In this paper we have proposed an initial methodology for designing situational information systems. Using traditional methodologies, designers decompose the world into object correlates for implementation in databases where information about real world objects is held. A situational approach is, by contrast, likely to result in data about activities and situations being recorded so that action can be undertaken. The systems designed are likely to be radically different from those resulting from traditional methodologies. For example, it is highly unlikely that an IS designer trained in existing design methodologies would design a system with the simplicity and elegance of the Kanban system whereas an operator on the shop floor would probably see Kanbans as a logical system for controlling stock. This observation, which is in principle testable, dramatically highlights the gulf between information design theory and operations practice, which our methodology addresses. Furthermore, the use of Kanbans cannot be dismissed as merely a quaint or anachronistic manual system because it is now very widely used in the high-tech automotive manufacturing industry where it often replaces computerised systems based on the deliberative approach. Thus, this methodology has the potential to revolutionise information systems and, we expect, will lead to much more effective information systems in specific contexts.

We built the initial methodology by examining concepts from the situational systems literature and by building a tentative methodology based on these concepts. We then examined some existing (manual) situational systems found in the literature involving human actors to deepen our understanding of the characteristics of situational systems and thereby to strengthen the tentative methodology.

We have found, in examining human actors as part of evolved situational systems, that environmental structuring and the role of physical tokens in an actor’s environment are critical to designing situational systems. Representation of situations and activities in an agent’s environment helps situated reasoning and enables action with little deliberation. Physical tokens help agents to hand over situations and to solve problems. We also expect that, in computerised information systems, specific physical information and communications technology may be required to support physical tokens for human actors in these systems.

In future work, we intend using three action research cycles (Baskerville and Wood-Harper, 1996; Lau, 1999) to successively refine the methodology from here. In each cycle, an already implemented system in an organisation will be analysed and changed by applying the situational systems methodology. The methodology used in each cycle will be the output methodology of the previous cycle (or that emerging from the pilot cycle, if it is the first). The system selected in a specific cycle will be one that has been implemented using a traditional design and development methodology, involves routine work, and has been deemed to be ineffective.

The methodology being refined in this project is likely to add a much deeper understanding to disparate attempts at designing information systems for difficult contexts involving repetitive routine activity. Soft Systems Methodology, Human-computer Interfaces, and Ubiquitous Computing are all examples of other possible approaches but they lack unifying theory. For the first time, there is prospect of a methodology for building situational information systems based on firm theoretical foundations.