Business Intelligence

(1)

(2)

Robert Poortman

Studentnumber: 1479210

robertpoortman@gmail.com

28 February 2007

RijksUniversiteit Groningen, the Netherlands Faculty of Management and Organization

Master of Science Business Administration, Business & ICT Supervisors:

Dr. E. Harison Faculty of Economics RijksUniversiteit Groningen Drs. R.M Oosterbeek

Solution Center SAP BI & Analytics Company X Utrecht

Business Intelligence

(3)

PREFACE

This thesis represents my research on the success and failure factors regarding BI-system implementation and thereby the end of the Masters program in Business and ICT at the

University of Groningen. This research is performed at Company X in Utrecht within the time scope of five months.

It was hard for me to exchange the city of Groningen, where I was born, grow up and have all my friends, for Utrecht but it was definitely worth it. I realized that the best opportunities for an internship and a career in ‘business consulting’ lies in the middle of the Netherlands. Deep research provided me with the information that Company X is market leader in Business Intelligence (BI) system implementation and is also quoted in the top 5 ‘best employer’ and even the best employer in the IT-branch in the Netherlands. Therefore I would like to thank the organization of Company X for giving me the opportunity to conduct research with full support, fully access to their resources and to deploy myself in the discipline of Business Intelligence. I hope this research will contribute to the success of BI-system implementation and will be a foundation for further research.

Personally, I would like to thank Rene Oosterbeek (Company X), Niels van der Zeyst (Company X) and Bram Broeks (ex-Company X) for their cooperation and supervision. I also want to thank all the people of Company X and Company Y who dedicated time and effort to this research via meetings, questionnaires and interviews.

Special personal thanks I would dedicate to Dr. Elad Harison for his support and supervision during this research. He provided me with insights which I would never had without him. It was a great and helpful time working with a man like him.

Last but not least, I would like to thank my parents, girlfriend and relatives for supporting me the last two years in continuing studying and finishing my research.

(4)

MANAGEMENT SUMMARY

Business Intelligence (BI) has become an important field in Information System (IS)

implementation, which is pointed out by major IT research and advisory organizations. These BI-system implementation systems give organizations better insight in their information and improve the decision-making process (Lönnqvist, 2006). Although many articles are written about the success of Business Intelligence and BI-system implementation, a direct link is never been made, until now.

BI adjusts its former technical approach towards a more human-oriented approach. Although the technical aspect (e.g. data-warehousing, data-mining and extraction tools) is still visible in the current approach, sharing data and supporting decision-making in a non-technical way are now integral parts of BI. With the influence of KM not only structured data but also non-structured data are included in the decision-making process. Therefore, BI is transformed from a tool to an approach.

Business Intelligence is based on data and with help of intelligence the information is

transformed to support the DM process. In this way, BI is defined as an approach that applies a set of tools to extract visibly presented information out of data to provide decision-makers the intelligence and ability to make solid and non-intuitive decisions to gain competitive advantage. Different process-models indicate that BI-system implementation is systematic and needs-driven. Most common motivations to implement BI-systems are: the increasing quality of information, better observation of threads and opportunities and the growth of available knowledge.

These elements of success are defined as constructs of success which are development success and product success. Development success covers the opinion of the contractor which is expressed in cost, time and whether the implementation meets its previous defined

specification. Product success is elaborated according to the D&M-model. The D&M model relies upon seven success categories. System quality (reflects the technical and performance oriented success), Information quality (usefulness and relative importance of the output produced by the system), Service quality (the manner in which service is provided and received), Use (variety and frequency of use), User satisfaction (how well the information need of users is satisfied to their perception), Individual impact (to what extent the system changes impact on user performance) and Organizational impact (to what extent the system improves previous set organizational-wide targets).

BI-system project implementations are measured but in a minimum way (Watson, 2006). The critical success factors (CSF) of BI-system implementation are brought together in

measurement categories and placed into a measurement model (figure A). This measurement model gives insight in important BI-system implementation factors and is used to measure the success of a SAP Business Warehouse (BW) implementation at Company Y with help of consultants of Company X.

Construct Category Element

(5)

2. Flexibility 3. Integration 4. Accessibility 5.Timeliness Product Service quality 1. Tangibles

2. Reliqbility 3: Responsiveness 4. Assurance 5. Empathy Product Use / intention to use 1. Use frequency

2. Use variety Product User satisfaction 1. Content

2. Accuracy 3. Format 4. Timeliness 5. Ease of use Product Individual impact 1. Task productivity

2. Task innovation 3. Customer satisfaction 4. Management control Product Organizational impact 1. Operation cost reduction

2. Staff reduction

3. Overall productivity gains 4. Increased revenues 5. Increased sales 6. Increased market share 7. Increased profits 8. Cost benefit ratio 9. Increased work volume 10. Product quality

11. Contribution to achieving goals

Development Costs

Development Time

Development Meet specification Figure A: Summarized measurement model

Company Y is a company which operates in the energy market. The firm is market leader in delivering energy in the Netherlands. Two SAP BW systems are implemented and support the management with information about the holding and about the service and sales process. Revenues are the primary Key Performance Indicators (KPI).

Company X is a large international IT-company that specializes in three disciplines: consulting, technology and outsourcing and played an important role in both BI-system implementation processes at Company Y.

The methods being used in this research are interviews and two questionnaires that were enhanced after the results from the interviews were assimilated. The questionnaires (for Company X and Company Y) were formulated out of the measurement model, which was derived from the literature (figure A).

The questionnaire is divided into two parts. The first part consists of measuring the product itself in categories as: information quality, system quality, service quality, use, user

satisfaction, individual impact and organizational impact. The second part measures the implementation process in the development construct. This construct is divided into the categories: costs, time and whether the implementation meets its specification.

These questionnaires were sent to Company Y and Company X to determine whether both BI-system implementations completed successfully. The questionnaires are primarily filled in by end-users of the systems and managers who mainly use the systems to support their day-to-day activities.

(6)

The impact of the BI system on the individual and on the organization was assessed as negative. The system for the holding only gained market share and improved the quality of the product, but the overall productivity, staff-size and revenues decreased since the

implementation of this system. The system for service and sales scored well on productivity, costs and quality of the product. Nevertheless, revenues and market share were disappointing. The results of the second part of the questionnaire (i.e. development construct) differ from each other perspectives. According to Company Y, the BI-system implementation did not meet its previous set up specification. Company X contradicts that statement but agrees with Company Y that the implementation process did not finish in time and within the planned budget.

(7)

CHAPTER 1: INTRODUCTION

This chapter is the foundation of the thesis and in this chapter the structure is described. To clarify the purpose of the research questions, the motives for research and the conceptual framework will shape clarity.

1.1 Motive

Major IT-research and advisory organizations (e.g. Company X, Atos Origin, and Accenture) point out that in the last decade Business Intelligence (BI) has become an important field in Information System (IS) implementation. In particular the last five years BI has gained ground. Deploying BI systems is now one of the top three priorities of CIOs1. Gartner also pointed out that the amount of companies using or intent to use BI software shall yearly increase with 6.5% in Europe till 20092_{. To achieve successes like: better quality information,}

increased efficiency and better observation of threads and opportunities, a careful and valid implementation is required. But what makes an implementation valid or successful? This is one of the most important factors this thesis will investigate and explain and is written to help recognizing the success factors of implementing BI-systems.

But not only pay attention to the factors, to indicate the success of the implementation the use of thorough measurements are given to support this success. Clients can be showed the success of the BI-system implementation and this success can be used to secure even more future clients for BI-system implementation. There is a lack of hard evidence of the success of BI-system implementation (Watson, 2006) and this thesis will provide a framework for

proving a successful implementation. In that way organizations not only sell a vague IS but a well based and substantiated BI-system.

1.2 Problem statement

The research objective is to get a clear definition of BI and what BI does in the environment of IS implementation. At the moment, no common accepted definition of BI is available. Every organization and even different parts within an organization use separate definitions for BI. IT emphasis a more technical line of approach and the business side of an organization uses a more managerial definition. In the current situation, one coherent enclosed definition is missing.

With a coherent definition of BI, it is easier to understand the purpose of BI-system implementation. Every implementation has got factors which have influence on the

implementation process. To indicate a successful implementation of BI-systems, success and failure has to be taken into account supported by measurement results.

Therefore, the research question stated as follows:

• What are the factors affecting the success and failure of BI-system implementation?

1_{Mobile technology is the fastest growing area of IT expenditure. By: Saran, Cliff. Computer Weekly, 4/19/2005, p32-32, 1/3p} 2_{Gartner Says Business Intelligence Software Market to Reach $3 Billion in 2009}_:

(10)

Inside this research question, several parts need to be investigated (e.g. what is a successful implementation and how can an implementation be measured?). Factors are particular related influences on the BI-system implementation and can affect the implementation process in positive or negative way. These factors contribute to the overall success of the

implementation of the BI-system. They support or hinder the chance to show the measurements contribute to the overall success of the implementation project. This will result in some other major questions:

• What factors affect the success of implementing BI-systems?

• How can organizations measure if the BI-system implementation is successful?

This covering research question and its major questions will lead to several practical research questions and sub questions:

1. What is the definition of Business Intelligence?

2. What motivates firms to implement BI Systems?

3. What is the definition of success in implementing BI Systems?

4. What are the attributes of the implementation leading to success, partial success or

failure?

a. What are the factors/attributes leading to success?

b. In which cases and conditions success/failure can emerge?

c. Is there a difference in perception between business people and ICT-people

and does this affect the success of implementing BI Systems? 5. How can success in implementing BI Systems be measured?

1.3 Conceptual framework

The conceptual framework defines the user’s conceptions of the research. It is represented via written description and visual representation of the relationships between the different

components of the research. This conceptual framework is based upon the problem statement and showed in figure 1.1.

(11)

Figure 1.1: Conceptual framework

1.4 Research methodology

De Leeuw (2003, p70) makes a distinction between scientific research and research studies. The used methodology in this thesis is based upon scientific research. Knowledge in general will be realized by scientific research and scientific research will contribute to knowledge (figure 1.2). This research starts with gathering data from academic resources like journals. Based upon these findings a theoretical framework is formulated and is used to compare the theoretical findings with a practical situation (the case study in chapter 6). The results from this comparison give feedback to the theoretical conclusion and knowledge is acquired. This feedback will answer the main research question.

Although this research is a scientific research, it has some links with the research study (de Leeuw’s distinctions): the organizations are non-fictive and have direct benefits from the results of this research because this research gives suggestions of the factors on which success depends. Especially the case study will give organizations a feeling of recognition and they can easily understand the success factors in this research in their organizations.

Figure 1.2: Scientific research

(12)

Continuing with the methods how answers to the research questions will be realised. The definitions of Business Intelligence and success are derived from academic definitions. Several statements are combined and completed with my own findings. Links are made with correspondent articles and parts are added. In this way the definitions are formulated and have a reliable academic foundation. To explain the necessity of measuring success, the next metaphor is appropriate. The success of implementing BI-system implementation is like measuring the success of a football club. If the football club is performing well and having success, success can be seen as the amount of points they have got, the amount of goals they’ve scored, and the attendance etcetera. The amount of points can be measured to count the wins, draws and loses. The amount of goals per game and the attendance per game can be count. These indicators contribute to the success of a football club, but they have to be measured. The same counts for the success of implementing BI-systems. Several indicators (like attendance, goals and points in football) indicate the success of an implementation of a BI-system and have to be measured.

Now back to the thesis, success factors are factors which will lead to success but success (the definition) cannot be verified if theses haven’t been measured. This thesis is not completed without the measurement of success of implementing BI systems.

1.5 Structure

This thesis includes several research questions that are interrelated to each other and are assessed as such. All the research questions are classified in the conceptual framework and addressed in a chronological order. Then, the empirical findings are presented. Finally, conclusions are derived and recommendations regarding Company X’s BI-system

implementation are provided. Below, every research question is elaborated in chronological order per chapter.

1. What is the definition of Business Intelligence?

The definition of BI is derived from several academic resources and described in chapter two. This part of the research covers a desk research and the definition is extracted out of the available academic literature from the internet and books. These resources shape a valid basis for a decent definition. Part of this basis is the difference in vision of business and ICT people. This phenomenon is addressed in a small part of chapter two.

2. What motivates firms to implement BI-Systems?

Before going on to the success of BI-system implementation, the elaboration of the motivation for organizations to implement BI-systems is important for categorizing the success factors of Business Intelligence. Motivation, in this case, is a tendency to expand effort to achieve a goal (i.e. successful implementation of BI-system)3. This is also a literature research and described in chapter two. This paragraph will function as the introduction to chapter three.

(13)

3. What is de definition of success in implementing BI-Systems?

The answer to this research question will be generated also by desk-research. The definition of success will act as the foundation for chapter three.

4. What are the attributes of the implementation leading to success, partial success or failure?

This research question continues with the answer on previous research question and elaborates with the attributes which are leading to success and its emerging conditions. Academic resources are again the basis for the answer to this research question and finishes with the perspective of the difference in perception between business and ICT people. Answer to this question will come up with critical success factors (CSFs).

5. How can success in implementing BI Systems be measured?

(14)

CHAPTER 2: BUSINESS INTELLIGENCE

Information is the most important source in decision-making (DM) in the twentieth century. Having the right information at the right time is crucial to the ability of organization to make reliable decisions (Gibson, 2004). In literature, BI is divided into a product part and a

development part. The development construct is only a small part of measuring BI-system

success. This construct has to be taken into account but the primary focus is on the product construct (chapter 3). This chapter provides a definition of Business Intelligence that is based on a literature survey and particularly on contemporary sources discussing BI and its use (element 1 and 2 in figure 1.1). The background of BI is technical and therefore a technical basis is made in chapter 2.1. This paragraph explains different aspects of BI like data-warehousing and Knowledge Management (KM). This technical bias forms up a BI-process model at the end of chapter 2.1. With help of the process model and the (technical) theoretical background a definition of BI will close paragraph 2.1. The end of this chapter explains what motivates a firm to implement BI-systems.

2.1 Background

In order for an organization to make reliable and well-founded decisions, voluminous data needs to be managed, analyzed and fed into the DM-process. Data-mining in this case is a way to access the stored data and is explained in the next part about data-mining. BI also plays an important role in DM. Analytical tools like OLAP (OnLine Analytical Processing tools) are used to extract information out of loads of data to perform as a well based foundation for every decision made. Due to the large amounts of data, not all the data is structured anymore. And it is encouraged with the fact that human organizational issues are becoming more important in the success of high-tech IT systems. This also results in large amount of unstructured data because of the knowledge sharing focus (knowledge

management). This will result in the following aspects of business intelligence: 1. BI data warehousing & data-mining.

2. BI in decision-making.

3. OLAP-tools.

4. BI and knowledge management.

Those aspects and their sources of importance are elaborated as follows:

BI data warehousing & data-mining

(15)

Data warehousing was developed to gather and process the information that is already stored in existing databases (i.e. historical data from its original sources) so that new facts could be revealed. In that way, the information in the data files is used in a better and faster way. A feature of data warehousing is that the original database is left untouched and that no new databases are created. Information must be derived from these databases without adding new data to it. The information from these databases is copied into the data warehousing system where it is processed and used to identify aspects like market trends, competition alerts or to give more information about customers and suppliers. One method to access information stored in a data warehouse is via OLAP tools.

It is clear that the use of data warehouses is set to grow significantly with help of

understanding the customer to gain competitive advantage (i.e. going from the organizational level to the individual transaction level) (Corbitt, 2003). Data warehouses are always aimed at ‘decision-support’ on a typical subject or user group and function as one place where

information can be found. The original data sources are always accessed via the data warehouse system.

The costs of implementing data-mining tools and data warehousing systems are made in the integrity of the data information. In other words, correct data entry is one of the most

important aspects for effective use of the data warehouse system: a black ant with four feet is technically not the same as a four legged ant with the color black. This analogy means to indicate the same thing: an ant, but the description is different. The data warehousing system will also see this as two different objects. The more of these inconsistencies occur in the system, the more time it cost to locate, identify and correct them. How can a system know you’re meaning the same thing when it is described in two ways?

Data mining access the stored information by using complex statistical analysis to discover the relationships in the data (e.g. credit card companies use data mining to identify and localize fraud). Because of the decreasing numbers of fraud and the complexity of credit card fraud, it is even more difficult to identify incidents of fraud and detection is a multistep process of identifying suspicious transactions via database checks and data mining between sources like: social identification number, address, birthdates etcetera (with help of data mining algorithm).

Another example is to identify the churn-chance of existing mobile phone users. Extracting data from its original sources and identifying links between them will bring up a chance percentage for every user to leave the current provider and change over to an other provider (frequency of outgoing calls, average invoice etc.). Linking those sources will examine individual transactions exclusively. Contextual comparisons can also be made if a transaction occurs in two different places (New York and Amsterdam) on almost the same time.

However, not all the tools are sales focused. Data warehouses are ideal environment for performance measurement, where long term data and their analyses are increasingly important.

BI in decision making

The history of DM shows that decisions do not necessarily depend on perfectly calculated rational considerations. Complex circumstances, limited time and growing amounts of data are major factors that affect the DM-process (Buchanan, 2006). It is believed that, for the BI of an enterprise, only 20% of the information can be extracted from the formatted data stored in relational databases4. The remaining 80% of the data is hidden in unstructured or

(16)

structured documents. Those findings emphasize the importance of the ability of use and organization of data structures within data warehouses.

Managers of organizations in all levels are dependent on detailed, accurate information when they make decisions. As described in previous part about data warehousing and data-mining, data warehousing can satisfy this need of information not only to renegotiate contract but also to make decent and well-based decisions about future investments and strategy. With help of analytical tools like OLAP, queries can provide managers information about efficiency and effectiveness of every business unit or even on operational level. With decent and valid information available. The chance of a well-founded decision increases, because incorrect input will never result in a correct output. There is no other conclusion possible that there is no room for guesswork when making good decisions and those decisions have to be made on basis of careful data analysis (Corbitt, 2003). Judges have to go deeply into the lawsuit to pass a valid judgment on someone. In DM, a good preparation also applies to a valid decision. BI will help to sort out the relevant information from the irrelevant documents.

To adopt Business Intelligence into DM, at the operational level, automated DM can eliminate the need for human involvement in decisions that are either highly routine (e.g. ordering standard inventory), or so numerous that they prohibit close oversight step by step (e.g. implementing custom-made tools in a large organization). Whether aimed at organizational level issues, advanced analytic applications (i.e. data mining tools) can dramatically improve the consistency and the quality of DM throughout the organization. These decisions are based on quantified costs, benefits and risks rather than mind-guessing and instinct. At the same time, the creation of extra time because of the accelerating decision-making process can be used by managers to focus on more pressing issues. Connecting to the bottom-line, the BI-tools like OLAP in combination with data warehousing systems offer a superior return on investment, especially compared with their non-decision-oriented tools and systems. A recent study indicates that projects involving predictive analysis offered a return on investment (ROI) of 145%, compared with 89% for the non-predictive decision-making5.

Concluding, executives can set strategy, identify key business drivers and know more about the effects of strategic decisions across the organization. Line managers can align their daily DM with the changing market conditions, customer behavior and actions of competitors. Throughout every level within the organization, users can understand why certain decisions are made and predict how their way of acting influences the results where teamwork, targeting and embed solutions, enterprise integration and adaptive analytics are strategic imperatives.

OLAP-tools

OLAP involves the use of queries to investigate predetermined relationships. For example, management may begin with a query that breaks down sales by region in the last three years compared to the investments. This may followed by additional queries that unravel to lower levels (e.g. by grouping sales by different customers and by quarters). This is called ‘drilling-down’. Companies can use drill-down-capabilities to develop profit and loss statements for individual customers and then use this information to renegotiate contracts. With these drill-down-queries, efficiency figures cannot only be acquired on organizational level, but also on individual level. Codd et al. introduced the term OLAP in 1993 as ‘the dynamic synthesis, analysis and consolidation of large volumes of multidimensional data’. OLAP technology can organize data in multidimensional tables called data cubes and provides access to the data warehouse through an interactive GUI (figure 2.1). Some of the common capabilities of

(17)

User interface

OLAP include: multidimensionality, aggregation, drill-down and roll-up (view detailed and aggregated data), and slicing and dicing. The most common types of OLAP technology are Multidimensional OLAP (MOLAP) and Relational OLAP (ROLAP).The differences between the two types concern data processing capability and data currency (Hasan, 2001). In

MOLAP, the data is cleaned, aggregated in multiple dimensions, and uploaded into a data cube periodically. The data is stored in multidimensional arrays, thus the database has precompiled organization and data arrays chat can be accessed directly and relatively fast. In ROLAP, data is aggregated and stored along with relational databases. ROLAP relies on indices to be built on tables for data access. Users generate queries on the fly (i.e. using Structured Query Language), offering more flexibility in query generation and data currency. The decision whether to use MOLAP or ROLAP is affected by terms of time and space. MOLAP uses more space because it duplicates data to increase the speed of the OLAP-server. MOLAP is also best used for users who have ‘bounded’ problem sets (They want to ask the same range questions every day/week, e.g. finance). On the other hand ROLAP uses less space but is relatively slower. ROLAP is best used for users who have ‘unbounded’ problem sets (they don’t have any idea what they want to ask from day to day, e.g. marketing) (Hasan, 2001).

Figure 2.1: Data warehouse and OLAP

The use of analytical tools like OLAP by organizations is often associated with gaining competitive advantage. However, these tools are not only used for achieving competitive advantage but can also improve the consistency and the quality of DM throughout an organization (Arena, 2005). In order to achieve those benefits (i.e. competitive advantage, quality improvement and increasing consistency), usually huge amounts of data need to managed, analyzed and put into the decision-making process, as described above. Data warehouses assist decision making in organizations by incorporating analytical databases and online analytical processing (OLAP) tools into internal processes in the organization. These tools demand provision of tools to end-users in order to use the data warehousing system correctly. Lack of end-user tools often results in failure to achieve benefits (Hasan, 2001). The cause of failure, besides the missing end-user tools, is the technology acceptance, the

perceived usefulness and ease of use6.

BI and Knowledge Management

The difficulty of taking advantage of the existing data and information increases because of its increasing volume. The volume increases because of the transformation of organizations to ‘knowledge-centric’ organizations. Knowledge and correct information is important for

6_{Davis, D.G. 1989. Perceived usefulness, perceived ease of use, and user accepranee of information technology.}

(18)

knowledge-centric organizations and these organizations build their company around the sources of information to acquire knowledge. Knowledge has got a central role in these organizations. In these ‘knowledge-centric’ organizations, a large number of employees need access to a greater variety of information to be effective. The gaining territory of the World Wide Web enhances this problem.7

Knowledge Management (KM) and BI had only a limited common scope in the past. BI has a long history and KM a somewhat shorter history. BI’s primary focus is technological (i.e. extraction, transformation of data). Although KM had a technological focus in the past, KM has a primary focus on information nowadays (i.e. capture, sharing and distribution of information). Another difference is the structure of information. BI deals with structured and quantitative information and KM with unstructured and graphical information (Seeley, 2006). BI and KM have in common is that they both deal with knowledge. Knowledge in KM

context is generally derived from humans and knowledge in BI is derived from data (figure 2.3), previously analyzed by humans. Both BI and KM have human dimensions to their representation. Though, those aspects are rarely discussed in the context of BI.

KM and BI can be used together by combining a set of tools, nicknamed ‘the dashboard’. The dashboard is a reusable set of BI tools that wraps critical systems that store records and data. Technology is the primary explanation to indicate the difference between BI and KM.

However, the two have in common the front-end technology for access and display of content (i.e. portals). Those portals can easily display data, data-derived knowledge or human-derived knowledge. The use of these portals certainly decreases the gap between KM and BI in a technological stand (Seeley, 2006).

Chapter two builds upon the basic assumption that the success of modern enterprises depends on their ability to generate value from available information. This is far more complex as the volumes of available information grow (i.e. internal and external information). Hence, there is a trade-off between the amount of data and the advantage that can be extracted from the information.

Literature provides different definitions of BI. Some scholars define it as a tool (Graves, 2005). Others perceive BI as a technology (Gibson, 2004). Not only the difference in opinions about BI is remarkable, but also the process models of the scholars who share the same

opinion about BI diverge.

Despite the differences in the process models, the similarities are visible (figure 2.2). The Microsoft process model and Viva’s process model are two models which demonstrate the differences and the similarities but could also be replaced by Philips’ BI-cycle-model (Philips, 1999) or Sahgal’s process model (Sahgal, 2005). A BI-process model like Viva’s model starts with observing the data and the needs from the environment8. Philips calls this stage aiming and determines the information needs. After the data is observed and the needs from the environment are visible, data and information from internal and external sources can be collected (also the collection-stage in Philips’ model). The collected data must be filtered because of the distorting of the data localized from many sources. In this analysis-stage, strategic information and intelligence is produced out of the filtered information. With this strategic information, reliable and integer decisions can be made to take action or

communicate changes. Feedback is used to examine and evaluate the decision being made (figure 2.2).

Bennet, A. 2005.Exploring aspects of knowledge management that contribute to the passion expressed by its thought leaders. PhD in Human Organizational Systems. Fielding Graduate University. 132-154.

(19)

Viva’s process model and Philip’s cycle model can be simplified in a model like Microsoft’s process model9_{. This model can be compared to a general process-model for other}

circumstances. Four general phases are visible and can be equate to both previous set models: analyzing data, insight of information, action according to the collected information and measurement of the taken action.

The DeLone & McLean model (D&M-model) provides the analytical framework because it merges the BI process models (Philips, 2006; Sahgal, 2005) with the factors of success that are described in chapter three (DeLone, 2003).

As described above, the definitions of BI are not unambiguous and can be defined as a set of tools or as a technology (Graves, 2005; Gibson, 2004). However, hardly any scholar denies where BI refers to (Lönnqvist, 2006):

1. Relevant information and knowledge describing the business environment, the

organizations itself and its situation in relation to its markets, customers, competitors and economic issues.

2. An organized and systematic process by which organizations acquire analyze and

disseminate information from both internal and external information sources

significant for their business activities and for DM. Decision-making can take place at different levels: strategic, tactical and operational (Gibson, 2004). Each level has its own properties but all fit into the process model and its success factors.

Figure 2.2: Business Intelligence process models

! " !!#$ % ! &

Microsoft’s process model (simplified) Viva’s process model (generic)

(20)

To come back to formulating a definition, BI is not only be seen as a technology as described earlier. BI used to be only a technical tool, but with the impact of KM the BI tools are

gradually shifting towards the human aspects (e.g. decision-making and formulating

strategies). Nowadays, BI can be seen as both a technical and a human relational technique: gaining competitive advantage through processing of stored data to assist employees in DM. The ability to give decision-makers the intelligence is possible because of the BI-system. A decent data warehouse with intelligent tools like OLAP gives decision-makers this

opportunity.

This explains, with confirmation of academic sources, that terms as data, information, intelligence and decision are company Yial for the definition of BI (figure 2.3). The fact is that there is little academic research on BI and every definition of BI is placed in a separate environment of factors. To come up with a valid and consistent definition of Business Intelligence, a solid research that forms up strong links to many valid academic sources is required (Sahgal, 2005; Graves, 2005; Samel, 2006; et al). Therefore, despite existence of current definitions and factors that elaborate them, they all seem insufficient and suggest that new insights on BI are required.

Figure 2.3: Definition of BI

Merging the BI process models (Philips, 1999; Sahgal, 2005) with the academic sources of Graves (2005), Samel (2006), Lönnqvist (2006) and others in line with the analytical framework (D&M-model), provides the next definition of Business Intelligence:

Business Intelligence is an approach that applies a set of tools to extract visibly presented information out of data to provide decision-makers the intelligence and ability to make solid and non-intuitive decisions to gain competitive advantage.

2.2 Motivation for BI

Viva’s process model (figure 2.2) shows that a BI system must be systematic and needs-driven. The BI-cycle starts with planning based on corporate needs (see also Philips cycle-model). If the football players in a team are not motivated or hasn’t got any motivation/need to win the football match, the chance for the team to win the game or to achieve any success is negligible. Also in implementing BI-systems, success is closely related to motivation and the right focus. But what are the drivers behind BI-system implementation? What motivates firms to implement BI-systems? What is the focus of these companies and what are they trying to achieve?

The main priority for companies is to acquire better quality information, information that is integer and valid. The second largest expected benefit from BI-system implementation for companies is to develop their own observation to improve identifying threats and

opportunities. With help of valid information and integer links between different types of

data intelligence decision

information

(21)

information (e.g. marketing information and market information), opportunities in marketing policy is particular market segments occur. With this information, the management can better decide how to divide the investment money among the market segments. Not only

opportunities are becoming visible, also threads can be identified faster. Threats from new entrants in a market segment or changes in government policies can be noticed earlier. Unlike the Enterprise Resource Planning (ERP) segment (which BI can be compared to), time-savings and cost-savings are not in the top 7 expected benefits from BI-system implementation. Only 30 percent of the interviewee feels that time-saving is the most significant benefit provided by BI-system implementation and 14 percent see cost-saving as the most significant benefit. The most significant benefits are: growth of available knowledge (76%), sharing information (73%), improved efficiency (65%), easier information acquisition and analysis (57%) and faster decision-making (52%) (Thomas, 2001; Hannula, 2003). Concluding this chapter, BI adjusts its former technical approach towards a more human-oriented approach. Although the technical aspect (e.g. data-warehousing, data-mining and extraction tools) is still visible in the current approach, sharing data and supporting decision-making in a non-technical way are now integral parts of BI. With the influence of KM not only structured data but also non-structured data are included in the decision-making process. Therefore, BI is transformed from a tool to an approach.

Business Intelligence is based on data and with help of intelligence the information is

(22)

CHAPTER 3: SUCCESS OF IMPLEMENTING BI-SYSTEMS

It is important to define the concept of success to identify the factors which are leading to a successful implementation of BI-systems. The introduction of this chapter is composed to fulfill this aim. Chapter two provided a definition of BI, but this chapter will not discuss the success of BI but the success of system implementation. In that way, a definition for BI-systems is required. BI is defined in chapter two as an approach that applies a set of tools to extract visibly presented information out of data to provide decision-makers the intelligence and ability to make solid and non-intuitive decisions to gain competitive advantage. BI-systems are not only BI-systems, which fulfill to this approach, but can also be a part of a system (i.e. a tool to extract information out of data). After constructing the definition of success, the factors which contribute to the BI-system implementation success are set-out and de

conditions in which success emerge. This chapter will end with differences in perception between business-people and IT-people (element 3 in figure 1.1) and how this difference is affecting the success of implementing BI-systems.

3.1 Success

In chapter two, the definition of BI has been created to give this thesis a valid foundation. The definition of success will be the basis for this chapter. It is inconceivable to start with the factors leading to success without a clear definition. In general success is being described as: ‘an event that accomplishes its intended purpose’10. To be more specific and apply success in the world of management and organization, success can be defined as: a comparison whether

the targets which are set in the beginning of the project have been met (Lönnqvist, 2006).

Success of ICT-projects and even the success of this thesis have been set-out by

pre-formulated goals and targets. But to interface Business Intelligence, success is more uncertain and far more specific and dependable of different factors. These pre-formulated targets only evaluate the success of the development construct and not the success of the product itself. But in BI-system implementation, the development process is at least as important as the product itself. The definition in chapter two of BI confirms this statement: “…data to provide

managers the intelligence and ability to make integer and non-guessing decisions to gain competitive advantage”. For example: the opinion of the users is not part of the success of the

product, but is part of the success of the process of the project. The project is not a black box or a static entity, but an empiric object which interacts with users.

What is successful implementation?

It is difficult to explain how implementation becomes successful. This explanation is hard to identify because success is relatively subjective. A coach of a football club sees success as winning games and gaining points. A physiotherapist of the same team sees success as getting no injuries after the match and the owner of the club sees success as getting more profits than last year. With the success of BI-system implementation (or in this case a football club), a lot of parties are involved and do not share the same opinion (i.e. success). Lu (2006) argues that there is no general agreement or an absolute indicator that defines successful IS

implementation. Evaluation of the implementation project is never a totally rational judgment, independent of the situation or the stakeholders in the project. Evaluation and considering the

(23)

success of BI-system implementation is a subjective thinking process, where the relevant evaluators make comprehensive considerations based on understanding of firms, market, techniques (i.e. experience), perception and interpretation such as user satisfaction and ease of use.

Success constructs

Following its definition, success exists in both product success and development success. In academic research, different writers confirm this subdivision (Wateridge, 1998; Ogilvie, 2001; Watson, 2006). These constructs are also used to define the success of BI-system implementation (figure 3.1). The first construct (i.e. development) covers the common assessment of success of IS where project managers have to aim for satisfaction of cost, time and user specification to achieve success.

• Development cost: the cost of developing/implementing and maintaining the

BI-system should be appropriate and within budget.

• Development time: the time of developing/implementing the BI-system should be

appropriate and within budget.

• Development meets its specification: the development and implementation of the

BI-system is in accordance with the in advance formulated specification.

At the end of the BI-system implementation process (or interim), process evaluation and product success is measured according to whether the project budget is not exceeded, the implementation process succeeded in time and the specification is in accordance with the expectation of the party involved.

However, these three criteria are primarily a view of the contractor and other views (e.g. user, consultant) are not taken into account in this construct. In addition to this success construct, several attributes (i.e. service, quality, use, impact etcetera) are not taken into the

development construct but are important factors for BI-system implementation success (DeLone, 2003; Watson, 2006).

Nevertheless, service, quality, use and impact are attributes that also contributes to the success of BI-system implementation and are classified under the construct product (figure 3.1). Instead of the development construct, the product construct does not evaluate the results of the project (i.e. cost, time and specification), but is closer related to operational user satisfaction, quality and impact. Watson develops Wateridge’s framework adding these attributes to the product construct. Remarkable here is that these constructs are similar to the categories of the D&M-model (figure 2.2). Every attribute of the product construct according to Watson (2006 and DeLone (2003) is explained briefly below. The product construct is more elaborated in chapter five.

• System quality: system quality reflects technical, performance oriented and

engineering criteria and is focused on the desired characteristics of the BI-system which produces information.

• Information quality: the quality of information covers the output produced by a system

and the value usefulness or relative importance attributed to it by the user.

• Service quality: service quality is the manner in which service is provided as it

influences the degree of satisfaction with a good or service which is tangible or intangible.

• Individual impact: individual impact describes to what extent the BI-system changes

the productivity, tasks of the users and their satisfaction and the control of the management.

• Organizational impact: organizational impact describes to what extent the BI-system

(24)

• User satisfaction: user satisfaction is defined as how well the information need of

users is satisfied to their perception.

• Use: use can be classified into the dimensions use frequency and use variety, whereas

frequency covers the amount of time spend on the system and variety the amount of purposes of using the BI-system.

Figure 3.1: Success constructs

Concluding, the definition of success in implementing BI-systems has got two constructs: development success and product success. Development success reviews the implementation of the BI-system and compares is to its expectation.

The other success construct is the development construct. Development success refers to the creation, use and consequences of the product (BI-system) (figure 3.1).

3.2 Factors leading to a successful BI-system implementation

The success of implementing BI-systems is important to every part and level of the

organization. Failure to implement a BI-system may prove fatal to organizations: either by wasting enormous amounts on the system or by destroying the competitive advantages of the organization.

After defining success, the following question rises: what makes a BI-system implementation successful or in other words; what is leading to the success of the implementation of BI? The factors of success were investigated by academic researchers before and they are described below (Burns, 1991; Hasanali, 2002; Hong, 2002; Wong, 2005; Lu, 2006 et al.). Factors of success or CSFs can be defined as: areas in which results, if they are satisfactory, will ensure successful competitive performance for the organization (Wong, 2005). This definition of CSFs is fully in line with the definition of success that is brought in chapter 3.1. The CSFs of BI include:

1. Management, leadership and support: managers and leaders are important in acting as

role models to exemplify the desired behavior.

SUCCESS

product development

cost -

time - - system quality - information quality - service quality - individual impact - organizational impact meet specification - Watson, H.J. 2006. Watson, H.J. 2006. Wateridge, J. 1998. Watson, H.J. 2006. Watson, H.J. 2006. DeLone, W.H. 2003 Watson, H.J. 2006. Watson, H.J. 2006. - user satisfaction

- use Wateridge, J. 1998. Wateridge, J. 1998.

(25)

2. Culture: it defines the core beliefs, values and behaviors that are shared by the

members of the organization.

3. Information Technology (IT): containing, processing and presenting information with

help of a good underlying IT-infrastructure.

4. Strategy and purpose: providing the foundation for development success and how to

reach the formulated goals and the purpose to follow this strategy.

5. Measurement: giving information about data in a situation to ensure that objectives are

being fulfilled.

6. Organizational infrastructure: Establishing roles and groups within the organization

to perform knowledge-related tasks.

7. Processes and activities: coordinating standardization of processes and activities in

daily work.

8. Motivational aids: establishing the right incentives, rewards or motivational aids to

encourage people to share and apply knowledge.

9. Resources: consists in tangible and intangible resources (finances, labor, time,

attention).

10. Training and education: managing knowledge to improve motivation to share

knowledge with help of training and education.

11. Human Resource Management (HRM): focusing on recruitment, development and

retention to expand, improve and prevent knowledge from leaving the organization. Management, leadership and support has a significantly influence on the chance of success of a BI-system implementation (Horak, 2001; Ribiere, 2003). Leaders are important to

exemplify the desired behavior on the other employees, not only by words but also through deeds. Maintaining morale, steering, creating culture (CSF 2) and communicate the

importance of BI are competencies of leadership. Management commitment is company Yial for a successful BI-system implementation (Davenport, 1998; Jarrar, 2002; Truch, 2001). Without commitment, the importance of BI cannot be carried across the employees and its implementation is more likely to fail.

Organizational culture is another key factor of successful BI implementation (Davenport, 1998; Martensson, 2000). Culture defines the core beliefs, values and norms that govern the ways individuals within the organization act and behave. Culture may have a positive or a negative influence on implementing BI-systems. Adaptability culture (i.e. the ability of culture to change to fit the continuously changing internal and external environment of the organization); will encourage implementation because of the willingness to change. On the other side, bureaucratic culture demands a more stable and internal focus. Hence, BI-system implementation is difficult to carry out in organizations that are characterized by strong bureaucratic culture.

Information Technology is one of the key drivers of BI. Containing, processing and presenting information depends on good IT-infrastructure and it also connects between humans and information (Lee, 2002; Wong, 2003). Important aspects of IT are closely related to the other CSFs: simplicity of technology, ease of use suitability to users and relevancy. Without the awareness of these aspects, BI-system implementation is inevitable to overcome. However, a footnote must be made that IT is only a tool and not an ultimate solution.

(26)

purpose of all employees to follow this strategy and the belief that it works. Clear objectives, (interim) goals and supporting vision are examples of a well planned strategy. All of these elements need to be carefully developed before a BI investment is made.

Measurement is not the kind of measurement elaborated in chapter four, but is the

measurement that will give certain information about data in a particular situation. Measuring BI is necessary for ensuring that its objectives are being fulfilled (Arora, 2002; Ahmed, 1999). Measurement enables organizations to track the progress of the implementation of BI-systems and to determine its benefits and effectiveness. It also provides organizations a basis (like strategy) for evaluating, comparing, controlling and even improving their performance. Measurement also helps to advance leadership and to satisfy stakeholders.

Another CFS for implementing BI-systems is the development of a well based organizational infrastructure. In contrast to IT and HRM, this infrastructure is all about establishing roles and groups within the organization or creates new ones to perform knowledge-related tasks (e.g. knowledge broker: bringing the provider and de inquirer of knowledge together). Examples of typical infrastructural roles are: manager, controller, innovator, leader, information keeper (Davenport, 2001).

Processes and activities as a CSF contains the same attributes (Alavi, 2001) as the simplified BI-process model (figure 2.2). Time to time interventions and capable mechanisms need to be in place to ensure that these BI-processes are placed in a structured and systematic way (e.g. in knowledge sharing, technological networking tools should be supplemented with face-to-face discussion to exclude interpretation mistakes). These structured processes can be seen as standardization of daily work activities and it becomes part of the design of operational management in organizations. Coordination of these processes and activities is crucial to BI-system implementation (Alavi, 2001; Holsapple, 2000).

Another key success factor is motivation. Without motivation, no investment, infrastructure, or technological system will work successfully (Hauschild, 2001). Motivation can be

rewarded by monetary or non-monetary means to convince people to share knowledge. Both benefits can be placed into a reward system to that supports the implementation of BI-systems.

Resources are divided into tangible resources and intangible resources and have an

indispensable role in implementing BI-systems. Tangible resources are resources which are perceptible (i.e. finances and labor). Intangible resources are incorporeal (i.e. time and attention), which means that they have no material or physical existence. Financial resources are required for the purchase of a system. Human resources are necessary to coordinate and manage the implementation process (Davenport, 2001). Time must be scheduled in for the change-process and attention must be made by management to emphasize the importance of the BI-system implementation (Martensson, 2000). These resources are fully in line with Watsons’ success factors in the development construct (figure 3.1).

Training and education, like motivational aids, play an important role in managing

(27)

HRM is closely related to both education and training and resources as critical success factors. The main focus of HRM is on employee recruitment, development and retention. Effective recruitment is crucial because it ensures that knowledge is brought into the organization (e.g. recruitment programs). Finding the right people also reduces the risks of conflicts of cultures and business processes (also CFSs). Employee development will improve and enhance the personal value of the employees (e.g. improving skills and competences). Retention will prevent knowledge from leaving the organization by providing opportunities for the employees to grow and advance their career (e.g. providing a good working

environment).

3.3 D&M model

The D&M-model is presented in DeLone and McLean (1992) and originally named the

Information System Success Model (figure 3.2). Due to changing techniques and upcoming

technologies, like Business Intelligence, the model was revised and updated in 2001. Its revised version, the D&M success model, provides a comprehensive framework for measuring BI and its system implementation success (Qian, 2005; Skok, 2000; Watson, 2006 et al). The D&M-model only aims at the product construct, which is described in chapter 3.1 and leaves the development construct outside (figure 3.1). The D&M-model exists in seven elements. A brief description follows where a detailed description continues in chapter four.

• Information quality – focuses on the information product for desired characteristics

such as accuracy and timelines.

• System quality – focuses on the desired characteristics of the information system,

which produces the information.

• Service quality – the service quality element was added ten years later than the

original D&M-model (DeLone, 2003). The reason for adding this element was the development of organizations toward more service oriented enterprises. Pitt et al. (1995) observed that the current measures (i.e. system quality and information quality) only focuses on the products of the IS function rather than the service of the IS

function (e.g. providing users with support and information).

• Use - focuses on the interaction of the information product with its recipients.

• User satisfaction – focuses on the sensibility of the recipients’ information product and

to what extent they are satisfied with the current information.

• Individual impact – focuses on the influence which the information product has on

management decisions.

• Organizational impact – focuses on the effect of the information product on

organizational performance.

Figure 3.2: D&M-model (source: DeLone, 2003).

(28)

The relations in the D&M model are statistically significant and the creators of this model claim that the D&M-model is a causal model. Causality means that ‘A causes B’ and

increasing the effect of ‘A’ would increase ‘B’ too (or decrease it). For example, if the quality of Information increases, users would be more satisfied and therefore the organizational

impact would also increase.

The relations between the D&M-model and the CSF

Qian (2005) elaborates the links between the D&M-model and the CSFs as follows (figure 3.3): The top layer contains four levels of information system outputs measurement:

functional, technical, semantic and pragmatic. Only the last three levels of Masons’ influence theory (1978) are used in the D&M-model. The reason that the functional level is leaved out is because it was developed in the nineties, IS were only limited to transactional information systems and information processing process was automated by machine. The exclusion of the functional level is superseded because the BI-systems are more and more interrelated with human aspects and not only technical related. This is also the reason for adding ‘service quality’ to the D&M-model as explained earlier and therefore should also be added to the conceptual paradigm (figure 3.3). BI-managers cannot limit their attention to only hardware and software components ignoring the effects of and on the employees (e.g. motivational problems, personal development etcetera). Therefore, the D&M-model is expanded with the functional level, which analyzes information output in term of the processes which produce it. Markus (2001) defines the process of knowledge reuse in terms of four steps: acquiring

knowledge, refining knowledge, distributing knowledge and reusing knowledge. These four

steps can be placed on Viva’s process model (figure 2.2) which is exactly the BI-process model that is explained in chapter 2.1. Acquisition of knowledge covers the first three

elements of Viva’s model (1998): investigating the needs, observing data and the collection of data. After the data is collected unnecessary information is disposed and the valid information is analyzed (Markus’ refinement). Intelligence is generated out of the valid information and safely stored (Markus’ distribution). With help of feedback and (re)utilization, the BI-process is started again (Markus’ reuse).

Mason’s functional level is to analyze how information is produced in information systems. After acquisition and refinement, knowledge is “produced” and ready for use. It means that these two steps of knowledge reuse are placed in the functional level. Knowledge acquisition and refinement are supposed to directly affect the quality of the knowledge stored in

repositories which belongs to semantic level and is represented by information/service quality in the D&M-model. After knowledge is “produced”, the next step is knowledge distribution in which the repository content is made accessible to BI-system users through information technologies such as intranet and database. In this stage, the focus of success is mainly

technical issues, corresponding to DeLone and McLean’s system quality at the technical level. The last stage is knowledge reuse which is oriented toward the consumption of the output of BI-systems, equivalent to use in D/M model. Finally, the consumption of knowledge will have a series of influence on knowledge recipients, such as satisfaction and perceived impact, which belongs to measures in influence level.

(29)

System Quality • Information technology • Measurement • Processes/activities • Resources

Use / Intention to Use

• Processes/activities • Motivational aids Service Quality • Management / Leadership/ support • Culture • Org. infrastructure • Processes/activities • Motivational aids • Resources • Training / education • HRM User Satisfaction • Processes/activities • Motivational aids Individual Impact • Resources • Training / education Organzational Impact

• Strategy and purpose • Measurement • Organizational infrastructure • HRM Information Quality • Management / Leadership/ support • Culture • Org. infrastructure • Processes/activities • Motivational aids • Resources • Training / education • HRM

Figure 3.3: Conceptual paradigm (based on Qian, 2005).

Figure 3.4: CSFs categorized by the D&M-model.

Information quality focuses on the information product for desired characteristics such as

accuracy and timelines. Getting the desired characteristics on place on time requires good management. To acquire support from the employees to guarantee the quality of information, employees have to be motivated to work with the system. Not only motivation is required, but also proper training to perform day-to-day work with support of the system. If the

Functional

level Technical level Semantic level Pragmatic (influence) level

acquisition refinement Reuse distribution Information/ knowledge _User satisfaction Impact Use System Mason, 1978 Markus, 2001 DeLone & McLean, 1992

(30)

organizational culture and infrastructure does not support employees to succeed in the change process (i.e. BI-system implementation), a good foundation becomes a utopia. Coordination of the chance process is necessary to prevent the implementation from slipping away.

Improving quality requires resources. Not only financial resources but also time and attention to allow employees to improve the quality of information. Finally, HRM prevents decreasing quality of information with preventing lose of knowledge and acquiring new knowledge in terms of new employees and elaborating the current level of knowledge.

System quality focuses on the desired characteristics of the information system, which

produces the information. Processes around the BI-system must be structured to understand the underlying structure of the system. Changes to the system can easily be made when processes are standardized. This will result in the fewer technical specialists are required to add changes to the system itself. These adjustments require time, money and attention, which are necessary resources to guarantee certain system quality. Measurement in this category covers the tracking process of data. Therefore it is not part of information quality and service quality, although information and service also should be measured, but not in the way it is intended here.

Service quality can be compared with information quality in this context (see above). Use focuses on the interaction of the information product with its recipients. With

standardized processes and activities, efficiency and effectiveness of use will increase. When standardizing processes and activities, motivation of employees is important. People dislike doing the same work every day.

User satisfaction focuses on the sensibility of the recipients’ information product and to what

extent they are satisfied with the current information. The same CSFs as Use are applicable to this category.

Individual impact focuses on the influence which the information product has on management

decisions. Motivational aids and training and education are company Yial guarantee positive individual impact on employees in terms of efficiency and satisfaction. In that way, their job performance is maximized.

Organizational impact focuses on the effect of the information product on organizational

performance. A well defined strategy and a corresponding organizational infrastructure are formulated in accordance with the goals and targets of the organization. For example: a cost leadership strategy is chosen when the target is profit maximilization. Other factor to receive financial benefits is with a good HRM program (e.g. contract only academics to increase knowledge). Only the right information is acquired with decent management information systems when measuring the organizational impact.

Business Intelligence

Business Intelligence

PREFACE

MANAGEMENT SUMMARY

TABLE OF CONTENTS

CHAPTER 1: INTRODUCTION

CHAPTER 2: BUSINESS INTELLIGENCE

CHAPTER 3: SUCCESS OF IMPLEMENTING BI-SYSTEMS

SUCCESS