Chapter 3: Information Systems as Aids of Managerial Decision Making 3.1. A Priori Assumptions This perspective is based on the following assumptions: • the effectiveness of an organisation is determined by the quality of decisions made; • the quality of decisions depends on the availability of necessary information; • in organisations managers make the decisions; • the mission of information systems is to supply managers with the information necessary for making their decisions. The managerial decision support perspective was born at the beginning of the 1970’s, and has probably been the most popular one until the present day. Its success is shown by the widespread use of the expression ‘MIS’, as referring to a field of research, to the information systems department of companies, to information systems management courses etc. 3.2. Literature Review It seems logical that a comprehensive analysis of the nature of managerial decisions provided the basis for the development of the first information systems aiming at the support of managers. In reality, this happened just in reverse order. In their exceptionally famous study, Gorry and Scott Morton claimed that the failure of the first projects in managerial information systems was caused by a wrong approach to the work of managers and a misunderstanding of the nature of managerial information. It fact, until the end of the 1960’s only rather stereotype answers existed about the required characteristics of managerial information. Just to mention a few, they were expected to be relevant, current, and reliable. This made many researchers and specialists think that it was sufficient to collect information from every department of the company to one central place for the managers to make correct decisions. However, Gorry and Scott Morton called attention to the fact that the information produced in elementary transactions is relevant only for a few types of decisions, in other cases they are completely useless. Based on Anthony’s (1965) model 1, they used opposites to determine the characteristics of ‘good information’ related to the operational and strategic planning/control activities of a company (see Figure 6). From this, they could get a picture of the information requirements of tactical planning and control tasks too. Characteristics of information Operational planning and control Tactical planning and control Strategic planning and control Source largely internal external Scope well defined, narrow very wide detailed aggregate historical future highly current quite old high low very frequent infrequent Level of aggregation Time horizon Currency Required accuracy Frequency of use Figure 4: The required characteristics of information in different planning and control activities (Gorry – Scott Morton, 1971; Gorry – Scott Morton, 1989) The authors also analysed managerial decisions according to Herbert Simon’s (1960) ‘structuredness’ dimension2, during which they introduced the category of semi-structured decisions (problems). 1 Gorry and Scott Morton interpreted Anthony’s planning and controlling categories in the following way: • strategic planning and control: decisions on the objectives of the organisation, on changes in these objectives, and on the resources used to attain these goals; • tactical (or managerial) planning and control: decisions that assure the effective and efficient supply and usage of the resources in the accomplishment of organisational objectives; • operative control: decisions that assure the effective and efficient execution of specific task. 2 The ‘structuredness’ dimension can be understood on the basis of another model of Simon, namely the model of the decision making process. According to this, decision making has several stages: • intelligence: revealing the characteristics of a problematic situation (or of a favourable opportunity); 2 Their conclusions are the following: • Different types of decisions can be supported efficiently by different types of information. • The different types of information can be provided by different types of information systems. Followers of Gorry and Scott Morton tended to connect planning and control activities, as well as unstructured decisions with the top of organisational hierarchy, while linked structured decisions to the bottom level. This lead to the emergence of the so-called pyramid models, which can be considered as the ‘emblems’ of this perspective. In the pyramid models, the different levels of organisational hierarchy are supported by different types of information systems. The pyramid models have strong roots in reality as hierarchical levels indeed separate several corporate IT applications. Moreover, information systems are often isolated by functional boundaries as well, which results in ‘islands of systems’ – the typical architecture of the second half of 1980’s. Finally, the pyramid model seems appropriate also because in most of the cases it is really the lower level systems that provide the necessary inputs of the higher levels. • • design: modelling the problem, then finding, elaborating and evaluating alternative solutions (series of actions); selection: choosing the most appropriate alternative. A decision (problem) is said to be structured if all the three stages of the decision making (problem solving) process are structured. A stage is structured, if its procedures are standardised, its objectives are clear, and its inputs and outputs are defined. 3 EIS Executives Analysts, experts Functional middle managers Employees charged with operative responsibilities DSS MIS EDP/TPS DSS MIS EDP/TPS DSS MIS EDP/TPS Figure 5: A typical information system pyramid of leading companies in the second half of the 1980’s. Although there is certainly a need for input-output relations in the horizontal direction as well, pyramid models do not really emphasise them. A good illustration of the horizontal and vertical segmentation of corporate information systems as well as their vertical interdependence is given by a computer-aided cash desk, which registers payments in a special computer application. The system, similarly to other data processing applications, provides inputs for a program, which creates corporate financial reports. The head of the financial department operating this reporting tool then downloads some data into a spreadsheet, from which he prepares cash-flow forecasts and liquidity calculations. His results, together with aggregated information from the other functional departments, are finally sent to the computer of the chief executive officer. 4 3.3. Related System Categories The typical pyramid model exhibiting the vertical and horizontal segmentation of information systems is shown on Figure 7. The types of information systems displayed in the model can be defined as follows: • The EDP/TPS systems have already been defined in Chapter 2. The diagram clearly shows that they do not play a direct role in supporting managerial work. • MIS (Management Information Systems) aim at providing information for middle managers. They present regular reports in standard form with a certain functional focus. MIS can also be used for solving simple optimisation problems in a noninteractive way (e.g. production planning). • DSS (Decision Support Systems) are designed for problem modelling, simulation, and statistical analysis in semi-structured decision making situations. DSS are characterised by interactivity, a user-friendly environment, and a graphic user interface. It is important to mention that within the broad category of managerial decision support DSS includes group decision support applications and expert systems as well3. • The most aggregated information is provided by EIS (Executive Information Systems). At the end of the 1980’s only a few top executives used these systems, which required huge hardware and personnel apparatus. Early EIS mostly gave critical information on the internal performance of the organisation. 3 Group decision support systems as well as expert systems have been discussed in the doctoral dissertation of the Author in detail and especially from the perspective of managerial decision support (Drótos, 1991). Therefore, in this study no further analysis is provided. 5 DATA Executives WAREHOUSE AND OLAP Analysts, experts Functional middle managers ENTERPRISE RESOURCE Employees charged with operative responsibilities PLANNING (ERP) SYSTEM Figure 6: A typical information system pyramid of leading companies in the second half of the 1990’s. By the second half of the 1990’s, this strong vertical and horizontal segmentation has already disappeared in many organisation. A new structure emerged with only two vertical levels and with horizontally communicating functional elements (see Figure 8). ERP systems (discussed partly in Chapter 2) form the basis of the new pyramid. Horizontally, an ERP system comprise several functional and business areas, while vertically they cover the levels of EDP and MIS. The dotted vertical lines indicate that ERP systems still retain some functional specialisation, although this segmentation is much less rigid than before. The different modules are interconnected, share their database, and business processes can be traced along in them. It can be claimed that in the vertical direction EDP and MIS have almost completely integrated. Every time an elementary data is recorded, each report using this 6 data is overwritten automatically, unless purposely limited.4 Separate data processing systems beside ERP are operated only in special cases. 5 However, in their original form ERP systems could not fulfil the requirements of top managers and business analysts whose demands go beyond simple predefined reports. The reasons for this failure are the following: • in ERP systems ad hoc analyses would take up too much of the hardware’s capacity which is fitted to the predicted number of business transactions; • even if the capacity of the hardware is adequate, ad hoc analyses usually require a lot of programming in ERP systems, and some complex inquiries cannot be conducted at all; • comparisons, which are especially interesting for managers, are not always available because ERP systems often do not contain the necessary plan and historical data, only the actual results; • even if the ad hoc analysis is successful, the performance of the ERP systems are often modest concerning their graphic display, speed, and available statistical instruments. It seems that it is the OLAP (On-Line Analytical Processing) technology that remedies the above problems best. As a core function, the OLAP technology enables complex, quick, and user-friendly inquiries on the databases of ERP and other transaction processing systems. In one form of OLAP systems data are stored in relational databases (relational OLAP or ROLAP), while in an other form data are converted into multidimensional data cubes to prepare the analysis (multidimensional OLAP or MOLAP). These data cubes can be updated every month, every week, or in some special cases every day (in finance, for example). 4 For example, in some companies actualisation of the ledger takes place only at fixed intervals because of security reasons. 5 The most frequent reason is an industry- or company-specific activity, which is not covered by standard ERP system modules. 7 Closely related to OLAP technology there is another fashionable concept, called data warehouse. The data warehouse is an organised collection of data, which also store information about the relationships existing among primary data. The storage can be physical or logical; in the latter case, the data warehouse refers only to the source of data. The standard features of OLAP applications are the following: • multidimensional analysis, building on existing corporate databases; • standard (regular or irregular) reports; • user-driven ad-hoc analyses through the combination of the available dimensions (slice and dice); • what-if simulations and goal-seeking; • exception reports, automatic alarm if pre-set tolerance is exceeded; • drill down (breaking down of aggregated data into their components); • commentary on data; • individually designed user interface, advanced graphical display. In the last few years developers of ERP systems started to extend their services to the whole of the corporate pyramid. They integrate data warehouses and OLAP functions into their information systems, as well as create so-called industry-specific modules. The function of industry-specific modules is to satisfy those unique requirements of certain organisations (public utilities, hospitals, banks etc.), which standard ERP systems do not handle. While ERP systems are expanding both upwards and downwards, developers of the stand-alone OLAP systems wish to increase the intrinsic value of their service. They want to sell complex systems with built-in managerial know-how (e.g. Business Plan, Balanced Scorecard), rather than ‘empty’ packages. 8 3.4. System Development and Implementation During the development of management information systems, the programmers encountered serious dilemmas. Perhaps the most critical of all was how to determine the relevant information requirements. According to Wetherbe (1991), at the beginning three different solutions were used to determine managers’ information requirements. The first solution was asking the manager what he/she needed. Since managers usually could not give a definite answer to this question, the early projects often ended in failure, or devoured huge additional investment because the developed systems had to be changed to fit real needs. These unpleasant experiences have led system developers to ask for the manager’s signature explicitly, proving his expectations. However, this was not enough for success and did not prevent managers from accusing programmers of deliberate deception. The third solution called ‘catalogue method’ did not prove to be a success either. According to this method the developers showed a complete set of reports which the program could possibly generate, and asked the managers to choose from them. However, the answer was generally disappointing: managers wanted them all. More effective means of determining information requirements have been developed since then, of which the Critical Success Factors (CSF) method is probably the most famous (Rockart, 1979). Other proposals have focused on the better involvement of users more during the system development process (Land, 1987), and how to overcome difficulties in communication (Oliver – Langford, 1987; Valusek – Fryback, 1987; Mittermeir – Hsia – Yeh, 1987). Mantey and Teory (1989) summarise behavioural methods, which can be used at the different stages of the development process. Benbasat and Taylor (1978) discuss system designs corresponding to different cognitive styles. The prototype method also aims to improve the efficiency of user-developer relations. According to this method, the user has an opportunity of see and check a demo version of the designed information system, which allows him to make changes in time. As development progresses, prototypes become get more advanced and approach the user’s 9 demands. Based on 15 years of research, Beynon, Avies, Tudhope and Mackay (1999) evaluate the prototype method as follows: Advantages: • better communication between user and developer; • better determination of information requirements; • greater commitment of the user to the project; • more user-friendly information systems; • simpler software, easier maintenance; • shorter development period. Problems: • new types of tools and skills are needed for the development; • costs are at least 5-10 per cent higher; • project management is more difficult; • extreme expectations may arise; • impracticable in the case of highly complex systems; • users’ enthusiasm may be decreasing; • the system design is conservative (often starts from the present situation). The different types of information systems shown on Figure 8 follow different development techniques. ERP systems are implemented according to standardised methods. These methods, which are defined by the system developer (in some cases with the help of a consultant), are in most cases a combination of the sequential (life cycle model) and the prototype techniques. Data warehouses and OLAP-based systems are typically introduced with the prototype technique: some system developers take an already functioning prototype to the first meeting with the potential client. Besides ready-made and half-finished systems it is worth noting that with the proliferation of spreadsheets and user-friendly development tools more and more systems 10 are developed by the users, themselves (end-user computing). However, such systems can only fulfil local and/or simple analytical requirements. 3.5. The Information System Organisation In the case of decision support systems emphasis is laid on the determination, adequate supply, and flexible changes of unique managerial information, rather than on standardised processing of data based on simple algorithms. ‘One-off production’ suited to managers’ unique personal characteristics and special needs plays a prominent role instead of ‘mass production’. It is important to note, however, that ‘mass production’ still important for it provides the necessary input of ‘one-off production’. According to this perspective, the information system organisation, either staff or line, is close to managers, most probably near the top of the hierarchy. Basic expectations from its members are that they know the general business models (e.g. those of the financial, controlling area), be aware of the critical success factors of the industry, and also be able to communicate effectively with users. Although assistance from external experts or expeditious colleagues taking part in the development of their own information systems can certainly prove to be very helpful, the prominent role of the IT organisation could not be questioned. 3.6. Strengths and Limitations It is indisputable that the fulfilment of managerial tasks, especially of planning and control, depends heavily on correct and suitable information processing. Therefore, thinking of information systems as managerial decision support devices is clearly justified. 11 However, the exclusiveness of the MIS approach cannot be upheld any longer. Experience shows that even the a priori assumptions (listed in sub-chapter 3.1.) have often failed in practice: • First of all, the effectiveness of organisations can be improved in several ways, not only by means of supporting managers’ planning and control decisions. Advance can be achieved directly and in shorter run as well, for example through the development of strategic applications (see Chapter 4). • Secondly, it can be doubted whether the provision of the requested information, in itself, is sufficient for managers to make good decisions. A careful assessment requires that we consider the experience of researchers of organisational behaviour (Feldman – March, 1987), which show that: − some information necessary for a decision and available somewhere in an organisation is only revealed to decision-makers when the decision is already explicitly made; − even if the requested pieces of information are known to decision-makers, a large part of them are ignored when making the decisions; − independently of how much information we presently possess, there is always a need for additional information; − complaints about not having enough information come parallel with neglecting available data and facts. The above patterns of organisational behaviour only make sense if we place ourselves in other perspectives. From power point of view, the collection and transfer of information always take place in political arena, within a context of conflicting interests. Decisions made can either favour or damage existing position. Under such circumstances the likelihood to delay, (intentionally) misinterpret or distort information increases. Certainly, decision-makers are aware that a part of the information they receive is manipulated. This is one of the reasons why they ignore a set of the information available for them (Feldman – March, 1987). 12 According to another approach good decisions are characterised by the use of much information. That is to say, information in itself has value, so persons or organisations are considered to be ‘better’ if they have more information. In this sense the consumption of large amounts of information increases the base of legitimacy and leads to favourable judgement about the organisations and decision-makers concerned. This example shows that information as well as the information system that provides it can often be a signal, a symbol (Feldman – March, 1987) beside (or instead of) its primary function of reducing uncertainty (see Chapter 7). • Finally, claiming that decision-making is assigned to managers exclusively has also proved false. More and more companies apply the empowerment conception and allow their employees to make important decisions (see Chapter 5 and 6 on process and knowledge management). The information systems pyramid models based on the previously challenged a priory assumption can invoke further critiques: • The availability and quality of managerial information depends on the correct measuring of elementary data and their proper insertion into the database, with the appropriate dimensions (Bodnár, 1997). Pyramid models turn attention away from these important elements by disregarding the level of the EDP/TPS systems. • Pyramid models accept and implicitly support the vertical and horizontal segmentation of information systems. They do not realise that copying the organisational structure is not a value in itself, and that the segmentation of information systems is largely a result of technical problems. With the emphasis on the pyramid structure they promote a conservative, rigid concept from organisational theory point of view as well. • Pyramid models suggest that only the top levels of the organisational hierarchy require ‘intelligent’ decision support (Sprague, 1980). However, such support (embodied in DSS/OLAP systems) can be helpful on all levels (let us remember the popular empowerment conception again). 13 • Pyramid models cannot handle systems which do not belong to a specific level of hierarchy, e.g. electronic mail systems, intranets, groupware applications, or personal productivity tools. • Strategic information systems, which increase competitiveness through the direct influencing of the business activity rather than good managerial decisions, cannot be inserted into pyramid models either. Some authors (for example O’Brien, 1997, p. 11.) try to fit these systems into the pyramid anyway, probably for the sake of completeness. They put them to the top, ‘strategic’ level, which shows a misunderstanding of the nature of strategic information systems. • Pyramid models forget about inter-organisational information technologies (remote terminals, EDI, Internet, etc.), which play a significant role in business. They consider incoming information to be important only for top managers, so they wish to create channels only through the EIS systems. This contradicts both to the actual demands and the actual functions of organisational information systems (Finnegan – Murphy – O’Riordan, 1999). 14 15