Table 4 describes four engagements, in which an observed gap in practice drove theory development. Two engagements describe situations characterised by an absence of practice predicted by current theory. The other two describe situations where the observed practice was inconsistent with current theory.

Bannerman (2004) provides a capability-based explanation of IS project management performance outcomes, as an alternative to the traditional factor and process view. It presents a theory of performance as the contested outcome of drivers for success (learning) and drivers for failure (liability of newness).

Vlasic and Yetton (2004) provide a time-based explanation of how the variance of tasks on a project generates a cumulative variance in project performance. Drawing on the Total Quality Management literature, they present a theoretical framework to explain poor performance driven by the relationship between task inter-dependence and task variance.

Thorogood and Yetton (2005) provide an explanation of how the currently dominant IT investment model, Net Present Value (NPV), drives the bundling of project delivery. The authors propose an alternative Real Options-based model to unbundle IT investment decisions, with the IT infrastructure investment as the premium paid by an organisation to execute a portfolio of business project options. The business units then assess each optional business project over time, resulting in decisions to execute, delay or discard.

Real Options provides the IT investment decision framework but not ‘how’ to unbundle projects. Reynolds (2006) addresses IS project complexity and uncertainty, and argues that modularity can unbundle projects to reduce the technical and organisational complexity of IS-based business transformations.

Table 4 presents multiple theories, each of which addresses a gap in practice with new theory, drawing from different reference disciplines. This range of theories has been used to provide insight into problems in IS project management performance and to develop new theory that can be applied to make sense of, predict or prescribe practice in IS project management.

If only a single theory were required to fill the IS project management gap, the contention is that it would be easy to develop. Academics and practitioners together would have rapidly applied the theory to solve the identified problem. Instead, this research shows that the IS field requires multiple theories to support the management of projects, rather than a single theory of project management.

The ability to draw on multi-disciplinary thinking as described above has three major benefits. First, it enables easy access to alternative theoretical frameworks. Second, it provides access to a wide-range of research methods. Third, it supports deep immersion in the problem, generating strong engagement with practitioners.

The diverse theoretical backgrounds of the researchers supported the search for alternative theoretical frameworks and their initial evaluations. For example, from production engineering, the project critical path was treated as analogous to one run down a production line. The findings from Total Quality Management concerning variance-driven scheduling performance were then evaluated and integrated into the program. Similarly, Real Options Pricing was imported from investment theory to restructure the IS investment decision, with strong implications for both governance and the structure of the project and with both impacting directly on project performance. Looking in different places and through different lenses identified novel and powerful success factors.

A wide range of research methods can be applied. The selection of each is dependent on the research context and has included predominantly qualitative methods such as grounded theory, action research and interpretive case studies. It has also encouraged the research team, in other areas, to draw on quantitative methods such as structural equation modelling to allow simultaneous fitting of the data to the model and of the model to the problem.

Deep immersion in practice, with a multi-disciplinary team, supported a rich dialogue with practitioners. The managers involved in the projects evaluated all insights and this provided an early test against practice. Managers would know whether a proposal had already been tried and failed elsewhere in their industry. It also provided a guard against developing unnecessarily complex explanations, responding to Einstein’s call to keep it simple, or as simple as possible. All this illustrates how, within this research approach, there is a natural tension between the need to develop richer theory while, at the same time, maintaining simplicity to explain and guide practice.

Following the exploratory practice-driven approach described in this paper requires researchers to address three major challenges:

The first major challenge requires that researchers remain both problem-focused and theory-focused, even when deeply immersed in practice. Without this discipline, it is easy to become solution-bound. The danger is that the practical problem is solved but the researchers do not generate new theory.

The second major challenge is to improve performance in practice and not just to reflect what is already known. The danger is that the researchers may explain only what is already known in practice. Lee (1999) states that ‘with few exceptions, none of much significance, the scientists who turned to [practical needs] for their problems succeeded merely in validating and explaining, not improving, techniques developed earlier and without the aid of science’. This is almost certainly true for mature disciplines and practices. However, in immature areas with poor performance, such as IS project management, this is less of an issue. In addition, the approach of applying multi-disciplinary thinking allows new skills to be applied to practical problems.

The third major challenge is to ensure that the application of insights and models from other fields brings about new theory. To make a theoretical contribution, it is not sufficient to apply a theory from one field to a new context and to show that it works as expected. Whetten (1989) explains that the ‘common element in advancing theory development by applying it in new settings is the need for a theoretical feedback loop. Theorists need to learn something about the theory itself as a result of working with it under different conditions. That is, new applications should improve the tool, not merely reaffirm its utility’.

The application of the approach in this paper addresses this by providing deep immersion to evaluate both data and theory. It allows the simultaneous fitting of data to the theory and fitting of the model to the data. In this way, theory is adjusted to reflect the empirical data and, at the same time, it is tested against that data.

Finally, the approach presented above is oriented around the developing of new theory using insights and existing theory from other fields. This, in itself, does not address calls for new theory in the ‘core of IS’. Some, including Weber (2003), would argue that the IS discipline relies too much on theories borrowed or adapted from other disciplines. Instead the unique IS theory now becomes the integration of these theories, perhaps to the extent that others will want to borrow it.