5. Structured Decisions and Decision Processes

The first phase is a review of how decisions of relevance are made in the organisation. Then one must look to understand what can be improved. During this rough analysis, the following questions should be posed.

• What is the decision basis? How does the collection of information take place? Is there anything of relevance documented by previous decision-making processes?
• How do the strategies relate to the organisation’s goals and how are the overall objectives affected?
• Which sub-decisions are involved in the decision? What types of decision should be made during the various decision-making processes? Only long-term strategic ones? Or quick ones? Are all relevant criteria covered by the different decision types?
• Which decisions are repetitive? Which components of the strategy can be used in future decisions? Can parts of the information be reused, and if so, which parts and when?
• Which methods are used today? Are there elements that make these methods structured? Or do most things just happen?
• Who makes strategic decisions? Is it primarily the CEO, management group, board, or IT manager? Or are there other decision-makers who make important decisions affecting the business as a whole and impacting the strategy?
• Is there a documented process for implementing the strategy? Which decision components should be employed? Which informal processes can be formalised?
• What are the experiences and outcomes of previous decisions? How do the organisational structure and delegations look? What are the responsibilities and reporting paths?
• Are there adequate information systems and other information management processes?
• Are there well-specified quality assurance and rules of procedures for different staff categories as well as for the board and management groups?
• Who should be in the groups that will be involved in the different phases of the project, and how should they be involved?

The final point is very important for the result of the decision-making process. Developing well-composed groups that are anchored in the business will have a major impact on the decision quality and implementation. It is also fundamental that multiple end users are involved. A well-composed group of active and interested members who feel involved and have representative views is integral to the analysis.

Finally, the conditions for structured processes and the potential for improvement are examined during the strategy work. Often, there is much to be done to address risk analyses, transparency, dependence on individuals, and how fact-based the decisions usually are.

• Is there enough potential for implementing and introducing a structured process at all, or is significant preparatory work required? What should be included in the analysis? What boundaries should be defined?
• How can we learn from different experiences in the organisation?
• Who is perceived as beneficial to the organisation?
• What are the qualitative and quantitative goals of the organisation?
• What are the risks and conditions of managing new strategies that appear during the process?
• Are all of the above sufficiently documented?

After the rough analysis, a more formalised decision-making process should be carried out. It is normally advisable to start with a few select decision problems that are particularly important to the organisation.

A decision-making process should constitute a complete process of data collection, investigation, analysis, and recommendation. It may look different, but a common one would look as follows:

• Identification of the decision problem (or problems), so one knows what to do
• Structuring the problem so that the decision components and their relationships are clearly visible
• Information collection resulting in a detailed information basis
• Modelling the problem around the information basis
• Evaluation of the model whereby existing information is aggregated and analysed
• If necessary, feedback on and repetition of previous steps
• Creation of a decision basis consisting of instructions and recommendations

It is surprising how often the identification of the problem itself is missing, even though it is central to reasonably tackling the decision problems. In this step, we must clarify the purpose of the decision, what it should include, and what it should omit. The problem can often be seen from several perspectives, and assigning values to different strategies thus depends on the perspectives from which they are considered. The perspectives are represented by various criteria. All of this should be documented in a project specification based on the rough analysis outlined above.

The structuring of an actual decision problem involves specifying the relevant criteria and possible strategies for obtaining a more precise problem formulation to support the collection of information. After this, one should have a rough summary of the criteria based on the specified objectives and a list of the strategies that are considered feasible along with descriptions of what they mean at different levels.

To make the discussion more concrete, we will consider an example of how an organisation might develop an efficient and secure physical IT infrastructure with high transmission capacity nationwide, granting people easy access to interactive public e-services.

After the process described above, the most important criteria are believed to be security, efficiency, availability, and transmission capacity (throughput). There are also three main strategies: i) to outsource the entire development and operation, ii) to develop the necessary services in-house as needed, and iii) to develop more qualified services in-house and outsource simpler services.

Financial aspects are generally central to both the investment itself and the continuation of the operation. Personnel satisfaction and development, as well as changes to work tasks, are further perspectives to consider. We will soon proceed with this example, but first we must explain how to model the information involved.

During this phase, precise criteria are developed and ranked, and the strategies are specified. The criteria must be prioritised. This is usually achieved through the assignment of weights indicating importance to each, but the criteria could also simply be ranked. Thereafter, the consequences of the different strategies are analysed and valued. Finally, scenarios are analysed and the probabilities of different strategy consequences are estimated.

There are several difficulties at play here. The first time one gathers information, it is often difficult to find out what is really required for the analysis, that is, to correctly model the relevant criteria, priorities, and possible strategies with consequences as well as utility values and weights.

Further complications usually appear when trying to estimate the probabilities of the different scenarios. Furthermore, there are often preferences that may conflict with each other, such as maximising the quality and at the same time obtaining a high financial return on investment. Such goal conflicts lead us to make some trade-offs between our goals.

There are thus often contradictions. For example, the strategy with the best economic forecast might entail issues with the work environment. Often, one strategy is the most suitable for the short-term profitability of the operation and another is most suitable for accessibility. In this case, which strategy should be chosen? One way of managing the choice would be to evaluate all strategies solely on the basis of the most important criterion, but this approach ignores the information that other perspectives provide. Taking all perspectives into account can yield somewhat absurd effects, such as one strategy being merely slightly better than another from an economic perspective, but obviously worse from a quality perspective.

In order to progress in the analysis, we must prioritise how different criteria relate to each other and express this so that it can be calculated. This can be achieved by assigning significance weights to the criteria that express their relative importance. The greater the weight, the more important the criterion. Then we can assess the strategies based on the criteria by aggregating everything in an evaluation. This approach, whereby criteria are weighted or ranked according to importance, is called a multi-criteria analysis. We also usually study the effects of any uncertainty in the background information in order to assess the reliability of the solutions.

Some of the prerequisites for successful decision-making are thus a real ability to describe the strategies and also a clear sense of how well they fulfil the different criteria. We also need to understand how important the different criteria are in relation to the operational goals. There are better and worse methods for doing so. We will begin by considering a kind of base case, in which criteria weights and strategy values are numbered precisely.

The range of values assigned to strategies can differ depending on the situation. A scale from 1 to 5, for example, where 1 indicates that a strategy has a very poor rating under that criterion and 5 indicates that it has a very high rating, is not uncommon. We then obtain weighted values for the strategies where the values under the criteria are weighted with the criteria weights. The weights must be greater than or equal to zero and the sum of all weights must add up to 100%.

Suppose that in our e-service example above we consider that safety is more important than efficiency, which in turn is more important than accessibility, which is more important than transmission capacity. In order to carry out the analysis, we must formulate this more concretely. Let us specify our criteria preferences on a scale from 0% to 100%, to reflect how important the different criteria are. This is of course impossible to do precisely and reasonably, but let us pretend for the sake of argument that it is actually possible. The importance of:

• Security is 40% (0.4)
• Efficiency is 30% (0.3)
• Availability is 20% (0.2)
• Transmission capacity is 10% (0.1)

The strategies under the criteria are valued in terms of utility (i.e., how “good” they are), represented as numbers. A common method of valuing strategies is, as in Table 5.1, for the lowest possible utility for each criterion to be 0 and the highest 5. Intermediate strategies allow for utility values between 0 and 5. Note that the full utility scale range [0, 5] need not be occupied.

Table 5.1. Development of an e-service structure – valuation of strategies.

The values of the strategies under the criteria are then weighted together. If plainly unreasonable strategies have already been sifted out, then a fairly simple and well-established method is to weigh together criteria and values using the following formula:

V(A₁) = w₁ · v₁₁ + w₂ · v₁₂ + w₃ · v₁₃.

Here, w_j is the importance of criterion j, and v_ij is the utility value for strategy A_i under criterion j. We then look at which strategy is the most suitable by calculating the weighted values for all strategies:

V(Outsource) = 0.4 · 1 + 0.3 · 3 + 0.2 · 4 + 0.1 · 5 = 2.6

V(In-house) = 0.4 · 4 + 0.3 · 3 + 0.2 · 5 + 0.1 · 4 = 3.9

V(Mixture) = 0.4 · 3 + 0.3 · 4 + 0.2 · 4 + 0.1 · 4 = 3.6

We select the option with the highest weighted mean value, which is the strategy of developing the necessary services in-house as requirements arise. See Figure 5.1 which also shows the contribution of each criterion to the total value of each strategy.

Note that it is in reality in most cases very difficult, if not impossible, to produce precise weights and values for the strategies. Instead of expressing utilities or weights as precise numbers, we can instead work with comparisons and intervals. This is especially important when handling qualitative criteria such as those we have discussed here. Even without the requirement for precise values, people often find it difficult to express their preferences and therefore use different methods. Sometimes the criteria are compared to a so-called reference criterion, such as cost, and the decision-maker works out which trade-offs he or she is willing to make on that basis. For example, a cost reduction of EUR 10,000 can be perceived as having as much worth as a certain reduction in quality. Needless to say, this requires that everything is measured on well-defined scales with well-defined units.

Once we have obtained a result, we may find that we need more information or that we want to expand the analysis for other reasons. For example, we often need to give a more detailed description of the problem by building so-called criteria hierarchies with additional sub-criteria. This is to further facilitate assessing the goal fulfilment of different strategies. In this way, we can often better capture the structure of the decision problem. Criteria hierarchies are applicable when there may be sub-criteria to some or all of the criteria.

We now assume that development efficiency has a weight of 30% (0.3) and operational efficiency a weight of 70% (0.7), and that the values for the respective strategies are as shown in the table below.

When we consider ourselves finished with the analyses, we compile them into a structured decision basis.

Table 5.2. The values of the strategies under the sub-criteria.

Once we have come this far, it is possible to deepen the analysis further. Multi-criteria analyses are in this context a good tool for gaining a clearer picture of the decision situation, but do not normally reflect the fact that there is often a great deal of uncertainty involved. When working with risk analyses, which try to assess the probability of various threat scenarios, we must perform more detailed analyses and also estimate the probabilities of the different scenarios, as well as assess how the different consequences of the strategies will affect the situation. We must also work out how to value these consequences in order to get a clearer picture of the characteristics of the possible strategies. The components of a more detailed decision formulation are as follows:

The result of the evaluation forms the basis for the decision, along with other documentation from the decision process and well-founded and well-motivated recommendations regarding development and implementation.

A large part of the total work time in organisations is spent gathering, processing, and compiling information in the light of a set of business goals. The purpose is usually to create a basis for making organisational decisions. Much is gained if the decisions—or at least the important ones—are made in a rational and not intentionally biased manner, and if the decision-maker takes all relevant available information and all reasonable possibilities into account. Unfortunately, this is not usually how it works. People often make decisions on the basis of unclear reasoning, and this chapter has thus provided an introduction to how professional decision-making and quality-assured decision-making processes in companies and authorities should work.