information, hence decrease the IPR. Store managers indicate that they

Pg. 67

MSc thesis M.J. van Strien

Conclusion & discussion

do not feel overloaded with information because their experience enables them to determine what information is relevant and filter the information that is considered important – obviously reducing IPR.

Also it was found that decision responsibility is an important determinant of IPR. Although store managers indicate that they are formally responsible for the decisions made, variability in the perceived degree of responsibility (i.e. informal responsibility) was found. It is argued that informal responsibility is a function of organisational design and management style. Informal responsibility varies from “Ik overleg wel veel met ze, maar ze mogen zelf de beslissingen maken” to “Ik geef de afdelingsmanagers wel taken, maar wil dat ze toch alles met mij bespreken”. Obviously the latter increases the IPR.

8.1.3

Overall conclusion

From the case study, various variables are deducted that were found to be integral to the concept of information overload with store managers. The empirical model that can be drawn from these conclusions is presented in Appendix VIII. The inexistence of the information overload with store managers implies that the present research is not possible to address how business intelligence (BI) systems can mitigate information load. It is however argued that not the future application, but rather the existence application of BI systems, allows us to cautiously draw inferences on how BI systems mitigate information load.

Although not formally being considered as BI systems, it is argued that the OAS matches the characteristics of a BI system; the system gathers data from multiple dispersed sources (e.g. inventory management system, sales systems, and pricing systems), and allows users to interactively generate the information they need for decision making use. It is particularly found that the ability to specify areas of attention allows store managers to adopt the information to their specific information needs – which may vary per decision task and situation. Herewith the OAS reduces the information supply from approximately 1.500 order lines per day to the 150 order lines that require evaluation, obviously reducing the search effort and IPR. The evidence

Pg. 68

BI systems to mitigate information load with store managers

does however not contain information load measures in a situation without use of BI systems. Therefore comparison is disabled.

Concluding, it is suggested that BI systems allow store managers to present the information into the decision context. Thereby BI systems are suggested to decrease information amount and time pressure. Store manager‟s experience was found to be a prerequisite to chunk the information and operate the IS‟s. Additionally decision responsibility was found to be an important variable in explaining store manager‟s information load. Low (informal) decision responsibility was found to decrease IPR because decision – and information - responsibility was delegated.

8.2

Discussion & limitations

This paragraph will address some discussions and limitations for the present research. These will primarily be based on evaluation of the research validity.

8.2.1

Research validity

A fundamental question regarding the results of the research is how valid the results are. Construct validity was addressed by extensive literature study that provided various validated measures for collecting the evidence (see Appendix II). This allowed the researchers to measure exactly what was intended to measure. Additionally, the reliability of the findings was addressed by gathering evidence from multiple sources (i.e. interviews, questionnaires and observations) and triangulation. Supplemented with control of the exogenous factors, this assured internal validity for proposition 1 (p1, p1a, p1b, and p1c). It is noted however that this was only possible with subjective measurement of the decision quality.

This is because business performance could not be used as an adequate measure for the objective quality of a decision. It is argued that the exogenous factors for the relationship between decision

Pg. 69

MSc thesis M.J. van Strien

Conclusion & discussion

quality and business performance cannot be controlled. Therefore, construct validity for the objective decision quality was poor, and consequently the research design failed to address internal validity for proposition 2.

Since the research was conducted within an organization, the internal validity was considered to be more important than the external validity. This disables to draw general inferences for the information overload concept. The comparison of IS designer philosophy however allows to cautiously present a view on how and whether IS designers at retail chains address information. Additionally this allows to present recommendation for C1000 Retail‟s designer.

From the evaluation of the research validity, some limitations and recommendations for future research are presented.

8.2.2

Disability to test propositions

The case study provides few variability for these variables, and consequently it is failed to formally test the propositions that these variables result in information overload. The relative low variability in the evidence was the case for the independent, mediating as well as the dependent variables. As was found through triangulation, there is convergence of the qualitative and quantitative data for most of the variables. We suggest that the low variability may be a result of (1) a lack of heterogeneity in the sample pool, or (2) the consistent use of a five point-scale for our questionnaires. We did however not have any information (e.g. personal traits) about the subjects prior to selecting the pool.

Due to this low variability is was disabled to formally test the propositions. It was failed to test the relationship between (1) experience, time pressure, information amount and information load, and (2) information load and decision quality. Consequently this prohibited to draw conclusions regarding the prediction of variability in the mediating and dependent variables.

Pg. 70

BI systems to mitigate information load with store managers

8.2.3

Improper measurement for decision quality

The aim of the research design was to objectively measure decision quality. The attempt to do so was fostered by Raghunathan‟s (1999) notion that previous research, that tried to measure decision quality, had examined a manager‟s perceived decision quality instead. Moreover, Hwang and Lin (1999) argue that improper operationalization of the decision quality measure may lead to incorrect conclusion drawing. It was attempted to measure decision quality by analysing quantitative inventory performance data. Unfortunately, due to accessibility restrictions, it was failed to collect the data. It is argued however that – even if the data was available – business performance is not a proper measure for decision quality. Business performance cannot be directly related to the quality of a decision. It is found that various exogenous factors may (also) affect business performance. This disabled to properly measure decision quality, and hence the research fell back on the subjective measure.

8.2.4

Search costs of BI

It is conclude that the application of BI systems decreases store manager‟s information load by tailoring the information to the individual needs. Also BI systems lower the effort put into acquiring the relevant information. From a cost-benefit perspective, this enables the decision maker to employ a more accurate decision strategy while keeping effort at the same level. The research however did not include the additional effort a decision maker has to put into handling the system in order to be able to find the correct information. It is argued that this additional search costs might even be similar to those when a decision maker is not using BI systems. It is propose that these costs decrease when the decision maker is acquainted with the BI system.

Pg. 71

MSc thesis M.J. van Strien

8.3

Conclusion & discussion

Recommendations for future research

The present research has conducted an extensive literature investigation in order to find the appropriate measures for the variables that were proposed to be integral to the situation being investigated. This has produced a protocol to collect both rich contextual and concrete quantitative evidence for the variables. It is argued that a combination of this evidence is needed to deeply investigate the concept of information overload in organisations.

This research has again proven that decision quality is hard to assess. The present study has done an attempt to create a measure to objectively assess decision quality by proposing a relationship between decision quality and business performance. This attempt has however failed, and given the failure of previous attempts, it is argued that future research should not be directed towards the relationship between information overload and decision quality. Rather information overload research should aim at explaining the occurrence of information overload in various contexts instead of trying to evidence its detrimental effects.

Due to lack of variability in the independent and mediating variables, the present study has failed to explain the occurrence of information overload. It has however conducted a deep case study which has proposed additional variables that may explain the variations in the information load variable. Particularly experience, the use of IS‟s and (informal) decision responsibility were found to be important in the concept of information overload. Also is it believed that BI systems have the potential of reducing information load because of their ability to interactively present the information. This relationship should however be formally tested through controlled experiments. These experiments should include varying use of BI systems, but also varying experience of using the system.

Pg. 72

BI systems to mitigate information load with store managers

References

Ackoff, R. (1967). "Management misinformation systems." Management science: 147-156. Ahituv, N., M. Igbaria, et al. (1998). "The effects of time pressure and completeness of information on decision making." Journal of Management Information Systems 15(2): 172. Azvine, B., Z. Cui, et al. (2005). "Towards real-time business intelligence." BT Technology Journal 23(3): 214-225. Azvine, B., Z. Cui, et al. (2006). Real time business intelligence for the adaptive enterprise, IEEE. Bacharach, S. B. (1989). "Organizational theories: Some criteria for evaluation." Academy of management review 14(4): 496-515. Baddeley, A. (1994). "The magical number seven: still magic after all these years?" Psychological Review 101(2): 353-356. Bawden, D. (2001). "Information overload." Library & information briefings(92): 1-15. Bawden, D., C. Holtham, et al. (1999). Perspectives on information overload, ASLIB. Bawden, D. and L. Robinson (2009). "The dark side of information: overload, anxiety and other paradoxes and pathologies." Journal of Information Science 35(2): 180. Bazerman, M. H. (2006). "Judgment in managerial decision making." Benbasat, I. and R. N. Taylor (1982). "Behavioral aspects of information processing for the design of management information systems." IEEE Transactions on Systems, Man, and Cybernetics 12(4): 439-450. Brueggemann, C. E., S. Reader, et al. (2008). Mitigating Information Overload: The Impact of" Context-Based Approach" to the Design of Tools for Intelligence Analysts, NAVAL POSTGRADUATE SCHOOL MONTEREY CA.

A

MSc thesis M.J. van Strien

References

Cannon, A. R. (2008). "Inventory improvement and financial performance." International Journal of Production Economics 115(2): 581-593. Casey, C. J. (1982). "Coping with information overload: the need for empirical research." Cost and management 56(4): 31-38. Chan, S. H., S. G. Sutton, et al. (2009). "The paradoxical effects of feedback and reward on decision performance." Cheung, W. and G. Babin (2006). "A metadatabase-enabled executive information system (Part A): a flexible and adaptable architecture." Decision Support Systems 42(3): 1589-1598. Chewning, E. G. and A. M. Harell (1990). "The effect of information load on decision makers' cue utilization levels and decision quality in a financial distress decision task." Accounting, Organizations and Society 15(6): 527-542. Choo, C. W. (2006). The knowing organization: How organizations use informationto construct meaning, create knowledge and make decisions. Cook, G. J. (1993). "An empirical investigation of information search strategies with implications for decision support system design." Decision Sciences 24(3): 683-698. Courtney, J. F. (2001). "Decision making and knowledge management in inquiring organizations: toward a new decision-making paradigm for DSS." Decision Support Systems 31(1): 17-38. Datta, A. and H. Thomas (1999). "The cube data model: a conceptual model and algebra for on-line analytical processing in data warehouses1." Decision support systems 27(3): 289-301. Davis, D. (2005). Business research for decision making. Davis, F., R. Bagozzi, et al. (1989). "User acceptance of computer technology: a comparison of two theoretical models." Management science: 982-1003. de Vries, E. (2005). "Epistemology and Methodology in Case Research: a Comparison between European and American journals." Proceedings of the ECIS 2005 Regensburg, Germany: 1309-1320.

B

BI systems to mitigate information load with store managers

Dequech, D. (2001). "Bounded rationality, institutions, and uncertainty." Journal of economic issues 35(4): 911-929. Driver, M. and T. Mock (1975). "Human information processing, decision style theory, and accounting information systems." Accounting Review 50(3): 490-508. Driver, M. J. and S. Streufert (1969). "Integrative complexity: An approach to individuals and groups as information-processing systems." Administrative Science Quarterly 14(2): 272285. Edmunds, A. and A. Morris (2000). "The problem of information overload in business organisations: a review of the literature." International Journal of Information Management 20(1): 17-28. Eisenhardt, K. (1989). "Building theories from case study research." Academy of management review 14(4): 532-550. Elbashir, M. Z., P. A. Collier, et al. (2008). "Measuring the effects of business intelligence systems: The relationship between business process and organizational performance." International Journal of Accounting Information Systems 9(3): 135-153. Eppler, M. J. and J. Mengis (2003). "A framework for information overload research in organizations." report, Universita Della Svizzera, Italiana. Eppler, M. J. and J. Mengis (2004). "The concept of information overload: A review of literature from organization science, accounting, marketing, MIS, and related disciplines." The Information Society 20(5): 325-344. Fisher, C. W., I. S. Chengalur-Smith, et al. (2003). "The impact of experience and time on the use of data quality information in decision making." Information Systems Research 14(2): 170. Flyvbjerg, B. (2006). "Five misunderstandings about case-study research." Qualitative inquiry 12(2): 219. Gangadharan, G. R. and S. N. Swami (2004). Business intelligence systems: design and implementation strategies. Ge, M. (2009). "Information quality assessment and effects on inventory decision-making."

C

MSc thesis M.J. van Strien

References

Gigerenzer, G. and D. G. Goldstein (1996). "Reasoning the fast and frugal way: Models of bounded rationality." Psychological review 103(4): 650-669. Gilad, T. and B. Gilad (1986). "Business Intelligence The Quiet Revolution." Sloan Management Review 27(4): 53-60. Golfarelli, M., S. Rizzi, et al. (2004). Beyond data warehousing: what's next in business intelligence?, ACM. Gorden, R. L. (1980). "Interviewing: Strategy, techniques, and tactics." Homewood, IL: Dorsey: pp. 49-50. Gorry, G. A. and M. S. Scott-Morton (1979). "Decision support systems: emerging tools for planning." Report (Alfred P. Sloan School of Management. Center for Information Systems Research); no. 44. Grünig, R. and R. Kühn (2005). Process-based strategic planning, Springer-Verlag New York Inc. Grünig, R. and R. Kühn (2009). Successful decision-making: a systematic approach to complex problems, Springer Verlag. Hahn, M., R. Lawson, et al. (1992). "The effects of time pressure and information load on decision quality." Psychology and Marketing 9(5): 365-378. Harrison, E. F. (1987). The managerial decision-making process, Boston [etc.]: Houghton Mifflin. Harrison, E. F. (1999). "The managerial decision-making process." Hartley, J. (2004). "Case study research." Essential guide to qualitative methods in organizational research: 323-333. Henry, W. A. (1980). "The effect of information-processing ability on processing accuracy." Journal of Consumer Research: 42-48. Herbig, P. and H. Kramer (1994). "The Effect of Information Overload on the Innovation ChoiceProcess." Journal of Consumer Marketing 11(2): 45-54.

D

BI systems to mitigate information load with store managers

Himma, K. (2007). "The concept of information overload: A preliminary step in understanding the nature of a harmful information-related condition." Ethics and Information Technology 9(4): 259-272. Hong, I. B. and D. R. Vogel (1991). "Data and Model Management in a Generalized MCDM-DSS*." Decision Sciences 22(1): 1-25. Huang, M. H. (2000). "Information load: its relationship to online exploratory and shopping behavior." International Journal of Information Management 20(5): 337-347. Huber, G. P. (2003). The necessary nature of future firms: Attributes of survivors in a changing world, Sage. Hwang, M. I. and J. W. Lin (1999). "Information dimension, information overload and decision quality." Journal of Information Science 25(3): 213. Inmon, W. H. (2009). Building the data warehouse, Wiley-India. Iselin, E. R. (1988). "The effects of information load and information diversity on decision quality in a structured decision task* 1." Accounting, Organizations and Society 13(2): 147-164. Iselin, E. R. (1993). "The effects of the information and data properties of financial ratios and statements on managerial decision quality." Journal of Business Finance & Accounting 20(2): 249-266. Iselin, E. R. (1996). "Accounting information and the quality of financial managerial decisions." Journal of Information Science 22(2): 147. Ives, B., M. H. Olson, et al. (1983). "The measurement of user information satisfaction." Communications of the ACM 26(10): 785-793. Jacoby, J. (1984). "Perspectives on information overload." Journal of Consumer Research 10(4): 432-435. Janis, I. L. and L. Mann (1977). Decision making: A psychological analysis of conflict, choice, and commitment, Free Press New York. Jarvenpaa, S. L. (1989). "The effect of task demands and graphical format on information processing strategies." Management Science 35(3): 285-303. E

MSc thesis M.J. van Strien

References

Johnson, E. J. and J. W. Payne (1985). "Effort and accuracy in choice." Management Science: 395-414. Kahneman, D., P. Slovic, et al. (1982). Judgment under uncertainty: Heuristics and biases, Cambridge Univ Pr. Kaplan, B. and D. Duchon (1988). "Combining qualitative and quantitative methods in information systems research: a case study." MIS quarterly: 571-586. Kaplan, B. and J. Maxwell (1994). "Qualitative research methods for evaluating computer information systems." Evaluating the Organizational Impact of Healthcare Information Systems: 30-55. Keller, K. L. and R. Staelin (1987). "Effects of quality and quantity of information on decision effectiveness." Journal of Consumer Research: 200-213. Klausegger, C., R. R. Sinkovics, et al. (2007). "Information overload: a cross-national investigation of influence factors and effects." Marketing Intelligence & Planning 25(7): 691-718. Königer, P. and K. Janowitz (1995). "Drowning in information, but thirsty for knowledge." International Journal of Information Management 15(1): 5-16. Krippendorff, K. (2004). Content analysis: An introduction to its methodology, Sage Publications, Inc. Kumar, R. (2005). Research methodology: A step-by-step guide for beginners, Sage Publications Ltd. Lee, A. S. (1991). "Integrating positivist and interpretive approaches to organizational research." Organization science 2(4): 342-365. Leech, B. (2003). "Asking questions: techniques for semistructured interviews." PS: Political Science and Politics 35(04): 665-668. Levy, M. and B. A. Weitz (2007). Retailing management, Irwin. Lönnqvist, A. and V. Pirttimäki (2006). "The measurement of business intelligence." Information Systems Management 23(1): 32-40. F

BI systems to mitigate information load with store managers

Lurie, N. H. (2004). "Decision making in information-rich environments: The role of information structure." Journal of Consumer Research 30(4): 473-486. Maimbo, H. and G. Pervan (2005). Designing a case study protocol for application in IS research. March, S. and A. Hevner (2007). "Integrated decision support systems: A data warehousing perspective." Decision Support Systems 43(3): 1031-1043. Martinsons, M. and R. Davison (2007). "Strategic decision making and support systems: Comparing American, Japanese and Chinese management." Decision Support Systems 43(1): 284-300. Mascha, M. F. and G. Smedley (2007). "Can computerized decision aids do." International Journal of Accounting Information Systems 8(2): 73-91. Mayring, P. (2000). Qualitative content analysis, Citeseer. Miles, M. and A. Huberman (1994). Qualitative data analysis: An expanded sourcebook, SAGE publications, Inc. Miller, G. A. (1956). "The magic number 7, plus or minus 2: some limits on our capacity for processing information." Psychological Review 63: 81-97. Mitroff, II and H. A. Linstone (1993). The unbounded mind, Oxford University Press New York. Moss, L. and S. Atre (2003). Business intelligence roadmap: the complete project lifecycle for decision-support applications, Addison-Wesley Professional. Myers, M. D. (1997, last modified: May 13, 2010 ). "Qualitative Research in Information Systems." MIS Quarterly from http://www.misq.org/discovery/MISQD_isworld. Negash, S. and P. Gray (2008). "Business intelligence." Handbook on Decision Support Systems 2: 175-193. Newell, A. and H. A. Simon (1972). Human problem solving, Prentice-Hall Englewood Cliffs, NJ. G

MSc thesis M.J. van Strien

References

Olson, B. J. and S. Parayitam (2007). "Strategic decision making: The effects of cognitive diversity, conflict, and trust on decision outcomes." Journal of Management 33(2): 196. Olszak, C. M. and E. Ziemba (2006). "Business Intelligence Systems in the Holistic Infrastructure Development Supporting Decision-Making in Organisations." Interdisciplinary Journal of Information, Knowledge, and Management 1: 57-68. Paquette, L. and T. Kida (1988). "The effect of decision strategy and task complexity on decision performance." Organizational Behavior and Human Decision Processes 41(1): 128-142. Park, Y. T. (2006). "An empirical investigation of the effects of data warehousing on decision performance." Information & Management 43(1): 51-61. Patton, M. (2002). Qualitative research and evaluation methods, Sage Publications, Inc. Paul, S. and D. Nazareth (2010). "Input information complexity, perceived time pressure, and information processing in GSS-based work groups: An experimental investigation using a decision schema to alleviate information overload conditions." Decision Support Systems. Payne, J. W. (1976). "Task complexity and contingent processing in decision making: An information search and protocol analysis* 1." Organizational behavior and human performance 16(2): 366-387. Payne, J. W., J. R. Bettman, et al. (1993). The adaptive decision maker, Cambridge Univ Pr. Raghunathan, S. (1999). "Impact of information quality and decision-maker quality on decision quality: a theoretical model and simulation analysis." Decision Support Systems 26(4): 275-286. Rouse, W. and S. Rouse (1984). "Human information seeking and design of information systems." Information Processing & Management 20(1-2): 129-138. Russell, J. A. and A. Mehrabian (1974). An approach to environmental psychology, MIT Press. Rutkowski, A.-F. and C. S. Saunders (2010). "Growing Pains with Information Overload." Computer June: pp. 96, 94-95.

H

BI systems to mitigate information load with store managers

Saaty, T. L. and M. S. Ozdemir (2003). "Why the magic number seven plus or minus two." Mathematical and Computer Modelling 38(3-4): 233-244. Schick, A. G. and L. A. Gordon Susan (1990). "Information overload: A temporal approach* 1." Accounting, Organizations and Society 15(3): 199-220. Schneider, S. (1987a). "Information overload: Causes and consequences." Human Systems Management 7(2): 143-153. Schneider, S. C. (1987b). "Information overload: Causes and consequences." Human Systems Management 7(2): 143-153. Schroder, H. M., M. J. Driver, et al. (1967). Human information processing, New York: Holt, Rinehart & Winston. Sekaran, U. and R. Bougie (2010). Research methods for business; a skill building approach, Wiley India Pvt. Ltd. Shannon, C. E. and W. Weaver (1949). "The mathematical theory of information." Urbana: University of Illinois Press 97. Shields, M. D. (1983). "Effects of information supply and demand on judgment accuracy: evidence from corporate managers." Accounting Review 58(2): 284-303. Shim, J. P., M. Warkentin, et al. (2002). "Past, present, and future of decision support technology* 1." Decision Support Systems 33(2): 111-126. Simnett, R. (1996). "The effect of information selection, information processing and task complexity on predictive accuracy of auditors* 1." Accounting, Organizations and Society 21(78): 699-719. Simon, H. (1997). Administrative behavior: A study of decision-making processes in administrative organizations, Free Pr. Simon, H. A. (1955). "A behavioral model of rational choice." The quarterly journal of economics 69(1): 99-118. Simon, H. A. (1957). Models of man: social and rational: mathematical essays on rational human behavior in a social setting, Wiley New York. I

MSc thesis M.J. van Strien

References

Simon, H. A. (1977). The new science of management decision, Prentice Hall PTR Upper Saddle River, NJ, USA. Sparrow, P. (1998). Information Overload. Legge, K., Clegg, C., Walsh, S.(Eds.) The experience of managing: a skills workbook, London: MacMillian. Sparrow, P. (1999). "Strategy and cognition: Understanding the role of management knowledge structures, organizational memory and information overload." Creativity and innovation management 8(2): 140-148. Starbuck, W. (1983). "Organizations as action generators." American Sociological Review 48(1): 91-102. Staw, B. M. and Z. Shapira (1997). Organizational decision making, Cambridge University Press New York. Stemler, S. (2001a). "An overview of content analysis." Practical assessment, research & evaluation 7(17): 137-146. Stemler, S. (2001b). "An overview of content analysis. Practical Assessmen." Research & Evaluation 7(17) Retrieved 19-09, 2010, from ttp://PAREonline.net/getvn.asp?v=7&n=17. Stigler, G. J. (1961). "The economics of information." The Journal of Political Economy 69(3). Streufert, S. C. (1973). "Effects of information relevance on decision making in complex environments." Memory and Cognition: pp. 224-228. Svenson, O. (1979). "Process descriptions of decision making* 1." Organizational behavior and human performance 23(1): 86-112. Taylor, R. N. (1975). "Age and experience as determinants of managerial information processing and decision making performance." Academy of Management Journal 18(1): 74-81. Thierauf, R. J. (2001). Effective business intelligence systems, Greenwood Publishing Group. Tipuric , D. and J. Prester (2003). "Decision Making Under Risk and Uncertainty: Risk Propensity of Croatian Managers." Business & Economics Society International 2003. J

BI systems to mitigate information load with store managers

Todd, P. and I. Benbasat (1991). "An experimental investigation of the impact of computer based decision aids on decision making strategies." Information Systems Research 2(2): 87-115. Todd, P. and I. Benbasat (1992). "The use of information in decision making: an experimental investigation of the impact of computer-based decision aids." MIS quarterly: 373393. Todd, P. and I. Benbasat (2000). "Inducing compensatory information processing through decision aids that facilitate effort reduction: an experimental assessment." Journal of Behavioral Decision Making 13(1): 91-106. Tremblay, M. C., R. Fuller, et al. (2007). "Doing more with more information: Changing healthcare planning with OLAP tools." Decision Support Systems 43(4): 1305-1320. Trochim, W. (1985). "Pattern matching, validity, and conceptualization in program evaluation." Evaluation Review 9(5): 575. Trusov, M., A. V. Bodapati, et al. (2006). "Retailer promotion planning: Improving forecast accuracy and interpretability." Journal of Interactive Marketing 20(3-4): 71-81. Turban, E., J. E. Aronson, et al. (2007). Decision support and business intelligence systems, Pearson Prentice Hall. Tushman, M. L. and D. A. Nadler (1978). "Information processing as an integrating concept in organizational design." Academy of management review 3(3): 613-624. Van Donselaar, K., T. Van Woensel, et al. (2006). "Inventory control of perishables in supermarkets." International Journal of Production Economics 104(2): 462-472. van Strien, M. J. (2010). Interview with Liesbeth Snabel. Amersfoort. Vassiliadis, P., A. Simitsis, et al. (2002). Conceptual modeling for ETL processes, ACM. Vessey, I. (1994). "The effect of information presentation on decision making: A cost-benefit analysis." Information & Management 27(2): 103-119. Waller, M., M. E. Johnson, et al. (1999). "Vendor-managed inventory in the retail supply chain." Journal of business logistics 20: 183-204. K

MSc thesis M.J. van Strien

References

Watson, H. J., B. H. Wixom, et al. (2006). "Real-time business intelligence: Best practices at Continental Airlines." Information Systems Management 23(1): 7-18. Williams, M. L., A. R. Dennis, et al. (2007). "The impact of DSS use and information load on errors and decision quality." European Journal of Operational Research 176(1): 468-481. Wilson, T. D. (1999). "Models in information behaviour research." Journal of documentation 55: 249-270. Wohlin, C., P. Runeson, et al. (2000). Experimentation in software engineering: an introduction, Springer Netherlands. Woods, D. D., E. S. Patterson, et al. (2002). "Can we ever escape from data overload? A cognitive systems diagnosis." Cognition, Technology & Work 4(1): 22-36. Woods, D. D., E. S. Patterson, et al. (1998). "Can We Ever Escape from Data Overload? A Cognitive Systems Diagnosis." Woodside, A. G. (2010). Principles for Case Study Research. http://gfa2010.univie.ac.at/fileadmin/user_upload/GFKconference/GFA2010/documents/jbr/JBRexample.pdf. Yin, R. K. (1989). Case study research, Sage Publ. Zaman, M. (2005). "Business Intelligence: Its Ins and Outs." Retrieved 19-04, 2010. Zunde, P. (1978). On information value.

L

Appendixes

Appendix I: Research strategy In all academic research, researchers need a strategy that supports them in the development and conduction of the research (Maimbo and Pervan, 2005). Research involves using valid and reliable procedures, methods and techniques, which are performed in a careful, systematic and patient manner. Generally, research is considered as a way of thinking, finding answers, and creating new knowledge (Kumar, 2005). Drawing on this definition, the author argues that research should be controlled, rigorous, systematic, empirical, valid and verifiable, and critical. This chapter will elaborate and justify the methodology used to conduct this research. Following Yin (1989), we refer to this as the research design – which represents a logical set of statements. Davis (2005; p.134) claims that “the formal specification of a research design is an integral part of the research”.

Research purpose Following Kumar (Kumar, 2005) we argue that the most suitable research method is contingent upon the research purpose. Therefore we first globally elaborate on the objective of this research. The objective is defined as “to present guidelines on how Business Intelligence (BI) systems can be employed in order to alleviate information overload”. The research will be conducted within the context of managerial decision-making activities of retail organizations‟ store managers. Since information overload is theorized to decrease decision quality the present research intends to provide guidelines that allow BI systems to increase store managers‟ decision-making quality.

When decomposing the research purpose, we will: (1) investigate the concept of information overload at store manager‟s, and (2) its relationship with the quality of their decisions, (3) investigate how BI systems can be deployed to alleviate information overload, and support store managers to incorporate all relevant information into their decisions, and (4) present recommendations on how information systems (IS‟s) designers might employ BI systems to fit decision makers‟ information-processing capacities.

I

MSc thesis M.J. van Strien

Whether a study‟s purpose is exploratory, descriptive, or hypothesis testing, depends on the stage to which knowledge about the research topic has advanced (Sekaran and Bougie, 2010). The concept of information overload already has a broad theoretical basis (Eppler and Mengis, 2004), hence we classify the present study as descriptive (i.e. explanatory) for the specific context.

Literature study The literature review is the core of the research process. We argue that the purpose of the literature study is twofold; we use it for (1) theory building, and (2) to gain a better understanding of the relevant theories and topics of investigation. For the prior purpose, we conducted a preliminary literature survey to verify the research relevance, define a clear problem statement, establish a conceptual framework, and to make sure that all the important variables for the research situation are investigated (Sekaran and Bougie, 2010). The literature study that addresses the latter purpose will introduce important concepts, and present relevant research work don e on the topic (i.e. answer research questions). More specific, it will provide a clear view on organizational decision making – as this is needed to research decision making behaviour in context (Choo, 2006), provide a theoretical insight into the effects that information overload has on managerial decision making, and elaborate on BI systems as countermeasure for information overload. The literature study will be based on literature from the following sources: Tilburg University scientific databases, Tilburg University library, Google Scholar databases, and reports which are of specific interest for the research and its context.

Conceptual framework The conceptual model is a theoretically based framework of how we theorize the relationships among several factors (i.e. variables), which are considered to be important to the problem. Hence these variables are identified to be integral to the dynamics of the situation being investigated (Sekaran and Bougie, 2010). In case study research (CSR; see paragraph 0), the development of a conceptual model involves both a comprehensive literature study and preliminary interviews (Maimbo and Pervan, 2005). Variables are important in bringing clarity and specifity to the conceptualization of a research problem, and to the development of a research

II

Appendixes

instrument. The research instrument affects how the data can be analysed, what interpretations can be made, and what conclusions can be drawn (Kumar, 2005).

We find some disagreement on whether CSR should define a conceptual model and variables. Kaplan and Maxwell (1994) argue that the CSR does not predefine dependent and independent variables, and a conceptual model, nor is it applicable to hypothesis testing. On the other hand, Yin (1989) includes proposition building into CSR. Although the terms “propositions” and “hypotheses” should not be used synonymously, they are both statements of relationships – propositions are more abstract (Bacharach, 1989). The use of propositions implies establishing a conceptual model. Compromising, we adopt Flyvbjerg‟s (2006) view which states that CSR is useful for both generating and testing of statements of relationship, but is not limited to these research activities alone. The established conceptual framework itself in presented in chapter 3.

Empirical study To ensure the purposefulness (i.e. to add value to existing scientific work) we need to draw our conclusions on evidence from real-life data (Kumar, 2005). The purpose of the empirical study consists of three complementary parts: (1) to investigate the information overload issue within a store manager‟s context, (2) to examine the relationship between information overload and a store manager‟s decision quality, and (3) to investigate how BI can decrease a store manager‟s information overload.

Interpretive research It is difficult to study information behaviour from a decision-making perspective independent from a particular process or context (Rouse and Rouse, 1984). More specific regarding information overload research, Eppler and Mengis (2003; p.35) argue that prior research has often been “too detached from the specific overload contexts”. Regarding empirical research, the authors advocate the use of more context-rich, qualitative research methods in addition to the already used experiments and survey. We support this view as we find that information overload is a phenomenon that is dependent on the context of both decision processes (Brueggemann et al., 2008), and individual differences in information processing (Rutkowski and Saunders, 2010).

III

MSc thesis M.J. van Strien

In contrast with the positivist approach, the interpretive approach believes that there are multiple realities (i.e. subjective constructions) in the mind, and assume the access to reality is only through social constructions (e.g. language, consciousness, and shared meanings) (Myers, 1997). Therefore interpretive researchers apply research methods that aim on understanding phenomena within their context, and through the meanings that people assign to these phenomena (de Vries, 2005; Hartley, 2004; Lee, 1991; Yin, 1989). We argue that our research purpose requires a context-specific approach; hence we adopt a CSR strategy to investigate the information overload phenomenon in the context of store managers.

Case study research CSR applies an interpretive approach to research problems and focuses on contemporary phenomena within a real-life context. For this reason CSR does not require control of behavioural events (Yin, 1989). This enables to measure phenomena (e.g. interactions between IT and users) in their natural context, making CSR a popular strategy in information system (IS) research (de Vries, 2005).

Case studies typically combine data collection methods such as archives, interviews, questionnaires, and observations. The way the variables are measured determines whether a study is considered to be predominantly „qualitative‟ or „quantitative‟. CSR evidence may be qualitative, quantitative or both (Maimbo and Pervan, 2005; Hartley, 2004; Eisenhardt, 1989; Yin, 1989). Qualitative evidence allows to generate concrete, practical, and context-dependent knowledge rather than general, theoretic, context-independent knowledge. Flyydbjerd (2006) rejects the conventional view that the latter is by definition more valuable. Qualitative evidence requires the use of context-specific measuring instruments (Kaplan and Duchon, 1988). We argue that the use of context-specific measuring instruments is not restricted to qualitative evidence. Following Lee (1991), we advocate the integration of qualitative and quantitative research methods. Combining multiple research approaches includes both testability and context into the research design. The author argues that collecting different kinds of data (i.e. from different sources, and through different methods) increases the robustness of the results.

IV

Appendixes

This can be achieved because the findings can be strengthened through triangulation (Yin, 1989) and allows a richer contextual basis for analysing the results (Kaplan and Duchon, 1988)

Research validity A fundamental question regarding the results of the research is how valid the results are. In order to plan for adequate results (i.e. results should be valid for the population of interest), Wolin, Runeson and Höst (2000) note that validity should be concerned in the design phase already. Because research design represents a logical set of statements, we can determine the quality of the proposed design, this is done by logical tests (Yin, 1989). The author finds that four tests have been commonly used to determine the quality of an empirical research: (1) construct validity, (2) internal validity, (3) external validity, and (4) reliability, or conclusion validity. Table 9 Criteria for judging research design quality; (Yin 1989; p.40)

Quality Activities challenge Construct validity Identifying correct operational measures for the concepts and variables being studied Internal validity Establish a causal relationship in which conditions are believed to result in other conditions External validity Defining a domain to which a study‟s finding can be generalized Reliability Demonstrating that the operations (e.g. data collection) can be repeated with the same results

Important in phase Data collection

research

Data analysis

Research design Data collection

Adequate validity does not necessarily imply most general validity. A research conducted within an organization (i.e. applied research) may be designed to answer some specific answers for that organization exclusively. Hence Wolin, Runeson et al. (2000) argue that it is sufficient when the results are valid within that specific organization. If the purpose is to draw more general conclusions, the validity must cover a more general scope as well. This implies that the way in which the validity is addressed, is contingent upon the research purpose.

V

MSc thesis M.J. van Strien

In conducting applied research - as the present study - the researcher would ideally like to maximize both the internal and external validity of a study. However, there must be compromises in both forms of validity due to considerations of practicality (Davis, 2005). Concerning CSR, Woodside (2010) emphasizes that accuracy comes first, not generality. We emphasize designing internal validity to increase accuracy of the conclusions. A more elaborate view on how we addressed the four threats to research validity is presented in the case study strategy.

Case study strategy Additional to the research design, which is discussed in chapter 2, this chapter will discuss the specific design of the case study we conducted, focusing on the deliberate choices that were made during the research process. The case study research (CSR) is a research strategy that focuses on understanding the dynamics present within single settings. This strategy allows researchers for the development of, for example, descriptions, test theory, generate theory (Eisenhardt, 1989), or to test questions and issues by setting these in a contextual and often causal context (Yin, 1989). We apply case study research to gain a thorough insight into the information overload concept in the context of retail store managers. Drawing on our conceptual model, this case study is conducted to (1) investigate the theoretical determinants for information overload (i.e. decisionmaker experience, time pressure, and information amount), and (2) examine the relationship between information load and decision quality.

Even though CSR has been found a suitable research strategy for the present research (see paragraph 2.4), there are some problems that researchers may encounter during the execution of CSR. These problems include, amongst others, the designing and scoping of the research to properly address the research question. Also the extensive ways of collecting data may easily result in an overwhelming amount of data (Yin, 1989). Therefore, the author calls for a structure that can be used to govern a CSR – he refers to this set of guidelines as the case study protocol (CSP). Maimbo and Pervan (2005) argue that the CSP outlines the procedures and rules governing the research. Defining clear guidelines ensures uniformity in data collection and analysis (Yin, 1989). The author argues that the use of a CSP is crucial to the proper execution of a case research.

VI

Appendixes

On the other hand Maimbo and Pervan (2005) note that a CSP is particularly useful when the research involves multiple researchers. For this reason we argue the need for a CSP, but acknowledge the relevant importance in the context of this research. Hence we distill a number of components from the CSP that are considered to be important for this research; unit of analysis (to address scoping), data gathering (to prevent from overwhelming amounts of data), and data analysis (to allow for accurate and valid conclusions). These components are presented in this chapter.

Unit of analysis CSR can include single- or multiple-case studies. For the present research we argue that information processing behaviour from a decision-making perspective, and more specific information overload, are hard to investigate independent from a particular context (Eppler and Mengis, 2004; Rouse and Rouse, 1984). Additionally, Eppler and Mengis (2004) argue that future information overload research should encompass an extensive contextual investigation, rather than yet another survey or experiment. To be able to conduct a detailed contextual case study, and include the user‟s and designer‟s point of view, we choose a single-case study strategy. In this design we follow Woodside (2010), who argues that in CSR accuracy (i.e. internal validity) comes first, not generality (i.e. external validity). However, because each research needs to address both internal and external validity, we include a comparison case to generate a more broad view from the designer‟s perspective. Following this approach, our respondent matrix is as presented in table. As noted in chapter 2 we focus our research on supermarkets. In this context we will refer to the store manager as the user, and the central IT designer as the designer. Table 10 Respondent matrix

Case 1 User Designer

Case 2 7 1

1

Data gathering This paragraph will address the challenge of how we will collect the data in order to measure the variables in the framework. We have argued that our conceptual framework has primarily

VII

MSc thesis M.J. van Strien

determined what we measured (Sekaran and Bougie, 2010). However, Myers (1997) claims that in qualitative research the analysis process also determines which data is gathered. The author argues that the questions posed to subjects largely determine what researchers are going to find out. The iterative analysis affects the data and the data affect the analysis in significant ways. Miles and Huberman (1994) support this view as they argue that data analysis is interactive with data gathering. This interplay affects both our gathering and analysis processes (see 7.3.2).

Throughout the collection process, a major objective is to collect data about actual human events and behaviour – captured in perceptions, attitudes, and facts (Hartley, 2004; Yin, 1989). Case study evidence may come from various sources; e.g. documents, archival records, interviews, direct observation, questionnaires, and physical artifacts. The variety in data types raises the challenge of restricting the amount of data, but yet gather sufficient evidence to generate accurate and reliable conclusions. No single source has a complete advantage over all the other sources, in fact they are considered to be highly complementary. The opportunity to use multiple data sources is considered both a major strength, and threat of case study data collection. In addition to the need to be familiar with the data gathering processes, we also need to continue addressing the case study‟s design quality challenges. As noted in paragraph 2.5, reliability and construct validity are relevant design challenges at the data collection phase.

Reliability A case study‟s finding or conclusion is likely to be more reliable when it was based on multiple data sources (Yin, 1989). Therefore, a good case study will use multiple sources to foster the data collection process. This allows for triangulationt, which is the cross-validation achieved when different kinds and sources of data converge and are found congruent (Kaplan and Maxwell, 1994). Triangulation is highly recommended by many researchers as a mechanism to improve the reliability of qualitative research. (Maimbo and Pervan, 2005; Hartley, 2004; Stemler, 2001a). Kaplan and Duchon (1988) note that, when using qualitative and quantitative data, triangualtion is recommended in order to increase the robustness of the results. Triangulation can be achieved through four distinct aspects, i.e. triangulation (1) of data sources, (2) among different evaluators, (3) of perspectives to the same data set, and (4) of methods

VIII

Appendixes

(Patton, 2002). The present discussion concerns the first type. For triangulation through multiple data sources, Yin (1989) distinguishes two conditions; when the data is really triangulated, and when the different sources address different facts. A clear distinction between the conditions is depicted in Figure 3. This figure shows that triangulation is achieved in the analysis process. We argue that gathering data from different sources allows us to achieve triangulation during the analysis, hence improve reliability.

Figure 3 Convergence and non-convergence; from Yin (1989)

Construct validity Construct validity pertains how we measure the variables and operationalize the constructs of the conceptual model (Sekaran and Bougie, 2010), hence it addresses the relationship between theory and measuring of actual behaviour (Wohlin et al., 2000). Put differently, construct validity deals with the degree to which the gathering process (effectively) allows to measure what is intended to measure. The construct validity challenge is also addressed by triangulation, because multiple sources of evidence provide multiple measures for the same phenomenon – increasing the chance that we have measured what was intended to measure (Yin, 1989).

Measurement of the variables is important because it makes the hypotheses testable. Following Sekaran and Bougie (2010), we distinct two types of variables; (1) objective variables which can be measured precise, and (2) subjective variables which can‟t be measured accurately as they often reflect an opinion. Since this research uses case study approach, we will primarily deal with subjective measures (Lee, 1991).

IX

MSc thesis M.J. van Strien

Generally, subjective variables are measured by (1) breaking them down in observable characteristic behaviour, or (2) by operationlizing (i.e. reduction of abstract and subjective concepts by measuring them in a tangible way). Operationalizing is done by looking at the concept‟s dimensions, facets or properties, and translating those into observable and measurable elements. Operationally defining the concepts describes its observable characteristics to measure the concept (Sekaran and Bougie, 2010). Bacharach (1989) refers to these variables as observed units. In order to translate our variables into observable observed units we have conducted an extensive literature research to replicate and extend scales, measures and instruments from previously tested work. The results of this study are presented in Appendix II: Measurement validity.

Sources of evidence As shown above, the literature study was the basis for the data gathering process. The study has resulted in a protocol for semi-structured interviews, with both qualitative (contextual) information and qualitative questionnaires. Additionally we asked the subjects to perform the inventory management decision-making process, this allowed us to conduct participant observations on the subjects. A brief description of the three types of evidence will be presented below. A more elaborate justification for the data gathering, together with the interview protocol and questionnaires can be found in Appendix III and Appendix IV.

Semi-structured interviews allowed us to collect detail, depth, and the respondents perspective, while also allowing hypotheses testing and qualitative analysis through additional questionnaires (Leech, 2003). Since a part of the interview was about decision quality, we acknowledged the need to put the respondents at ease. Following Leech (2003) we achieved this through question ordering, avoid “presuming questions”, and use grand tour and example questions (see interview protocol). Additionally, we asked the subjects to perform a order decision during the interview. The participant observation allowed us to gain a better insight in how the systems works, and how different users control the system. Also, following Gorden‟s (1980) scheduled-structured interview, we included an additional questionnaire to gain quantitative data. When considered necessary, the response bias was

X

Appendixes

controlled by arranging the adjective pairs, i.e. approximately half the time the adjective on the extreme left corresponded to judgments of higher information load. Due to access restrictions because of information confidentiality, we were unable to collect the data with which to objectively assess inventory management performance. As noted in Appendix II we therefore fall-back on the subjective measure of decision quality. Data analysis Data analysis consists of systematically analysing the data gathered in order to draw inferences and conclusions (Sekaran and Bougie, 2010). Yin (1989) argues that data analysis addresses the internal validity – especially for explanatory research. If the researcher incorrectly draws conclusions of a causal relationship between variables, e.g. because an external factor actually explains the concept, the research design has failed to address internal validity.

Yin (1989) defines data analysis as categorizing, examining, tabulating, testing, or otherwise recombining data in order to draw empirically based conclusions. The author notes that, for case study research especially, analyzing the evidence is difficult because the techniques are one of the least developed aspects of CSR and still have not been well defined. To overcome this, the author argues that case study analysis should follow a general analysis strategy - supporting the researcher to define priorities for what to analyze. Moreover, the author notes that a strategy will help the researcher to treat the evidence fairly, and rule out alternative interpretations – hence increase internal validity. A decent starting point for the analysis can be to put the information into different arrays, make a matrix of categories and assign evidence to the categories, tabulating the frequency of events, and calculating the relationships between categories (Miles and Huberman, 1994). We use these manipulations for data reduction, and to put the evidence in a preliminary order.

Analysis strategy Following Yin (1989) we apply a general analysis strategy. The author discusses four general strategies: (1) relying on theoretical propositions, (2) developing a case description, (3) using qualitative and quantitative data, and (4) examining rival explanations. Yin (1989) notes that “relying on theoretical propositions” is the most preferred strategy. This strategy follows the

XI

MSc thesis M.J. van Strien

theoretical propositions that have led to the case study research. As noted in the prior paragraph, our theoretical propositions have shaped our data collection process. Considering the structure of our data gathering process, we argue that relying on theoretical propositions is the most convenient analysis strategy. This strategy uses pattern-matching logic, which attempts to determine the extend to which the observations correspond to, or fit the conceptual model (Trochim, 1985). If the patterns coincide, the results can help a case study to strengthen its internal validity. For explanatory case studies - like the present research - the patterns are related to the dependent and the independent variables of the study (Yin, 1989).

Content analysis The strategy supported us in properly structuring the data for analysis purpose. Thereafter we used content analysis as a technique to conduct the analysis. Krippendorf (2004; p.25) defines content analysis as "a research technique for making replicable and valid references from data to their contexts". The technique entails a systematic reading of a body of texts, images, and symbolic matter. The essential idea is that words and/or signs can be assigned to primary conceptual categories (i.e. categorization). These categories represent important aspects of the theory to be tested (Myers, 1997). Hereafter, the material is to be analyzed step by step, following rules of procedure. Mayring (2000) acknowledges two central procedures in categorization: (1) category development, and (2) category application. The first procedure addresses the challenge of defining categories for the analysis. Stemler (2001b) distinguishes two approach in category development. A priori development derives categories from theoretical background, and emergent development in which the categories are established following some preliminary analysis of the data. We argue that, to align with our analysis strategy we apply an a priori approach. However, we also apply interaction between the gathering and analysis processes – implying an emergent development. Interaction between the gathering and analysis processes has especially drawn our attention – and data collection efforts – towards decision responsibility.

The second procedure of categorization (category application) aims to connect the case study evidence with the categories. The main idea is to provide explicit definitions for each category,

XII

Appendixes

dividing the material into content analytical units (Mayring, 2000). We use quotes derived from interviews to connect to the categories. The results are presented in Appendix VI: Content analysis.

Hereafter, the material is to be analyzed step by step, following rules of procedure. The researcher searches for structures and patterned regularities in the text and makes inferences on the basis of these regularities. The importance of an idea is revealed in the frequency with which it appears in the text (Myers, 1997). Concluding, content analysis allows for “empirical, methodological controlled analysis of texts within their context of communication, following content analytical rules and step by step models” (Mayring, 2000; p.2).

XIII

MSc thesis M.J. van Strien

Appendix II: Measurement validity This chapter discusses how we measure the variables and operationalize the constructs of the conceptual model. Measurement of the variables is important because it makes the hypotheses testable. Following Sekaran and Bougie (2010), we distinct two types of variables; (1) objective variables which can be measured precise, and (2) subjective variables which can‟t be measured accurately as they often reflect an opinion. Since this research adapts an interpretive or quantitative approach, we will primarily deal with subjective measures (Lee, 1991). Generally, subjective variables are measured by (1) breaking them down in observable characteristic behaviour, or (2) by operationlizing (i.e. reduction of abstract and subjective concepts by measuring them in a tangible way). Operationalizing is done by looking at the concept‟s dimensions, facets or properties, and translating those into observable and measurable elements. Operationally defining the concepts does not consist of delineating the reasons, antecedents, consequences or correlates, but describes its observable characteristics to measure the concept (Sekaran and Bougie, 2010). Bacharach (1989) refers to these variables - which have been operationalized empirically by measures – as observed units. In order to translate the variables into observable and measurable characteristics we have conducted an extensive literature research to replicate and – if necessary – extend scales, measures and instruments from previously tested work. The results of the replication and extension are presented below.

Decision quality Various researchers have investigated decision quality. The problem with decision quality is that we need to determine the best decision in order to assess whether the extent of correctness of the decision made (Keller and Staelin, 1987). Existing research has not developed a reliable measure of decision quality. Rather, decision performance has been measured indirectly via time taken to make a decision (Paquette and Kida, 1988), speed and accuracy, user satisfaction, decision quality, changes in heuristic bias, and shift in decision making toward a system‟s recommendation, – none of which is necessarily an accurate representation of decision quality (Chan, Sutton and Yao, 2009). Different from various researchers who examined decision quality, but actually measured a manager‟s perceived decision quality (Raghunathan, 1999), we

XIV

Appendixes

propose a method of measuring decision quality by assessing inventory performance. The proposed method is addressed in chapter 2

Inventory performance Inventory performance has previously been measured using a number of inventory ratios; inventory turnover (Cannon, 2008), inventory to sales ratio, and inventory to assets ratio (Ge, 2009). Ge (2009) argues that the three measures are highly correlated. Additionally he concludes that their use depends heavily on data (e.g. inventory and financial) availability and the objective of the research. Following the author, we propose that “inventory to sales ratio” is the best measure to assess the overall efficiency in inventory management. However, as Ge notes, application of this measure is dependent upon the availability of data. Since we have no guarantees about the availability and permission to use of both inventory and financial data, we advocate the need for a back-up measure.

If we cannot operationalize inventory performance (i.e. measure through ratios), we need to break inventory performance down into observable characteristics (Sekaran and Bougie, 2010). In order to do so we use Ge‟s (2009) research on the impact of information quality on inventory management decision quality. The author argues that the best inventory management decisions can be established using two inventory management policies: (1) Just in Time (JIT), and (2) Economic Order Quantity (EOQ). The aim of JIT is to eliminate the need for holding inventory items by emphasizing on order time and quantity. EOQ focuses on minimising total costs instead of minimising inventories, emphasizing the order quantity. Ge claims that these measures “can be used to measure decision quality” (Ge, 2009; pg. 76). We argue that, in order to break inventory performance down into observable characteristics, order quantity and time together constitute a measure for inventory performance.

Decision-maker experience Decision maker experience has previously been assessed by measuring age (Taylor, 1975), years of (management) experience (Ge, 2009; Fisher et al., 2003; Taylor, 1975), domain-specific knowledge (Fisher et al., 2003), and subjective experience. Fischer et al. (2003) categorized

XV

MSc thesis M.J. van Strien

experience into either novice or expert, where they considered a decision maker an expert when there is equal to or more than ten years of experience. Taylor (1975) has found high correlation between age and years of management experience. For this reason we will measure decision maker experience by years of management experience. Following Fisher et al., we relate this experience to a certain domain. Instead of applying a nominal scale to select mutually exclusive domains (categories), we will instruct the subject to restrict years of management experience to the retail domain.

Time-pressure Researchers typically study decision making without time constraints (Payne et al., 1993). Some researchers (Ahituv, Igbaria and Sella, 1998) have studied “time pressure” but measured “time pressure” by simply allocating specific time to perform a task. We distinguish between time constraints and time pressure. A time constraint is a specific allotment of time for making a decision, while time pressure is a subjective reaction to the amount of time allotted. Time pressure is experienced whenever the time available for the completion of a task is perceived as being shorter than normally required for the activity (Svenson and Edland 1987). Some people may feel pressure in a long time constraint while others may not feel time pressure in a short time constraint (Fisher et al., 2003).

Information amount To gain a more concrete view on the information amount of specific information categories, we present a simple nominal scale for indicating the amount of information (none vs. overload) per information category – see Table 11. The information categories (e.g. stock level reporting, sales forecast reporting, and IT-system support) have been determined during preliminary research and are verified during the interviews. Subjects have options to supply complementary information categories during the interview. Also, they are asked to indicate the amount of times they receive the information, in a nominal scale (too less vs. too often) - see Table 11. The scales contain a subjective measure to assess the specific information load for inventory management related information categories. However, when the subject indicates the perceived information amount, he or she also indicates whether this information is disseminated. Hence, the

XVI

Appendixes

scale in Error! Reference source not found.Table 11. This table also contains a basic objective easure for information amount. Additionally we ask the subject to indicate the time unit in which the information is disseminated (daily, weekly, or monthly). Concluding, the subject is asked to indicate whether the information is gained through a push or pull mechanism. This scale is not presented in the table, but was included in the questionnaire that was presented during the interviews. Table 11 Scale for received information amount

Information category Reports Stock level Sales forecasts Product pricing Sales Financial performance Steering reports Benchmarks IT system support Supply information Promotion/sales Trends Customers Colleagues Research News Local Nationwide ……………

None

Too few

Sufficient

Too much

Overload

□ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

Table 12 Scale for time indication of information load

Information category Reports Stock level Sales forecasts Product pricing Sales Financial performance Steering reports Benchmarks

Daily

Weekly

Monthly

Too less

Good

Too often

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

□ □

□ □

□ □

□ □

□ □

□ □

XVII

MSc thesis M.J. van Strien

IT system support Supply information Promotion/sales Trends Customers Colleagues Research News Local Nationwide

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □ □

□ □

□ □

□ □

□ □

□ □

□ □

Information load Our primary measure for information load is Rutkowski and Saunders‟s (2010) cognitive and emotional overload scale. To gain a more concrete insight into information load we also use Russel and Mehrabian‟s (1974) General Measure of Information Rate, which includes 14 semantic differential items for assessing the information rate (i.e. information load (Huang, 2000) of a situation. Russel and Mehrabian‟s scale is based on a number of familiar distinctions that are made in characterizing environments or specific stimuli, which the authors integrate within the concept of information rate. The authors considered that an advantage of using subjects‟ own judgments, instead of other measures of information load, is that the former automatically discounts the effects of familiarity and meaningfulness. Various researchers (e.g. Hwang and Lin, 1999; Iselin, 1988) have identified two major subdimensions for information load: (1) information complexity, and (2) information novelty. Complexity refers to the number of different elements or features of information, which can be the result of increased information diversity. Novelty involves the unexpected, surprising, new, or unfamiliar aspects of the information element (Huang, 2000).

Additionally we integrate a part of the information reliability scale (Ives, Olson and Baroudi, 1983) as trust in the received information is an important factor for the perceived value of information (Klausegger et al., 2007). The authors claim that higher (lower) trust in the information increases (decreases) perceived value. Perceived value, or its perceived meaningfulness affects the information load (Russell and Mehrabian, 1974). Moreover, Mascha and Smedley (2007; p.74) note that “failing to incorporate the recommendations of a decision aid may well be the result of failure to regard those recommendations in the first place”. They argue XVIII

Appendixes

that this implies that reliance on a decision aid is desirable and may stimulate decision-makers to incorporate the information into their decision. We argue that information reliance affects pertinence and thus information load. An overview of the composed scale is shown in Table 13. Table 13 General information load measurement scale

Information complexity (+) simple (+) patterned (-) large scale (+) good form (+) varied (-) contrasting (-) dissonant (+) pertinent Information novelty (-) common (+) meaningful (-) novel (+) homogenous (-) surprising Information reliability (+) complete (-) unimportant (+) reliable (+) precise

□ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □

complex random small scale bad form redundant consistent consonant impertinent

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

□ □ □ □ □

rare meaningless familiar heterogeneous usual

□ □ □ □

□ □ □ □

□ □ □ □

□ □ □ □

□ □ □ □

incomplete important unreliable imprecise

Response bias was controlled by arranging the adjective pairs so that approximately half the time the adjective on the extreme left corresponded to judgments of higher information loaf and the reverse was true for the other half of the measures. The plus (+) and minus (-) signs preceding each item in table 1-1 indicate the direction of the scores. These signs, of course, were omitted from the scale which was presented to the subjects, as were the category labels.

Appendix III: Interview protocol store managers 1. U bent volledig verantwoordelijk voor de keuzes m.b.t. voorraadbeheer Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

XIX

MSc thesis M.J. van Strien

□

□

□

□

□

2. U wordt gestuurd bij het maken van de beslissingen Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

3. U voelt zich ervaren met het maken van keuzes voor voorraadbeheer Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

4. Hoe lang maakt u al beslissingen voor het voorraadbeheer? 7 jaar

□

□

□

□

□

5. Als u keuzes maakt, doet u dat het liefst op basis van: (ordenen van 1 tot 3) Informatie

Ervaring

Gevoel

□

□

□

6. Hoeveel tijd is er (meestal) beschikbaar voor het maken van deze beslissingen? < 1 uur 1-4 uur 4-8 uur 1-2 dagen >2 dagen

□

□

□

7. U ervaart een tijdsdruk als u beslissingen moet maken Sterk oneens Oneens Neutraal

□

□

□

□

□

Eens

Sterk eens

□

□

8. Het is altijd duidelijk welke beslissing u moet maken Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

9. De informatie die ik bij de beslissing gebruik is altijd hetzelfde Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

XX

Appendixes

10. U ervaart het maken van de beslissingen als lastig Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

11. Het is altijd duidelijk wat het resultaat van de beslissing moet zijn Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

XXI

MSc thesis M.J. van Strien

12. Welke informatie (m.b.t. voorraadbeheer) ontvangt u, en hoeveel? Type informatie

Niks

Te weinig

Genoeg

Te veel

Overlade n

Rapporten Voorraadniveau

□

□

□

□

□

Sales forecasts

□

□

□

□

□

□

□

□

□

□

Sales

□

□

□

□

□

Financiële prestatie

□

□

□

□

□

Sturingsrapporten

□

□

□

□

□

Benchmarks

□

□

□

□

□

Advies ICT-systemen

□

□

□

□

□

Informatie over levering

□

□

□

□

□

Promoties/aanbiedingen

□

□

□

□

□

Klanten

□

□

□

□

□

Collega’s

□

□

□

□

□

Onderzoeken

□

□

□

□

□

Plaatselijk

□

□

□

□

□

Landelijk

□

□

□

□

□

……………

□

□

□

□

□

……………

□

□

□

□

□

……………

□

□

□

□

□

……………

□

□

□

□

□

Prijzen

van

producten

Trends

Nieuws

13. Wat vindt u van de mate waarin u de informatie ontvangt?

XXII

Appendixes

Type informatie

Te

Goed

Te vaak

weinig Rapporten Voorraadniveau

□

□

□

Sales forecasts

□

□

□

Prijzen producten

□

□

□

Sales

□

□

□

Financiële

□

□

□

□

□

□

□

□

□

Advies ICT-systemen

□

□

□

Informatie

□

□

□

□

□

□

Klanten

□

□

□

Collega’s

□

□

□

Onderzoeken

□

□

□

Plaatselijk

□

□

□

Landelijk

□

□

□

……………

□

□

□

……………

□

□

□

……………

□

□

□

……………

□

□

□

prestatie Sturingsrapporten Benchmarks

over

levering Promoties/aanbiedin Trends

Nieuws

XXIII

MSc thesis M.J. van Strien

14. De informatie die u krijgt is..

Simpel Gestructureerd Veel Helder weergegeven Gevarieerd Consistent Onduidelijk Bruikbaar Vertrouwd Betekenisvol Nieuw Alledaags Compleet Belangrijk Betrouwbaar Precies

□ □ □ □

□ □ □ □

□ □ □ □

□ □ □ □

□ □ □ □

Complex

□ □ □ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □ □ □

□ □ □ □ □ □ □ □ □ □ □ □

Dubbelzinnig

Ongestructureerd Weinig Onduidelijk weergegeven

Contrasterend Duidelijk Onbruikbaar Onbekend Betekenisloos Altijd hetzelfde Zeldzaam Incompleet Onbelangrijk Onbetrouwbaar Onnauwkeurig

15. De informatie ondersteunt u bij het maken van de beslissing Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

16. (u kunt)… de hoeveelheid informatie.. Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Hanteren

□

□

□

□

□

Overweldigt me

□

□

□

□

□

Niet omgaan met

□

□

□

□

□

Zet

□

□

□

□

□

Niet verwerken

druk

XXIV

me

onder

Appendixes

Verwart me

□

□

□

□

□

Irriteert me

□

□

□

□

□

17. U vindt het moeilijk te bepalen wanneer er besteld moet worden Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

18. U vindt het lastig te bepalen hoeveel er besteld moet worden Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

19. Het komt regelmatig voor dat er teveel producten op voorraad zijn Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

20. Het komt regelmatig voor dat er producten uit voorraad zijn Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

21. Er worden regelmatig producten weggegooid omdat ze over zijn Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

22. De waarde van de voorraad is gemiddeld te hoog Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

23. De voorraad heeft in verhouding een negatieve impact op het bedrijfsresultaat Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

XXV

MSc thesis M.J. van Strien

□

□

□

□

□

24. U bent tevreden over uw beslissingen Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

25. Als het u het eens bent met stellingen 17 t/m 23. Waardoor komt dit volgens u? (rangschikken van 1-4) Te veel informatie

Verkeerde informatie

Verkeerde levering

Bewuste keuze

□

□

□

□

XXVI

Appendixes

Appendix IV: Interview protocol system designers Semi-structured Goal: Is information overload considered during the design of information systems that support supermarket managers in their decision making?

Questions Volgens de website: “Het hoofdkantoor in Amersfoort ondersteunt de ondernemers in hun dagelijkse bedrijfsvoering”

Contextual (for case study) 1. Welke rol speelt C1000 IT bij het ondersteunen van de supermarktmanagers (SMM‟s)? 2. Wat is het doel van de IT ondersteuning (decision support, management control, efficiency)? 3. Wat is het belang van goede IT ondersteuning voor de SMM‟s? 4. Wat is het doel van C1000 IT (informatisering vs. automatisering)? Huidige doelstelling vs. Toekomstige doelstelling? Welke impact heeft het STORE traject op de informatievoorziening? Wie worden als stakeholders beschouwd?

5. Hoe worden deze doelen nagestreefd? Missie/strategie Goverance? Wie is verantwoordelijk voor design/ontwikkeling/implementatie/support?

6. Welke systemen zijn er om de supermarktmanagers te ondersteunen? In hoeverre zijn deze systemen geïntegreerd? Zijn er hiaten in de informatievoorziening Zijn er prestatieproblemen in de IT-diensten (te laat beschikbaar, kwaliteit)? 7. In hoeverre wordt business intelligence gebruikt bij het ondersteunen van SMM‟s? Waarvoor en waarom daarvoor?

XXVII

MSc thesis M.J. van Strien

Voordelen en problemen? Ontwikkelingen op BI gebied?

8. Uit literatuur blijkt dat de supermarktbranche erg gevoelig is voor invloeden van buitenaf. Hoe gaan jullie hiermee om in de informatievoorziening? Integratie van de informatiebronnen is hierin belangrijk. In hoeverre gebeurt dit?

9. Worden de doelstellingen nu behaald? Tevredenheid stakeholders?

Optioneel 10. Overige trends/ontwikkelingen in IT ondersteuning voor supermarktmanagers? Hoe zorgen jullie dat deze ontwikkelingen operationeel worden?

Questionnaire Design of the IS Het informatiesysteem (IS) is ontworpen rondom de informatiebehoefte van Supermarktmanagers (SMM‟s) Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Het IS toont de informatie vanuit één perspectief Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

We hebben het aantal afgebeelde kolommen/soorten informatie gelimiteerd Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

We hebben het aantal afgebeelde rijen gelimiteerd Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

XXVIII

Appendixes

De verstrekte informatie laadt ruimte voor dubbelzinnige interpretatie Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

We streven ernaar om alle informatie via het IS te verspreiden Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Informatie uit het IS is ook via andere bronnen te verkrijgen Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Wij bepalen welke informatie belangrijk is Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

De informatie wordt in weloverwogen batches getoond i.p.v. alles ineens Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Relevante informatie wordt

Irrelevante informatie wordt bewust niet getoond Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

We gebruiken het liefst een „push‟ principe voor de IS‟en Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Het is de bedoeling dat de SMM – naast het IS - zelf nog denk- en rekenwerk uitvoert Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

XXIX

MSc thesis M.J. van Strien

□

□

□

□

□

Ik begrijp de situatie waarin de SMM‟s werken en besluiten nemen Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Implementation of the IS Alle gebruikers krijgen training bij het implementeren van een (nieuw) IS Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Tijdens de implementatie wordt het IS aangepast aan de eigenschappen van de winkel Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Tijdens de implementatie wordt het IS aangepast aan de eigenschappen van de manager Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

IS evaluation SMM‟s kunnen aanvraag doen voor aanvullende functionaliteit/ Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Aanvullende IS‟en worden alleen op aanvraag ontworpen Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

De gebruiker wordt gedwongen met de nieuwe functionaliteit te werken Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

SMM‟s kunnen functionaliteit uit bestaande IS‟en verwijderen

XXX

Appendixes

Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Altijd

Een aanvraag wordt ... geaccepteerd Nooit

Zelden

Regelmatig

Vaak

□

□

□

□

Als een nieuw IS een bestaande functionaliteit vervangt, wordt de bestaande IS aangepast Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

De informatie uit de systemen is te complex voor de SMM‟s Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

De SMM‟s voelen zich vertrouwd met de informatie uit de systemen Sterk oneens

Oneens

Neutraal

Eens

Sterk eens

□

□

□

□

□

Measure validity Like we did with the store manager questionnaire, we also conducted a literature research to determine valid measures for the system designer questionnaire. Because we did not apply variables for the designer measures, this literature study aimed at aspects in which the designers can incorporate the information overload concept into the system design. Scoring these aspects allows us to determine whether information overload is considered during the system design. Table 14 Aspects of information overload in the system design

#

Characteristic

1

Number of items of information

Related to (independent variable) Information load

2

Uncertainty of information (available vs needed information)

Information load/decision maker experience

Addressed in (Bawden, 2001; Hwang and Lin, 1999; Herbig and Kramer, 1994; Jacoby, 1984) (Schneider, 1987a; Tushman and Nadler,

XXXI

MSc thesis M.J. van Strien

3

Information load

4

Information diversity/number of alternatives Ambiguity of information

5

Information novelty

6

Information complexity

7

Intensity of information

8

Information quality

Information load/decision maker experience Information load/decision maker experience Information load/decision maker experience Information load

Overabundance of irrelevant information 10 Push system

Information load

9

Information load

11 Context-based design

Information load/Time pressure Information load

12 Various distribution channels

Information load

XXXII

1978) (Hwang and Lin, 1999; Iselin, 1988) (Sparrow, 1999; Schneider, 1987a) (Schneider, 1987a) (Schneider, 1987a) (Schneider, 1987a) (Sparrow, 1999; Sparrow, 1998) (Ackoff, 1967) (Bawden, 2001) (Woods, Patterson, Roth and Ohio State Univ, 1998) (Edmunds and Morris, 2000)

Appendixes

Appendix V: Quantitative evidence Store managers Table 15 Questionnaire data for independent variables, including frequency table

Autonomy

Nijmegen (s7)

Werkendam (s6)

Mierlo (s5)

Veenendaal (s4)

Hillegom (s3)

Table 17 for scale explanation

Zoetemeer (s2)

See

Tilburg (s1)

Frequency table

1 2 3 4 5

Autonomy (1)

5

5

5

5

5

5

5

- - - - 7

Steering (2)

4

4

4

2

4

2

4

- 2 - 5 -

Subjective (3)

5

4

5

5

5

5

5

- - - 1 6

Objective (4)

5

4

4

5

4

5

4

- - - 4 3

Time

Time avail. (6)

3

4

2

2

2

4

2

- 4 1 2 -

pressure

Subjective (7)

4

5

2

2

2

4

2

- 4 - 2 1

Task

Subjective (10)

3

2

3

2

3

2

3

- 3 4 - -

complexity

Goal (8)

4

4

4

4

4

4

4

- - - 7 -

4

2

3

4

4

4

3

- 1 2 4 -

Experience

Infor. Novelty (9)

XXXIII

MSc thesis M.J. van Strien

Table 22 Descriptive statistics for independent variables

Autonomy

Experience

Time pressure

Information Novelty

Goal

Subjective

Subjective

Time avail.

Objective

Subjective

Steering

Autonomy

Complexity

Median

5

4

5

5

2

2

3

4

4

Mode

5

4

5

5

2

2

3

4

4

Standard Deviation

0

1,07

0,38

0,38

0,95

1,29

0,53

0

0,79

Sample Variance

0

1,14

0,14

0,14

0,90

1,67

0,29

0

0,62

Count

7

7

7

7

7

7

7

7

7

XXXIV

Appendixes

Table 17 Scale explanation for independent variables

Variable

Conclusion

Subjective/objective

Autonomy

High score means the store manager considers himself Subjective autonomous on the decisions

Steering

High score means the store manager perceives a high level of Subjective steering for his decisions

Subjective

High score means that the store manager considers himself

Subjective

experience

experienced

Objective

High score means that store manager is working in domain for a Objective

experience

long time.

Subjective

High score means the store manager perceives a low time

time pressure

pressure

Time

High score means the store manager has a lot of time available. Objective

availability

(see appendix B for scales)

Subjective

High score means the store manager perceives the low decision Subjective

complexity

complexity

Goal

High goal score means the decision goal is structured

Subjective

Information

High score means that the information used is always the same

Subjective

High scores mean low perceived information amount (see table )

Subjective

Subjective

novelty Information amount Information

High scores mean the subject experiences information overload

load

XXXV

MSc thesis M.J. van Strien

Hillegom

Veenedaal

Nijmegen

Mierlo

Zoetermeer

5= too much

Tilburg

1= too little

Werkendam

Table 24 Questionnaire data for information amount; perceived quantity

Frequency table

Reports

1

2

3

4

5

Stock level

2

2

3

2

3

2

3

-

4

3

-

-

Sales forecasts

2

3

3

2

4

3

3

-

2

4

1

-

Product prices

4

3

3

4

4

4

4

-

-

2

5

-

Actual Sales

3

3

3

3

3

3

2

-

1

6

-

-

Financial performance

4

2

3

4

4

4

4

-

1

1

5

-

Steering reports

4

2

3

4

4

4

3

-

1

2

4

-

Benchmarks

3

2

2

3

3

3

3

-

2

5

-

-

Advice IT-systems

3

3

3

3

3

3

3

-

-

7

-

-

Suypplier information

2

2

3

2

4

2

3

-

4

2

1

-

Promotions/sales

3

5

3

3

3

3

4

-

-

5

1

1

2

3

3

2

3

2

3

-

3

4

-

-

Colleagues/Competitors

2

3

3

2

3

2

3

-

3

4

-

-

(market) research

2

3

2

2

2

2

3

-

5

2

-

-

Local

3

3

3

3

3

3

3

-

-

7

-

-

National

3

3

3

3

3

3

3

-

-

7

-

-

2,8

2,8

2,9

2,8

3,3

2,9

3,1

Trends Customers

News

Average score

XXXVI

Appendixes

Hillegom

Veenedaal

Nijmegen

Mierlo

Werkendam

3 = too often

Zoetermeer

1= too less

Tilburg

Table 25 Questionnaire data for information amount; perceived frequency

Reports

Frequency table 1

2

3

Stock level

2

1

2

1

2

2

2

2

5

-

Sales forecasts

2

2

2

2

2

2

2

-

7

-

Product prices

3

2

2

2

3

3

3

-

3

4

Actual Sales

2

2

2

2

2

1

1

2

5

-

Financial performance

3

1

2

1

3

2

3

2

2

3

Steering reports

3

2

2

2

3

2

2

-

5

2

Benchmarks

2

2

2

2

3

3

3

-

4

3

Advice IT-systems

2

2

2

2

2

2

2

-

7

-

Suypplier information

2

2

2

2

3

2

3

-

5

2

Promotions/sales

2

2

2

2

3

3

3

-

4

3

1

1

2

1

1

2

2

4

3

-

Colleagues/Competitors 2

1

2

1

2

2

2

2

5

-

(market) research

2

1

2

1

2

1

1

4

3

-

Local

2

2

2

2

2

2

2

-

7

-

National

2

2

2

2

2

2

2

-

7

-

2,0

1,7

2,0

1,7

2,3

2,1

2,2

Trends Customers

News

Average score

XXXVII

MSc thesis M.J. van Strien

Actual Sales

Financial performance

Steering reports

Benchmarks

Advice IT-systems

Suypplier information

Promotions/sales

Customers

Colleagues/Competitors

(market) research

3 3 0,69 0,48 7

4 4 0,49 0,24 7

3 3 0,38 0,14 7

4 4 0,79 0,62 7

4 4 0,79 0,62 7

3 3 0,49 0,24 7

3 3 0,00 0,00 7

2 2 0,79 0,62 7

3 3 0,79 0,62 7

3 3 0,53 0,29 7

3 3 0,53 0,29 7

2 2 0,49 0,24 7

National

Product prices

2 2 0,53 0,29 7

Local

Sales forecasts

Median Mode St. Dev. Sample Var. Count

Stock level

Table 26 Descriptive statistics for information amount; perceived quantity

3 3 3 3 0,00 0,00 0,00 0,00 7 7

XXXVIII

Sales forecasts

Product prices

Actual Sales

Financial performance

Steering reports

Benchmarks

Advice IT-systems

Suypplier information

Promotions/sales

Customers

Colleagues/Competitors

(market) research

Local

National

Median Mode Standard Deviation Sample Variance Count

Stock level

Table 27 Descriptive statistics for information amount; perceived frequency

2 2

2 2

3 3

2 2

2 3

2 2

2,5 2

2 2

2 2

2 2

1 1

2 2

1 1

2 2

2 2

0,49 0,00 0,53 0,49 0,90 0,49 0,55 0,00 0,49 0,53 0,53 0,49 0,53 0,00 0,00 0,24 0,00 0,29 0,24 0,81 0,24 0,30 0,00 0,24 0,29 0,29 0,24 0,29 0,00 0,00 7 7 7 7 7 7 6 7 7 7 7 7 7 7 7

Appendixes

Table 28 Questionnaire data for information load measure

Information

Complexity

load

Nijmegen (s7)

Werkendam (s6)

Mierlo (s5)

table

1 2 3 4 5

Simple (16)

2

3

3

4

3

4

2

- 2 3 2 -

Patterned (16)

4

3

4

2

3

4

2

- 2 2 3 -

(16)

2

3

1

2

3

3

2

1 3 3 - -

Good form (16)

3

3

2

4

2

4

3

- 2 3 2 -

Varied (16)

4

3

4

4

4

2

4

- 1 1 5 -

Consistent (16)

4

3

2

4

4

4

2

- 2 1 4 -

Consonant (16)

4

5

3

4

2

2

3

- 2 2 2 1

Pertinent (16)

4

5

4

5

3

4

4

- - 1 4 2

Common (16)

4

5

3

4

4

4

4

- - 1 5 1

(16)

2

5

4

4

4

4

3

- 1 1 4 1

Familiar (16)

4

3

3

2

2

3

4

- 2 3 2 -

Usual (16)

4

5

4

4

3

4

4

- - 1 5 1

Complete (16)

4

3

3

4

3

4

4

- - 3 4 -

Important (16)

4

3

4

5

4

4

3

- - 2 4 1

Reliable (16)

4

3

4

4

4

4

3

- - 2 5 -

Precise (16)

4

3

3

4

4

4

2

- 1 2 4 -

Small

Novelty

Veenendaal (s4)

Table 17 for scale explanation

Hillegom (s3)

See

Zoetemeer (s2)

Tilburg (s1)

Frequency

scale

Meaningful

Reliability

XXXIX

MSc thesis M.J. van Strien

Table 29 Descriptive statistics for information load scale

Information load

Precise

Reliable

Complete

Usual

Familiar

Important

Reliability

Meaningful

Common

Pertinent

Consonant

Consistent

Varied

Good form

Patterned

Simple

Novelty

Small scale

Complexity

Median

3

3

2

3

4

4

3

4

4

4

3

4

4

4

4

4

Mode

3

4

2

3

4

4

4

4

4

4

3

4

4

4

4

4

St. Dev

0,82 0,90 0,76 0,82 0,79 0,95 1,11 0,69 0,58 0,95 0,82 0,58 0,53 0,69 0,49 0,79

Sample Variance 0,67 0,81 0,57 0,67 0,62 0,9- 1,24 0,48 0,33 0,9- 0,67 0,33 0,29 0,48 0,24 0,62 Count

7

7

7

7

7

7

7

7

7

Tabel 30 Descriptive statistics for information load variable

Confused

Emotionally pressured

Irritated

by

Emotional load Overwhelmed the effort

Cannot cope with

Cannot handle

Cannot process

Cognitive load

Median 2 2 2 2 2 2 2 Mode 2 2 2 2 2 2 2 Standard Deviation 1,11 0,98 0,95 0,95 1,40 0,79 0,38 Sample Variance 1,24 0,95 0,90 0,90 1,95 0,62 0,14 Minimum 1 2 1 2 1 2 2 Maximum 4 4 4 4 5 4 3 Count 7 7 7 7 7 7 7

XL

7

7

7

7

7

7

7

Appendixes

Table 31 Questionnaire data for information load variable; including frequency table

Tilburg (s1)

Zoetemeer (s2)

Hillegom (s3)

Veenendaal (s4)

Mierlo (s5)

Werkendam (s6)

Nijmegen (s7)

Frequency table

Cannot process

2

1

3

3

2

1

4

2

2

2

1

-

Cannot handle

4

2

2

2

2

2

4

-

5

-

2

-

Cannot cope with

4

2

2

1

2

2

3

1

4

1

1

-

1

2

3

4

5

Cognitive

Overwhelmed by

overload

the effort

2

2

4

2

4

2

3

-

4

1

2

-

Irritated

2

1

4

2

2

2

5

1

4

-

1

1

Emotionally Emotional

pressured

2

3

4

2

2

2

3

-

4

2

1

-

overload

Confused

2

3

2

2

2

2

2

-

6

1

-

-

Average score

2,6 2,0 3,0 2,0 2,3 1,9 3,3

XLI

MSc thesis M.J. van Strien

Table 32 Questionnaire data for dependent variables; including frequency table

Tilburg (s1)

Zoetemeer (s2)

Hillegom (s3)

Veenendaal (s4)

Mierlo (s5)

Werkendam (s6)

Nijmegen (s7)

Frequency

Decision

Frequency

20

4

2

4

4

4

4

5

- 1 - 5 1

Quality

Quantity

21

2

2

2

3

4

4

5

- 3 1 2 1

Subjective

27

3

4

3

4

4

4

4

- - 2 5 -

3,0

2,7

3,0

3,7

4,0

4,0

4,7

Q

table 1 2 3 4 5

Business

Overstock

22

2

2

1

2

2

4

3

1 4 1 1 -

performance

Understock

23

2

2

1

2

2

4

2

1 5 - 1 -

Waste

24

2

2

2

3

2

3

3

- 4 3 - -

Stock costs

25

2

4

2

2

3

4

2

- 4 1 2 -

26

2

4

2

2

2

5

3

- 4 1 1 1

2

2,8

1,6

2,2

2,2

4

2,6

4

3

3

3

4

-

3

- - 4 2 -

1

1

1

1

1

-

1

6 - - - -

3

2

2

2

2

-

2

- 5 1 - -

2

4

4

4

3

-

4

- 1 1 4 -

Financial impact Average scores Information Causes

quantity Information quality Delivery failure Deliberate choice

XLII

Appendixes

Table 33 Descriptive statistics for dependent variables

Financial impact

Stock costs

Waste

Understock

Overstock

Subjective

Quantity

Frequency

Decision Quality Business Performance

Median

4

3

4

2

2

2

2

2

Mode

4

2

4

2

2

2

2

2

Standard Deviation 0,90 1,21 0,49 0,95 0,90 0,53 0,95 1,21 Sample Variance

0,81 1,48 0,24 0,90 0,81 0,29 0,90 1,48

Range

3

3

1

3

3

1

2

3

Count

7

7

7

7

7

7

7

7

XLIII

MSc thesis M.J. van Strien

Systems designers

C1000 Retail

PlusRetail

C1000RetaiCC1000

Table 34 Questionnaire data for systems designers

(SMM‟s)

4

4

Het IS toont de informatie vanuit één perspectief

3

2

kolommen/soorten informatie gelimiteerd

3

4

We hebben het aantal afgebeelde rijen gelimiteerd

4

4

2

4

2

5

verkrijgen

4

4

Wij bepalen welke informatie belangrijk is

2

2

3

4

3

-

4

5

2

4

3

4

1= Fully disagree; 5=dully agree

Het informatiesysteem (IS) is ontworpen rondom de informatiebehoefte van supermarktmanagers

We

hebben

het

aantal

afgebeelde

De verstrekte informatie laadt ruimte voor dubbelzinnige interpretatie We streven ernaar om alle informatie via het IS te verspreiden Informatie uit het IS is ook via andere bronnen te

De informatie wordt in weloverwogen batches getoond i.p.v. alles ineens Relevante informatie wordt gehighlight Irrelevante informatie wordt bewust niet getoond We gebruiken het liefst een „push‟ principe voor de IS‟en Het is de bedoeling dat de SMM – naast het IS zelf nog denk- en rekenwerk uitvoert

XLIV

Appendixes

Ik begrijp de situatie waarin de SMM‟s werken en besluiten nemen Alle

gebruikers

krijgen

training

bij

4

4

4

4

4

4

1

2

5

4

2

4

4

4

1

2

3

4

2

4

2

2

4

4

het

implementeren van een (nieuw) IS Tijdens de implementatie wordt het IS aangepast aan de eigenschappen van de winkel Tijdens de implementatie wordt het IS aangepast aan de eigenschappen van de manager SMM‟s kunnen aanvraag doen voor aanvullende functionaliteit/IS‟en Aanvullende IS‟en worden alleen op aanvraag ontworpen De gebruiker wordt gedwongen met de nieuwe functionaliteit te werken SMM‟s kunnen functionaliteit uit bestaande IS‟en verwijderen Een aanvraag wordt ... geaccepteerd Als een nieuw IS een bestaande functionaliteit vervangt, wordt de bestaande IS aangepast De informatie uit de systemen is te complex voor de SMM‟s De SMM‟s voelen zich vertrouwd met de informatie uit de systemen

XLV

MSc thesis M.J. van Strien

Appendix VI: Content analysis Table 35 Mapping the quotes with categories

Category

Subcategory

Experience General

s1

s2

s3

2, 3, 8, 7,1

s4

s5

s6

s7

1

1,2,10

1,4,6

34

3,5

2,3,25

9 Domain

10, 34,

knowledge

35

Technology use 1 Trust

22,23

3,6,8,

5, 6, 7, 8,45

1,7

4 23

26, 28,

5,10,12, 3,5,15,2 3,4

3,15,17,

15

0

19

6,8

6,7,8,9, 5,6

7,8,9

29 Time

General

Pressure

13,14,1 2,13,16, 5

Interrupting Structuredness

18,51

25

9,14,31 12

17

10 5

7,21,22, 23,24

15

23,23

24 Information General

16,32

Amount

25,26,3 3,7,8,9, 9,11,13 10,12,1 7,9 0,32,34, 12,14,1 39,52

Complexity

9,11,12

3,18

5

4,11,20, 6,19,21, 6,8,10,1 12,14,1 14,18,1 8 23,25,

27,28,3 1,12

3-

2,36,47

6,19

12,17,8

9

,31,35, Novelty

18

17,24,2 24 6

Pertinence

21,22,2 22,23,2 24,16,1 14,19,2 17,21,2 13

12,16,2

7

1

4,25,27, 7,18,24, -,24 40,41,4 26

XLVI

2

Appendixes

2,44,45, 54,55 Quality

Relevance

28,53,5 22,26

15,11,1 14

4

5,16

17,19,2 20,33,3 1,8,9,17 10,17,1

12

13

14,2

4,30,33, 5,36,38, ,18,19,2 8 34

45,48,4 0 9

Table 36 Abbreviation for the categories of the content analysis

Code TU DK TR EX ST IN PE RE CO NO QU TP IA

Category Technology use Domain knowledge Trust General Structuredness Interrupting Pertinence Relevance Complexity Novelty Quality General General

Table 37 Content analysis quotes

Content analysis Subject Code quote s1 EX Als je dit ziet dan denk je 'hoe kan dit nou?', alleen de informatievoorziening is gegroeid in de loop van de jaren. Het is niet dat de informatie in 1x op ons af is gekomen s1 EX De informatievoorziening is gematigd geëxplodeerd s1 EX Het ligt niet aan C1000 dat ik alle informatie goed weet te delegeren, dat is een stukje ervaring dat je opdoet s4 EX Alleen door mijn ervaring weet ik informatie goed te filteren. Dan kan ik goed bepalen of ik het moeten weten of dat dit naar iemand anders moet

XLVII

MSc thesis M.J. van Strien

Content analysis Subject Code quote s4 EX De zaak is van mijn vader geweest, dus ik bene r eigenlijk ingerold" s5 EX Ik ben pas sinds dit jaar ondernemer. Ik ben begonnen als stagiair en heb hiervoor wel 8 jaar in de supermarkt gewerkt s6 EX Ik ben begonnen in de supermarkt als bijbaantje en ben er zo verder ingerold. Ik heb nu ongeveer 25 jaar ervaring met het werken in een supermarkt s6 EX Door mijn ervaring heb ik het gevoel dat ik meestal wel goed met de hoeveelheid om kan gaan s6 EX Daarmee ben ik volgens mij behoorlijk ervaren s7 EX "Ik werk al vanaf mijn achttiende in de supermarkt. Ik ben begonnen als vakkenvuller en daar is mijn interesse door gegroeid s7 EX "Het is een kwestie van de juiste vragen stellen"(5.15) s7 EX Ik vind de informatie belangrijk om een goede beslissing mee te maken. Ik kan wel alles vanuit ervaring doen, maar ik kan ook niet alles weten s1 TU Als je niet begaan bent met techniek, wordt het lastig s2 TU De input voor het bestelsysteem moet goed zijn. Het duurt jaren voordat bde hele winkel begrijpt hoe dat werkt s2 TU Ik ben onderdeel van de werkgroep ICT van C1000. Vooral omdat ik gevoel heb voor IT en de informatiestroom vanuit C1000 een interessant onderwerp vind s2 TU De systemen zijn toch wel pittig. Je moet er mee om leren gaan. Je ziet nu nog vaak dat ze blijven bestellen op de manier waarop ze deden s3 TU Negetig procent kun je door de bestelmodule laten doen en 10 procent kun je dan nog toevoegen s3 TU De informatie op C1000net is algemeen, dit is moeilijk om specifiek om aan proces te koppelen s4 TU Tijdens mijn carrière heb ik om moeten leren gaan met de ICT-systemen. Vroeger gebeurde alles nog anders en moesten we het werk zelf doen s7 TU Je moet er mee om leren gaan. Je ziet nu nog vaak dat ze blijven bestellen op de manier waarop ze deden s7 TU "Tijdens mijn carrière heb ik om moeten leren gaan met de ICT-systemen. In het begin as dat lastig en mede daarom heb ik nu niet altijd vertrouwen in de systemen s1 DK Ik ben de winkels langsgegaan om mensen wegwijs te maken in het systeem s1 DK Ik heb zelf in die sturende rol gezeten s1 DK Ik voel me sterk ervaren met de keuzes s1 DK De harde werkers zijn nog steeds nodig, maar daarnaast moet je ook de processen begrijpen s2 DK Als je snapt hoe dingen werken, dan gaat het goed s2 DK Het is een kwestie van de juiste vragen stellen s4 DK Ik werk al vanaf mijn achttiende in de supermarkt. Ik ben begonnen als vakkenvuller en daar is mijn interesse door gegroeid s4 DK Uit ervaring kan ik zeggen dat de afdelingshoofden toch wel naar me toekomen als het om belangrijke beslissingen gaat s4 DK Ik voel me zeer ervaren met het maken van beslissingen, vooral die over voorraadbeheer

XLVIII

Appendixes

Content analysis Subject Code quote gaan s5 DK Die stage was onderdeel van de opleiding retail-management. Daarna heb ik overigens nog meer opleidingen gevolgd s5 DK Ik ben pas sinds dit jaar ondernemer. Ik ben begonnen als stagiair en heb hiervoor wel 8 jaar in de supermarkt gewerkt s5 DK Met bestelbeslissingen heb ik 7 jaar ervaring. Ik hielp hier altijd mee tijdens mijn bijbaantje s7 DK "Ik voel me zeer ervaren met het maken van beslissingen, ook al is er de laatste tijd nog wel wat verandering in de branche. Daardoor moeten we rekening gaan houden met andere dingen s1 TR Ik ben daar heel praktisch in, gewoon dat ding laten berekenen s1 TR Op het moment dat wij het voortraject goed hebben uitgevoerd kunnen we met een gerust hart tegen het systeem zeggen, ga maar een bestelling maken s1 TR Verbeteren van het advies is puur emotie, eigenlijk moet je gewoon rationeel dat ding zijn werk laten doen s1 TR Als jij zorgt dat je voorraad goed staat, hoef jij niks te controleren (aan het besteladvies) s1 TR Als ik weet dat de voorraad informatie goed is, dan is het voor mij goed s1 TR Het vertrouwen (in het systeem) heb ik zelf voor gezorgd dat ik dat moest krijgen s2 TR De basis van het besteladvies is goed. 90% van de gevallen is het advies goed, is 10% behoeft wat meer aandacht. De managers moeten hierop focussen. Hiervoor zijn aandachtsgroepen gemaakt in het systeem s2 TR De input voor het bestelsysteem moet goed zijn. Het duurt jaren voordat bde hele winkel begrijpt hoe dat werkt s3 TR De informatie op C1000net is algemeen, dit is moeilijk om specifiek om aan proces te koppelen s4 TR Ik heb het vertrouwen in de systemen ontwikkeld omdat ik er zeker de meerwaarde van inzie s4 TR Ik heb in het begin slapeloze nachten gehad omdat ik niet aan die systemen kon wennen. Ik vertrouwde er nog niet op. Inmiddels heb ik dat vertrouwen wel s4 TR Ik moest wel wennen aan alle informatie die uit de systemen komt s4 TR Ik ben over het algemeen tevreden met de informatie die we krijgen, dus ik heb niet veel moeite om hiermee om te gaan s5 TR Ondanks dat ik het misschien pas een jaartje doe, ben ik wel overtuigd dat ik dit goed kan s5 TR Eigenlijk ben ik best wel tevreden over de informatiesystemen. s5 TR De ICT-systemen zijn nog niet zo professioneel als die van de concurrent. Vooral het feit dat ons systeem nog geen real-time mutaties kan verwerken is op dit moment een minpunt s5 TR Het liefst maak ik beslissingen op basis van ervaring of intuïtie. Informatie komt daarna omdat ik vaak ervaren heb dat de informatie niet klopt. s6 TR Voor het maken van beslissingen vertrouw ik het liefst op informatie, zolang deze goed is zou het resultaat ook goed moeten zijn s6 TR Ik vertrouw ook voor 95% van de bestellingen op het advies dat het systeem geeft s7 TR Ik denk alleen niet dat dit echt invloed heeft op mijn beslissingen. Het blijkt vaak toch goed

XLIX

MSc thesis M.J. van Strien

Content analysis Subject Code quote te gaan als ik het op op mijn ervaring doe s7 TR Eigenlijk ben ik best tevreden over de ICT-systemen, alleen ik verlies er soms het overzicht door s7 TR Het bestelsysteem is heel duidelijk. Ik maak 95% van de tijd gebruik van de adviezen. s7 TR "Tijdens mijn carrière heb ik om moeten leren gaan met de ICT-systemen. In het begin as dat lastig en mede daarom heb ik nu niet altijd vertrouwen in de systemen s1 TP Bestellen is core-business, dat moet op tijd gebeuren s1 TP We hebben nu 20 uur om te bestellen, met ons systeem is dat geen probleem s1 TP Het is geen probleem om op tijd te bestellen s2 TP Ik probeer zelf mijn operationele taken zo klein mogelijk te houden, anders hou ik geen tijd voor coaching s2 TP Het is wel eens gebeurd dat ik ergens geen tijd voor had, terwijl ik er dieper in moest duiken s2 TP Je kunt je niet permitteren om niet op de hoogte te zijn van trends, en concurrentie s2 TP Als medewerkers die berichten te laat leest, dan leidt dat altijd tot nee-verkoop s2 TP Als ondernemer, is mijn ervaring dat als ik hier rondwandel, dan is een uurtje of 25 niet ingevuld. Die laat ik gewoon vrij en laat ik afhangen van het proces s4 TP We worden dagelijks bevoorraad, daarom is het meestal niet heel erg als we iets te laat bestellen s4 TP Ik ervaar eigenlijk geen tijdsdruk en heb niet het gevoel dat mijn medewerkers dat we doen s5 TP Er is helemaal geen tijdsdruk bij het maken van bestelbeslissingen. Ik heb meestal meer dan een dag om de beslissing te maken, het is een kwestie van je zaakjes goed op orde hebben s5 TP Ik heb soms eerder het idee dat ik te vroeg ben met bestellen en dat het distributiecentrum nog niet klaar is s5 TP Nee, bij het maken van die beslissingen ervaar ik niet echt een tijdsdruk. Ik ben sowieso niet het persoon dat snel druk ervaart. s5 TP Ik vind goed voorraadbeheer zeker belangrijk. Het bepaalt je klantenservice en heeft ook een belangrijke impact op je financiële prestatie s5 TP Er komen tweemaal per dag een vrachtwagen binnen om te bevoorraden. 's Ochtends eentje met vooral verse producten en later in de middag met houdbare producten s6 TP Het is vaak lastig om alle informatie goed op orde te krijgen voordat we daadwerkelijk de beslissing moeten maken s6 TP Ik ervaar soms wel een tijdsdruk om een bestelling goed de deur uit te doen ja s7 TP Er komen per dag twee vrachtwagens voor de bevoorrading. Daardoor hebben we altijd zeker tot 4 uur om de beslissing te maken s7 TP Het komt zeker wel eens voor dat we alles nog niet op orde hebben toen we de bestelling door moesten geven s7 TP Ik ervaar bij het maken van de beslissing zeker wel eens een tijdsdruk. Dat komt vooral omdat het veel tijd kost om overal van op de hoogte te zijn s1 ST Je maakt je beslissingen van dag tot dag s2 ST Het type informatie dat je krijgt verschilt nogal. Je het dagelijkse informatie die en incidenteel, vernieuwd is. Dat vraagt natuurlijk nogal wat van ons

L

Appendixes

Content analysis Subject Code quote s3 ST "De activiteiten worden steeds complexer. Je moet rekening houden met regelgeving, klanten, concurrenten, ga maar door s4 ST Het maken van de beslissingen ervaar ik niet als lastig. Wel is het zo dat 10% van de artikelen afhankelijk is van de situatie. Hierbij moeten we rekening houden met het weer enzo s4 ST Het voorbereiden op bestellen begint meestal 's avonds met het tellen van de voorraden in het magazijn en de schappen. Op deze manier hebben we de volgende dag nog zeker 4 uur om alles op een rijtje te zetten s4 ST Er is voor mij altijd duidelijk welke beslissingen ik moet maken s4 ST Het doel altijd simpel. Zorgen dat alle producten voldoende op voorraad zijn s5 ST Het is voor mij altijd heel duidelijk welke beslissing ik moet maken. Ik heb wel vaak probleem om te bepalen wanneer en hoevewel er besteld moet worden s5 ST Het doel van de bestelling is altijd duidelijk, de producten moeten op voorraad zijn. Aan de andere kant moeten er ook weer niet teveel producten in het magazijn liggen. Dit kost alleen maar geld s6 ST Het is voor mij altijd duidelijk wanneer ik een bestelling moet plaatsen. s7 ST Het doel altijd duidelijk; we moeten zorgen dat alle producten voldoende op voorraad zijn. Of dat lukt is weer een andere vraag s7 ST Jawel, het is voor mij altijd duidelijk welke beslissingen ik moet maken. De beslissing heeft altijd een bepaalde structuur s2 IN Ik heb het wel eens dat als er veel belangrijke dingen zijn waarvoor ik gestoord wordt, er minder tijd over blijft voor de dagelijkse communicatie s2 IN Als ondernemer heb je het makkelijk, want je zit er de hele dag in s2 IN Het is heel belangrijk om je aandacht goed te verdelen, en dat gaat zeker wel eens mis s7 IN Het is een hectisch beroep, er komen altijd dingen tussendoor s1 IA Ik ben erg tevreden over de informatievoorziening s1 IA Alle belangrijke informatie komt via C1000net. Af en toe is het rustig, af en toe is het druk s2 IA Het is de bedoeling dat elke afdeling zijn berichten bijhoudt, maar het wordt wel eens veel s2 IA Het meeste vanuit C1000 gaat via C1000net, daarnaast ook veel via email s2 IA Voor sommige afdelingen wordt het wel eens te veel informatie, terwijl ze eigenlijk gewoon met de klant bezig moeten zijn s2 IA Inhoudelijk kan ik daar best mee omgaan. Het kan wel eens zijn dat je veel dingen krijgt en dat je daar veel tijd mee kwijt bent s2 IA Ik vind de hoeveelheid informatie die via C1000net wordt verspreid fors. Het is en onverwacht en vrij veel s2 IA De ondernemer ïs het lastig, die heeft een gigantische lijst met berichten en hij moet overal achteraan gaan bellen als hij denkt gaat dit wel goed s2 IA Het zijn best intensieve lijstjes,, waarvan ik denk die laat ik wel eens een paar dagen liggen s3 IA Ik vind niet dat de communicatie erop is verbeterd met C1000net s3 IA Als je kijkt wat we krijgen aan berichtgeving, is dat eigenlijk niet meer normaal s3 IA Ik maak daarmee de stapel van informatie voor mezelf erg hoog, maar doe dat wel bewust

LI

MSc thesis M.J. van Strien

Content analysis Subject Code quote s3 IA In C1000net is het al een hele klus om daar te komen waar je moet zijn s3 IA De afdelingsmanagers kunnen het soms helemaal niet meer bijhouden, dan moet ik voorfilteren s3 IA Ik moet zeggen dat het steeds lastiger wordt om goed met die informatie om te gaan. Ze proberen dat wel te stroomlijnen met C1000net, maar soms is dat bijna niet meer te volgen. Dan moet je echt gaan zoeken naar nieuws en moeten we steeds vaker gaan filte s3 IA De afdelingsmanagers moeten zelf ook de informatie van C1000net opzoeken s4 IA Het systeem, en dan zeker C1000net geeft soms erg veel informatie richting mij en de medewerkers. s4 IA Vier keer per jaar komt er iemand vanuit het hoofdkantoor praten iover hoe de winkel het doet. We kijken vooral naar onszelf, niet naar anderen s4 IA De systemen leveren ons veel informatie, maar halen dan ook veel van het reken- en denkwerk weg s5 IA Vooral de externe informatie speelt een belangrijke rol in mijn werk. Deze wordt niet altijd via het systeem geleverd s5 IA Veel van de informatie wordt via de systemen aangeleverd. Dat is toch wel twee-derde denk ik. s5 IA Als je kijkt naar informatie systemen dan hebben we een aantal losstaande systemen ter beschikking. Bijvoorbeeld C1000net, het bestelsysteem en het facturensysteem s5 IA Een paar keer per jaar komt een account manager vanuit het hoofdkantoor hier praten over de presentaties van de winkel. Ik haal hier vel informatie uit, maar het blijft adviserend s6 IA Er komt veel informatie op me af. Ik heb het gevoel dat de meeste informatie via C1000net komt s6 IA Ik voel me regelmatig onder druk gezet door de hoeveelheid informatie die ik moet verwerken s7 IA Dat klopt helemaal ja, het aantal informatie dat mijn kant op komt is enorm gegroeid s7 IA De informatie is zeker veel. Daarnaast moet ik ook vaak zoeken om alles bij elkaar te krijgen, ik vind dat C1000net een betere structuur kan krijgen s7 IA Ik ervaar bij het maken van de beslissing zeker wel eens een tijdsdruk. Dat komt vooral omdat het veel tijd kost om overal van op de hoogte te zijn s1 PE Aandachtsgroepen kun je zelf aanmaken en daarin kun je aangeven wat je belangrijk vindt dat er gecontroleerd wordt s1 PE De informatie voorziening op zich is goed. Alleen hij sluit niet altijd een op een aan bij wat er gebeurt s1 PE Wij moeten de bestelling nog helemaal aanmaken en beoordelen. Wij moeten zelf rekening houden met hiaten door bijvoorbeeld weersverwachting s2 PE Dat vind ik frusterend, want mijn medewerkers weten niet wanneer ze iets kunnen verwachten en weten niet wie er actie op heeft ondernomen s2 PE Als zij daar goed over na zouden denken, dan zouden ze het anders brengen s2 PE Het is niet negatief bedoeld, maar het gebeurd wel gewoon zo. Zij denken 'als wij de informatie hebben geleverd, dan hebben wij ons werk gedaan' maar dat is natuurlijk niet zo s2 PE Je ziet bij C1000 nog wel eens dat ze denken totaan de ondernemer. Terwijl ze eigenlijk

LII

Appendixes

Content analysis Subject Code quote verder moeten denken s2 PE De opbouw van C1000net is niet intuitief. Het is ouderwets met hoofdstukken s2 PE C1000net wordt de Schuitema als communicatiemiddel ingezet om het voor hen makkelijker te maken. Terwijl er nog niet helemaal wordt gekeken naar hoe het voor de ondernemer makkelijker wordt gemaakt s2 PE De basis van het besteladvies is goed. 90% van de gevallen is het advies goed, is 10% behoeft wat meer aandacht. De managers moeten hierop focussen. Hiervoor zijn aandachtsgroepen gemaakt in het systeem s2 PE De forecast wordt bepaald op 16 weken. Jaarlijkse trends worden hier niet in meegenomen. Dit is vakmanschap van de manager s2 PE De ondernemer ïs het lastig, die heeft een gigantische lijst met berichten en hij moet overal achteraan gaan bellen als hij denkt gaat dit wel goed s2 PE Op het moment dat zij dat bericht geproduceerd hebben, gooien zij dat over de schutting heen. Dat wordt dan op die internetsite neergezet, maar hier gaat het proces wel verder s2 PE Mijn bezwaar in wat zij doen is, ze gooien informatie over de schutting heen. En dat vind ik niet goed s2 PE Als je kijkt naar de inhoud ben ik wel tevreden. Maar als je kijkt naar het proces, daar heb ik wel wat opmerkingen over s3 PE Qua IT vind ik dat je mee moet met de laatste ontwikkelingen om aansluiting te kunnen vinden bij andere systemen s3 PE Er is een indeling naar assortimentsgroep, maar de dagberichten komen allemala door elkaar binnen. Degene waarvoor het is bedoeld, moet het dan maar uitzoeken s3 PE Er moet iemand zijn die, voordat het die trechter ingegooid wordt, bepaald dit is wel of niet interessant s3 PE Als ik het zelf niet weet, kan ik nooit weten of het überhaupt in de winkel goed gaat komen s3 PE Met de PDA's kunnen mensen zien wat het besteladvies zien en dat controleren op basis van wat ze in de winkel zien en wat ze denken dat er morgen gebeurt s4 PE Het maken van de beslissingen ervaar ik niet als lastig. Wel is het zo dat 10% van de artikelen afhankelijk is van de situatie. Hierbij moeten we rekening houden met het weer enzo s4 PE Het systeem houdt geen rekening bijvoorbeeld het weer voor verkoopvoorspelling. Dit moeten we dan zelf dus nog doen s4 PE De enige opmerkingen die ik over het bestelsysteem heb ik dat de informatie niet altijd duidelijk wordt weergegeven en dat het niet altijd klopt s4 PE Voor de artikelen die afhankelijk zijn van de omstandigheden hebben we zelf een logboekje aangelegd. Het systeem houdt namelijk geen rekening met deze omstandigheden s5 PE Als je vraagt of ik de informatie het liefst toegestuurd krijg ik zelf opzoek is dat afhankelijk van welke informatie. Ik denk dat het 50/50 is wat ik het best vind s5 PE Doordat de systemen niet real-time zijn, kan ik niet op elk gewenst moment inzicht krijgen in de voorraadstatus. Dit maakt het lastiger om goede beslissingen te maken s5 PE Voor de forecast houden de systemen geen rekening met de omstandigheden zoals weer en evenementen. Dit vind ik wel jammer, want nu moeten we dat handmatig gaan verwerken

LIII

MSc thesis M.J. van Strien

Content analysis Subject Code quote s6 PE Het komt best vaak voor dat de informatie elkaar tegenspreekt. Meestal zijn dit berichten op C1000net, dan hebben ze centraal een fout gemaakt die rechtgezet moet worden s7 PE De informatie is zeker veel. Daarnaast moet ik ook vaak zoeken om alles bij elkaar te krijgen, ik vind dat C1000net een betere structuur kan krijgen s7 PE Het intranet is eigenlijk alleen een berichtendoorgeefluik. Ik mis het vaak dat ik iets met het bericht kan doen, bijvoorbeeld weggooien, of juist doorsturen naar de verantwoordelijke s7 PE De andere vijf procent vind ik dat we (afdelingsmanagers) goed op moeten letten of ze wel kloppen. Deze vijf procent kan trouwens wel steeds anders zijn s1 RE Je kunt het systeem heel dynamisch maken, maar je moet er niet in doorslaan s1 RE Als je dit ziet dan denk je 'hoe kan dit nou?', alleen de informatievoorziening is gegroeid in de loop van de jaren. Het is niet dat de informatie in 1x op ons af is gekomen s1 RE Het informatiesysteem is toegespitst op verschillende afdelingen s1 RE Alle informatie wordt door de afdelingsmanagers zelf gelezen, ik doe er niks mee. Ik lees wel 90% van de informatie die binnenkomt s1 RE Sommigen informatie is essentieel voor je bedrijfsvoering s1 RE Bijna alle informatie die binnenkomt neem ik tot me, alleen ik ben niet altijd uitvoerende van alle informatie s2 RE Je kunt zelf bepalen of je de informatie uit C1000net wilt zien s2 RE Ik zorg wel dat wat van belang is binnen het bedrijf, wat van belang is, dat ik dat wel zie ergens s2 RE De basis van het besteladvies is goed. 90% van de gevallen is het advies goed, is 10% behoeft wat meer aandacht. De managers moeten hierop focussen. Hiervoor zijn aandachtsgroepen gemaakt in het systeem s2 RE Er is gene goed onderscheid tussen ondernemersberichten en medewerkersberichten. Hierdoor krijg ik de neiging om alles te gaan lezen. Daardoor kunnen medewerkers bij belangrijke documenten s2 RE De indeling naar afdeling (in C1000net) zou nog een stap gedetailleerder moeten. De indeling is ook niet in elke winkel hetzelfde, maar in C1000net wel s2 RE Ik zou er meer voor zijn dat je bijvoorbeeld inlogt als X en dat dat systeem dan weet welk bericht voor haar is s2 RE De meeste berichten zijn wel relevant. Ik ddenk bijna nooit dat ik een bericht niet had hoeven lezen s2 RE De informatie wordt wel snel een brei. Dan staan er acht berichten die niet belangrijk zijn, maar daar tussenin staat wel een hele belangrijke. Daardoor wordt het wel eens ononverzichtelijk s3 RE Als ik het zelf niet weet, kan ik nooit weten of het überhaupt in de winkel goed gaat komen s3 RE De afdelingsmanagers kunnen het soms helemaal niet meer bijhouden, dan moet ik voorfilteren s3 RE Zestig tot zeventig procent van de informatie is belangrijk voor mijn bedrijfsvoering, de rest lig ik niet wakker van s3 RE Er moet iemand zijn die, voordat het die trechter ingegooid wordt, bepaald dit is wel of niet interessant

LIV

Appendixes

Content analysis Subject Code quote s3 RE Ik zie het verband niet zo hoe de hoeveelheid informatie invloed heeft op mijn bestelling s3 RE In C1000net is het al een hele klus om daar te komen waar je moet zijn s3 RE Als ik zelf de informatie niet zie, kan ik ook niet bepalen of het in de winkel ergens wel goed terecht komt. Daarbij hoef ik het niet helemaal uit te pluizen, maar wil ik wel weten waar het over gaat en vooral weten of ik dat in de winkel terug zie s4 RE De financiële sturing vind ik niet altijd nodig, ik weet goed hoe het gaat en daarbij hoef ik geen cijfertjes vanuit Schuitema s4 RE Alleen door mijn ervaring weet ik informatie goed te filteren. Dan kan ik goed bepalen of ik het moeten weten of dat dit naar iemand anders moet s4 RE Ik heb het vertrouwen in de systemen ontwikkeld omdat ik er zeker de meerwaarde van inzie s6 RE De hoeveelheid maakt het lastig omdat eigenlijk alles wel belangrijk is om te weten s7 RE C1000net is niet zo ingericht dat het aansluit bij onze afdelingen. Dit zorgt soms dat er onbedoeld berichten bij iemand komen s7 RE Met C1000net heb ik wel vaak moeite. Daar staan veel berichten op die dan niet allemaal voor mij bedoeld zijn s1 CO C1000net is een goed medium, alleen het is vaak te omslachtig s1 CO De informatievoorziening is gematigd geëxplodeerd s1 CO Als ik in het systeem dan ooit alle parameters op wil zoeken, dat gaat niet s1 CO Je kunt het systeem heel dynamisch maken, maar je moet er niet in doorslaan s1 CO Het gebeurt eigenlijk bijna nooit dat er informatie buiten het informatiesysteem om komt s1 CO Voor de mensen die nieuw binnen komen is het verstandig niet alles in in1x te laten zien of te leren s1 CO Als er iemand nieuws is, dan ben je weken kwijt met hem wegwijs maken in het systeem (C1000net en bestelmodule) s1 CO De onderlinge link tussen processen zou er veel meer moeten zijn s2 CO De systemen zijn toch wel pittig. Je moet er mee om leren gaan. Je ziet nu nog vaak dat ze blijven bestellen op de manier waarop ze deden s2 CO Het meeste vanuit C1000 gaat via C1000net, daarnaast ook veel via email s2 CO De invloeden komen van de consument, C1000, de consument, en je eigen gevoel s2 CO De meeste informatie komt via C1000net s2 CO De opbouw van C1000net is niet intuitief. Het is ouderwets met hoofdstukken s2 CO Het is oude informatie in een nieuw jasje s2 CO De dagelijkse dingen voor bestelling doe ik niet, maar ik controleer wel wat ik zie in de winkel. Aandachtspunten uit de winkel zijn aanleiding om in de computer te kijken s2 CO De informatie wordt wel snel een brei. Dan staan er acht berichten die niet belangrijk zijn, maar daar tussenin staat wel een hele belangrijke. Daardoor wordt het wel eens ononverzichtelijk s3 CO Ik vind niet dat de communicatie erop is verbeterd met C1000net s3 CO Iedereen heeft zijn informatie. Op gebied van personeelszaken, wijzigingen in regelgeving, risicomanagement, assortiment en prijsverlagingen van de concurrentie. En dat wordt een

LV

MSc thesis M.J. van Strien

Content analysis Subject Code quote hele brei. En er wordt van mij verwacht dat ik weet waar ik alles moet vinde s3 CO Het wordt steeds lastiger om die informatie te behappen s3 CO De complexiteit en het vinden van, dat vind ik steeds moeilijker worden s3 CO De afdelingsmanagers kunnen het soms helemaal niet meer bijhouden, dan moet ik voorfilteren s4 CO Alleen soms is het bij bepaalde acties zo dat we overspoel worden met berichten. Dan is het wel even lastig om dit allemaal te begrijpen s4 CO De enige opmerkingen die ik over het bestelsysteem heb ik dat de informatie niet altijd duidelijk wordt weergegeven en dat het niet altijd klopt s4 CO Ik moest wel wennen aan alle informatie die uit de systemen komt s4 CO Voor de artikelen die afhankelijk zijn van de omstandigheden hebben we zelf een logboekje aangelegd. Het systeem houdt namelijk geen rekening met deze omstandigheden s5 CO Maar er komt ook veel informatie via mensen binnen. Ik heb bijvoorbeeld een aantal mensen die op de winkel letten zonder dat ze hiervoor in functie zijn. Je kunt ze spionnen noemen s5 CO Vooral de externe informatie speelt een belangrijke rol in mijn werk. Deze wordt niet altijd via het systeem geleverd s5 CO Gesprekken met klanten, en collega's zijn erg belangrijk om competitief te blijven s6 CO Gelukkig is de meeste informatie wel goed gestructureerd, maar soms is het dan wel weer onduidelijk weergegeven s7 CO Dan kan ik niet alle informatie goed verwerken s7 CO Eigenlijk ben ik best tevreden over de ICT-systemen, alleen ik verlies er soms het overzicht door s7 CO De informatie is zeker veel. Daarnaast moet ik ook vaak zoeken om alles bij elkaar te krijgen, ik vind dat C1000net een betere structuur kan krijgen s1 NO De ad-hoc verspreiding via C1000net is niet ideaal s2 NO Dat vind ik frusterend, want mijn medewerkers weten niet wanneer ze iets kunnen verwachten en weten niet wie er actie op heeft ondernomen s2 NO Ik vind de hoeveelheid informatie die via C1000net wordt verspreid fors. Het is en onverwacht en vrij veel s2 NO Het type informatie dat je krijgt verschilt nogal. Je het dagelijkse informatie die en incidenteel, vernieuwd is. Dat vraagt natuurlijk nogal wat van ons s3 NO Met de PDA's kunnen mensen zien wat het besteladvies zien en dat controleren op basis van wat ze in de winkel zien en wat ze denken dat er morgen gebeurt s2 QU Als je kijkt naar de inhoud van informatie via het systeem, dan is er niet altijd wat verbeterd aan de inhoud s2 QU Het is oude informatie in een nieuw jasje s2 QU Als je kijkt naar de inhoud ben ik wel tevreden. Maar als je kijkt naar het proces, daar heb ik wel wat opmerkingen over s3 QU Qua IT vind ik dat je mee moet met de laatste ontwikkelingen om aansluiting te kunnen vinden bij andere systemen

LVI

Appendixes

Content analysis Subject Code quote s3 QU Negetig procent kun je door de bestelmodule laten doen en 10 procent kun je dan nog toevoegen s4 QU Ik ben over het algemeen tevreden met de informatie die we krijgen, dus ik heb niet veel moeite om hiermee om te gaan s5 QU Wat ik alleen jammer vind is dat de informatie soms niet klopt. Dan denk ik vooral aan voorraadniveaus waardoor bestellingen verkeerd kunnen zijn s5 QU De informatie uit de systemen blijkt niet altijd te kloppen. Dit kan resulteren in foute bestellingen s5 QU Eigenlijk ben ik best wel tevreden over de informatiesystemen. s6 QU De meeste verkeerde bestellingen komen doordat de informatie niet klopt s7 QU Ik heb wel het gevoel dat ik alle informatie beschikbaar heb om een beslissing te maken. Helaas is de informatie niet altijd even precies. Vooral de voorraadinformatie klopt niet altijd

LVII

MSc thesis M.J. van Strien

Appendix VII: Quotes systems designers Table 38 Quotes from interview with C1000 Retail designer

Liesbeth Snabel: (Functional application manager C1000 Retail) Functioneel applicatiebeheerder van winkelautomatisering, daarna analysist informatievoorziening Nu service delivery manager van winkelautomatisering Aanspreekpunt voor business (retail) voor de IT diensten, niet op requirements gedeelte, maar meer op beheersgedeelte Rol C1000 IT: Leveren software, maken software (stopt), beheren backoffice, gebruikersondersteuning Retail expert centre contactpersoon voor IT wat doen en willen ondernemers. Betrokken bij IT projecten, functioneel ontwerp, requirement bepalen Installatie en implementatie van winkelautomatisering doet BASIS (dochter van C1000) Software over de winkels heen is standaard Zonder inmeninging van C1000 IT kan de ondernemer niets op de winkelclient installeren Wij zijn een van de weinige organisaties met een eigen ontwikkelorganisatie, wij willen gaan naar uitbesteden Voor mij is IT een middel om de processen goed te laten lopen. An sich helpt IT met het effectiever en efficienter uit laten voeren van proicessen Het is de informatie die je eroverheenstuurt dat het hem doet. IT kan hierbij helpen Het belangrijkste van de IT vind ik dat het werkt, de beschikbaarheid IT is een randvoorwaarde, dat werkt gewoon Door automatisering ga je af en toe processen veranderen, of dingen afdwingen in een proces De besteladviesmodule beloofde om tijd te gaan besparen, maar dwingt wel om alles bij te houden. Je moet je voorraden bijhouden, parameters instellen. In het verleden hadden ondernemers nog veel vrijheid in het systeem. Hierdoor werd het lastig om te beheren voor automatisering. De software was wel hetzelfde, maar de configuratie was anders We zijn duidelijk geworden in dingen die kunnen, moeten en mogen. Daarmee willen we veel meer dingen centraal oppakken De systemen hebben gevolgd wat de policy was. En dat was dat ondernemers de vrijheid hebben, die kun je alleen adviseren Er zijn nu 1200 artikelen gedefinieerd waarvan wij niet willen dat ze de prijs aanpassen. Die worden 's nachts automatisch terug gezet De IT systemen worden aangepast en er worden stukjes bijgemaakt Ik vind dat de beslissingsbevoegdheid van de ondernemers blijft, maar dat wij het makkelijker maken om niet meer te focussen op het detailgeneuzel Er zitten mensen centraal die overzicht en kennis van zaken hebben. Daarvan wordt verwacht dat zij het beter begrijpen dan de ondernemers De informatie die je voor het bestellen nodig hebt, de voorraad dus, kan niet uit de winkel. Maar vervolgens, dus wat hij moet gaan bestellen, dat kunnen we beter centraal Er wordt gekozen tussen opties waarbij er adviezen worden gegeven en die nog aangepast kunnen worden, of de bestelling plaatsen en de ondernemers er niks aan laten doen

LVIII

Appendixes

Nu moet de ondernemer zelf nog aan alle knopjes draaien Bijna elke ondernemer heeft de besteladviesmodule tot zijn beschikking Een belangrijk systeem is het C1000net. Dat is het portaal voor alle ondernemers Dat maakt het makkelijker voor de ondernemer, alhoewel er staat wel heel veel informatie op. Je hebt de winkelclient. Daar staat de backoffice of, en kassarapportagessystemen. Daarnaast staat er Exact voor de inkoopfacturen op Uiteindelijk zijn alle processen door ons systeem ondersteund C1000 is zoveel mogelijk ingericht op bundeling van de informatie. Daardoor is het wel gebundeld, maar wel erg veel Er staan veel berichten op waarbij er vanuit C1000 correcties worden gedaan richting de ondernemers Ze zijn er op zich tevreden mee, maar er zijn wel een paar punten voor verbetering. Bijvoorbeeld dat ze per bericht aan willen geven dat het gelezen is en eventueel actie op ondernomen is Je hebt dagberichten en weekberichten. Dagberichten zijn meer de urgente correcties en operationele dingen. Heel eerlijk gezegd zijn die dagberichten meestal dat wij centraal iets fout hebben gedaan Ik heb niet eens tijd om ze allemaal te bekijken, het zijn er zo veel Er is een categorisatie. Je hebt algemeen, en dan ook meer gericht op de afdeling zoals AGF, vers, etc. Weekberichten zijn de wat minder urgente problemen, die vooral gaan over trends en artikelbeheer Daarnaast is er vanuit IT nog een nieuwsbrief die ondernemers op de hoogte stelt over ontwikkelingen, instructies en de status van projecten Wat de ondernemers aangeven is dat zij vaak informatie bij elkaar moeten verzamelen. Het is heel veel, maar het moet ook tijdig zijn. Er gaat ook nog best veel via de mailk Het is heel veel informatie waarin ze toch iets moeten koppelen Dat komt misschien wel door de werkwijze hier, dat je hier gewoon afdelingen hebt die afzonderlijk werken. Dan krijg je blokjes. De nieuwsportal is ingezet om de informatie voor de ondernemers te bundelen Alleen het is gewoon hartstikke veel, dat vind ik wel Vroeger hadden de visie dat we alle informatie moesten brengen. Nu moeten ondernemers vaak zelf op het systeem kijken en dat vinden ze lastig Er gaat meer rapportage komen over de prestaties van de ondernemers en de totaalperformance van C1000. Hierdoor worden afgesproken SLA's gemonitord Door het STORE project heeft de ondernemer SLA's met verschillende afdelingen (retail, marketing) en daarop komen KPI's die gerapporteerd worden Dit zie ik meer als een verantwoording vanuit C1000, niet echt iets waarop de ondernemer kan sturen De operationele besturing doet de ondernemer. Daar krijgt hij informatie voor, voornamerlijk uit C1000net De extra aansturing vanuit tactisch niveau komt ook vanuit de ondernemer. Dat is het aansturen van zijn eigen onderneming op de situatie waarin hij zit De vertegenwoordiger gebruikt de BI systemen om rapportages uit te brengen richting de ondernemer. De vertegenwoordiger geeft aan hoe de ondernemer scoort ten opzichte van de norm. Uiteindelijk is de ondernemer zelf verantwoordelijk, maar omdat hij in de formule zit hebben wij ook wat te zeggen Onze beslissingen over de formule halen wij uit informatie die we uit de winkel krijgenl, op basis van postdata Vervolgens stellen wij normen op die door worden gegeven aan de ondernemer. De prestaties van de ondernemer worden langs de normen gelegd en daarop worden actiepunten uitgezet

LIX

MSc thesis M.J. van Strien

Voor mij zijn die BI systemen ervoor om de informatie die we nodig hebben te ontsluiten. En dat hoeft niet realtime Beslissingsruimte van de onderneme ligt in keuzes op assortiment, bestelling, personeel, personeelsplanning, marketing, promoties Wij geven een landelijke folder uit, maar daarnaast kan de ondernemer kiezen om extra acties te doen Voorraad wordt qua afzet niet real-time verwerkt. Voorraadmutaties van de afzet, dus wat langs de kassa gaat, worden gelijk verwerkt in de voorraden. Met als gevolg dat je niet gedurende de dag de voorraadniveaus kunt zien Goed voorraabeheer is zo belangrijk bij bestellingen en het voorkomen van overstock en out-of-stock De verantwordelijkheid voor de informatiekwaliteit ligt volledig bij de ondernemer, hier kunnen wij verder niet heel veel aan doen Een ontwikkeling in BI is dat we veel meer gaan sturen op basis van de informatie vanuit de winkel, terwijl we het vroeger deden op informatie vanuit onszelf

Table 39 Quotes from interview with Plus Retail desinger

Bert van den Brink (head application management Plus Retail) Ik denk dat ondernemers bij ons nog autonomer zijn dan bij C1000. Plus is een cooperatieve organisatie, de ondernemers zijn aandeelhouders van de organisatie We zorgen er wel voor dat alle ondernemers met dezelfde informatiesystemen werken Wij kennen voor onze ondernemers drie hoofdsystemen.: personeel & efficiency; goederen & assortiment, en finance. Daarmee ondersteunen wij de winkels Deze systemen worden ingezet om de processen efficient en effectief te laten verlopen In het personeelssysteem zitten alle regelingen waar een ondernemer rekening mee moet houden. De CAO's zitten er bijvoorbeeld in verwerkt IT ondersteunt de ondernemers door ze zoveel mogelijk werk uit handen te nemen. G&A is hier een goed voorbeeld van, dit zorgt dat er automatisch een besteladvies wordt gegenereerd Mensen kunnen dit alleen behappen door hun ervaring G&A zorgt voor 90-95% van de bestellingen. Dit neemt de ondernemers heel veel werk uit handen. De 5% die dan nog moeten worden gedaan is afhankelijk van het weer en dergelijke We proberen emoties van de ondernemer eruit te halen door op basis van historische gegevens advies te geven Op die manier hebben we ervoor gezorgd dat het voorraadniveau lager is, en dat er minder out-of-stock is. Dus beter resultaat, terwijl het minder werk kost Er wordt ook gezorgd dat de bestellingen altijd op tijd de deur uit gaan, het systeem dwingt dit af Er zijn een paar ondernemers die het niks vinden dat het systeem het overneemt. Maar het merendeel vindt het heel fijn dat ze tijd overhouden voor andere dingen Wij gaan altijd op stage bij de ondernemers. Hierdoor kunnen wij goed zien waar ze mee bezig zijn, welke dingen daar spelen. En bovendien kunnen we hen ter plekke supporten met bestaande systemen De systemen zijn geintegreerd. Als bijvoorbeeld iemand een pot jam laat vallen wordt dit in G&A geboekt en dan automatisch in de administratie (finance) verwerkt Dat is wat we beogen met IT, of we dat halen is altijd de vraag. We zien aan de servicedesk wel dat de indicenten afnemen G&A hebben we in 2008 geimplementeerd Het is een heel geintegreerd systeem. Dat heeft als nadeel dat bij alles wat we doen, we heel goed na moeten denken of het impact heeft op andere functionaliteit

LX

Appendixes

Daarvoor hebben we een veel meer rigide change procedure opgesteld. Nu zijn we bezig om de achterliggende architectuur vast te leggen, om het makkelijker te kunnen testen De laatste twee jaar zijn we veel meer bezig met het professionaliseren van de IT organisatie. De laatste tijd is testen veel belangrijker geworden Je moet wel omdat IT essentieel is geworden voor de bedrijfsvoering. Er gaat hier bijna niets meer zonder dat IT er bij komt kijken De meeste afspraken met leveranciers zijn gecentraliseerd. Zo proberen we de sinergien op te zoeken om concurrerend te kunnen zijn Het regionale DC komt aan de deur, diepvries DC komt aan de deur, de bakker komt aan de deur, dan komen er nog wat lokale leveranciers. Er komen 5-10 leveringen per dag binnen Vroeger was dat het dubbele, dat zijn voordelen die behaald worden met de centralisatie We vragen leveranciers om een digitale pakbon aan te leveren bij de levering. Die pakbon wordt dat ingelezen in het G&A systeem Het komt niet zo heel vaak voor dat informatie over de voorraad in ons systeem niet klopt We hebben een aantal alerts daarvoor ingericht Bijvoorbeeld als we verwachten dat we vandaag 8 van dat artikel verkopen, en we merken dat er maar 3 verkocht worden. Dan kan het goed zijn dat het schap leeg is. Daardoor wordt een telopdracht aangemaakt Ook als een artikel verkocht wordt terwijl er volgens het systeem geen voorraad meer is, dan wordt er ook een telopdracht gemaakt Er worden per dag zo'n 50 tot 100 gestuurde telopdracht voor een supermarkt gegenereerd Daarmee worden ook diefstal, derving en breuken gemonitord. Dat zorgt voor een goede voorraadprestatie. We merken dat de prestatie hiervoor wel is verbeterd Als er afwijkingen op de geleverde bestelling zijn, dan kunnen ondernemers dat handmatig wijzigen als er geen goede pakbon aanwezig is Als de pakbon een bepaalde tijd te laat is, dan wordt de bestelorder als achtervang voor de geleverde artikelen genomen Ik denk dat een heel verstandig onderzoek is, het onderwerp is wel iets dat duidelijk speelt binnen de retailorganisaties. Ik denk niet dat het voor ons heel veel nieuws zal brengen Ervaringen is heel belangrijk voor de ondernemers. Hierdoor kunnen ze goed hun filters zetten Maar ik geloof wel dat er altijd een overload is. Dat kan op hele kleine dingen zijn, maar er wordt zoveel informatie gepusht Wij hebben PlusWeb. Daarover gaan heel veel berichten heen-en-weer, eigenlijk veel te veel Daarom hebben we er nu een redacteur opgezet. Hierdoor kunnen mensen niet meer zomaar berichten erop plaatsen, maar de redacteur vraagt altijd wat bedoel je ermee En dan nog gaat het niet altijd even goed. Soms goed er een bericht dat ze de promotie zelf in het systeem moeten zetten. Dan zou het wel handig zijn als er gelijk bij zou staan hoe ze dat doen Dat proberen wij ook met die redacteur te bereiken, maar dan nog gaat het vaak fout Het gaat eigenlijk zowel om push-berichten die ondernemers moeten filteren, maar ook gerichte berichten. Maar er wordt zeker wel nagedacht over wie het bericht moet ontvangen PlusWeb is een webportal waarop vooral berichten richting de ondernemer worden gestuurd. Ze kunnen hierop aangeven dat ze het gelezen hebben We hebben er in ieder geval over nagedacht hoe we zoveel mogelijk werk kunnen besparen. Bijvoorbeeld

LXI

MSc thesis M.J. van Strien

pakbonnen moeten automatisch worden verwerkt Maar als het gaat om berichten die we versturen hebben we niet expliciet rekening gehouden met information overload. Gelukkig hebben we wel de goede ondernemers. Die zeggen gewoon je moet eens ophouden met die spam Een informatiesysteem neerzetten en dan maar denken dat dat voldoende is dat werkt niet. Je moet het product alsware verkopen Het gebeurt ons ook dat we een project opstarten omdat de klant daarom vraagt. Dat wordt ontwikkeld en een hoop geld in gestoken, maar soms is er dan geen draagvlak voor G&A is maatwerk en het warehousemanagementsystem ook Informatiesystemen sluiten aan bij wat ondernemers willen. Maar we hebben 270 bazen, het is moeilijk om iedereen tevreden te stellen Het systeem is vanuit meerdere functies (afdelingen) te gebruiken, maar qua interface willen we wel veel standaard hebben Ik denk dat we alleen relevante informatie tonen, dus highlighten is niet nodig

LXII

Appendixes

Appendix VIII: Empirical Model

Figure 4 Empirical model

LXIII