An independent study reference written by Dr Phuc V. Nguyen. It is not official subject material — for assessment requirements always follow your subject outline and vUWS.
Decision support systems
A decision support system is built to help a person decide, not to decide for them. It belongs where problems are semi-structured, meaning part of the task can be computed and part needs judgement that resists being written down. Most such systems combine three things: a store of data, one or more models that turn that data into consequences, and an interface that lets a user ask what-if questions quickly. Interactivity is characteristic of these systems rather than part of their definition, but it is usually what decides whether one gets used: whether a manager can pose a question, see the implication, and change the question in the same sitting.
Why it matters
Compare a calculator with an autopilot. The calculator waits to be asked, answers fast, and leaves the choosing to you. The autopilot holds the controls. Decision support sits at the calculator end of that range, built for cases where no rule covers everything and the person has to supply context the system never sees.
Which task is the weakest candidate for a decision support system?
Worked examples
A hospital wants to reduce elective surgery cancellations. Should the analytics team build an automated scheduler or a support tool?
The scheduling arithmetic is computable: theatre time, staff rosters, expected procedure length. The rest is not. A surgeon may know a patient is unlikely to attend, a ward may be closing for maintenance, a registrar may be halfway through a training requirement. A tool that proposes a schedule, shows which constraints are tight, and lets a coordinator override a slot and see the flow-on effect immediately fits that shape. Full automation would either ignore the unwritten knowledge or be quietly rejected by the people who hold it.
A category manager asks for a report of last month sales by product and region. Six months later she is still asking for variations of the same report.
The repeated requests are the signal. A static report answers one question, and every follow-up costs an analyst a day. Turning it into a support tool means letting her change the period, the region grouping and the comparison basis herself, and adding a small model that projects the effect of a proposed range change. The value is not that the report became prettier. It is that the loop from question to answer shortens from a day to a minute, which changes how many questions get asked at all.
Common mistakes
- ✗A decision support system replaces the manager. It is designed to do the opposite. The system handles the computable part, and the judgement, the context and the accountability stay with the person.
- ✗A dashboard is a decision support system. A dashboard reports state. It becomes support in the fuller sense once a user can interrogate it, change assumptions and see the consequences, rather than read a fixed set of numbers someone else chose.
- ✗The better the model inside, the better the system. Systems that are accurate but slow, opaque or awkward go unused, and an unused system supports nothing. Response time and the ability to ask a follow-up question often matter more than the last few points of accuracy.
- ✗Support systems suit structured problems best. Structured problems can be handled by rules, which makes them strong candidates for automation, so an interactive support layer usually adds less there. Support pays off in the middle band, where part of the task is computable and part is not.
Revision bullets
- •Support rather than replace: judgement and accountability stay with the person
- •Aimed at semi-structured problems, between pure rules and pure judgement
- •Usual parts: a data store, a model layer and an interactive interface
- •Speed of the what-if loop often beats the last decimal of accuracy
- •A fixed report is not support until the user can change the question
- •Structured problems are often candidates for automation, unstructured ones stay with people
Quick check
Which task is the weakest candidate for a decision support system?
A well-built support tool is used twice, then abandoned in favour of the old spreadsheet. The most likely cause is
Connected topics
More in Decisions and Models
Sources
- Keen & Scott Morton (1978)Keen, P. G. W., & Scott Morton, M. S. Decision Support Systems: An Organizational Perspective. Addison-Wesley, 1978.Early statement that these systems extend a manager judgement on semi-structured problems rather than automate the decision away.
- Gorry & Scott Morton (1971)Gorry, G. A., & Scott Morton, M. S. "A Framework for Management Information Systems." Sloan Management Review, 13(1), 1971.The structured, semi-structured and unstructured classification that marks out where support systems belong.
- Power (2002)Power, D. J. Decision Support Systems: Concepts and Resources for Managers. Quorum Books, 2002.Groups support systems by what drives them: data, models, documents, communication or codified knowledge.