Analytical models

Analytical model s are very useful where the complexity is manageable, and there is no policy resistance or interacting feedback loops. They are effective when objective s are clear, the options are well defined and the critical uncertainties are measurable. They are easy to develop when there is a fair amount of clarity on a problem or situation, the cause and effect relationships are clear and persistent, and patterns are recognizable patterns (Figures 3.6, 4.8, 4.9, 4.13, 8.2, 9.9). They also need enough historical information for assumptions on certain variables, such as costs, processing times and the load factors of resources.

Good examples of the use of analytical models are Service Desk and call centre staffing, which can be visualized as a system of queues. It is possible to gather data on the rate of arrival of requests (or incident s), how long it takes to process them on average, and how many requests are waiting to be handled. This level of knowledge is sufficient to build simple analytical models. Figure 8.6 shows an example for a single-stage, single-agent queue at a Service Desk, with certain assumptions about the arrival pattern of requests and the processing time.

Figure 8.6 Example of simple analytical model for the Service Desk

Service Desk modelling can become quite complex with the addition of numbers of service channels, multi-stage processes, dependencies and delays. However, it is useful to start with basic models and progressively elaborate them to reflect closely the reality of a problem or situation.

The following are commonly used sets of tools useful for decision making in Service Strategy:

• Decision trees, payoff matrices, analytic hierarchy process, etc.
• Linear programming (Figure 8.7) and integer programming, goal programming, etc.
• Queuing and network flow models (Figure 8.8)
• Clustering, forecasting, time-series analysis, etc.
• Analysis of variance, design of experiments, etc.

These methods can be applied to solve a variety of problems such as:

• Allocation of resources between services and contract s
• Analysis of demand patterns and segmentation of users
• Compression, correlation and filtering (Table 8.2)
• Scheduling of jobs, tasks and staff
• Location and layout of facilities and infrastructure elements
• Capital budgeting, pricing and purchase decisions
• Portfolio optimization
• Contingency planning and redundancy (coverage problems).

Figure 8.7 Simple LP model

Figure 8.8 Simple network model

There is depth and diversity in analytical model s, some of which have been in use for decades and have been instrumental to the maturity of disciplines such as operations management, project management and financial analysis. Service sectors such as telecommunications, transportation, logistics and financial services have achieved high levels of performance from the application of systems and industrial engineering concepts, methodologies and quality control processes to service function s and processes.41

There is a range of automation tools available for analytical modelling. The simplest tool available is a computer spreadsheet such as Microsoft Excel with its built-in solver function. Model s with a fair amount of sophistication can be built using spreadsheets. More sophisticated models can be constructed using tools, special purpose optimization programming languages (OPL) and optimization engines. Several commercial solutions for automation in service management include functions and modules for analytical modelling and visualization.

Service Strategy and other functions and processes in the Service Lifecycle can benefit similarly from such knowledge to improve performance in the presence of technical, financial and time constraints. Six Sigma™, PMBOK ® and PRINCE2 ® offer well-tested sets of methods based on analytical models. These should be evaluated and adopted within the context of Service Strategy and service management.

Дата добавления: 2015-11-14