Shifting burden to the intervenor
(in the context of managing outsource contracts)
This interpretation of the problem and its symptoms helped the group to produce more detailed causal loop diagrams of the present interaction between their organisations. They also generated a ‘wish list’ for an improved pattern of interaction, identifying the criteria to be used to assess whether a sustainable solution had been identified. Finally, they set about making suggestions for changes that would satisfy these criteria. At the end of this one-day meeting the group had resolved to move towards a greater participative role for contractors:
- using contractors as consultants at the project planning stage,
- joining together for benchmarking and to capture learning from experience that would be fed into updated training programmes, and
- collaborating to set ambitious but achievable performance targets and then sharing the
- financial benefits of improved project management.
The group’s suggestions for change, along with all their diagrams, were used to inform the development of some simple system dynamic models during the following day. When the group reconvened on day 3 they revisited their earlier thinking – testing their assumptions against the formulation and operation of the models provided. A number of additional insights emerged:
- The trend to ever more challenging projects would always stretch contractor teams and it would be difficult to counter this even with improved training. New suggestions were:
- collaborative research might help to identify future needs.
- reduced contractor turnover would help to build more experienced teams. Low turnover should become one of the criteria for selection of a service provider.
- To some extent the incidence of technical problems which could escalate to delay a project was random and outside contractor control. Improved monitoring procedures supported by statistical analysis would enable both contractor and client organisation management to better distinguish between random (intrinsic) variation and adverse performance trends.
These additional insights and change suggestions were carried forward from the meeting and were taken up by a wider change programme in the organisation. Feedback from participants was positive:
- they had found the workshop enjoyable and thought-provoking;
- they were leaving with an enhanced appreciation of how problems were arising and they would seek to spread this understanding;
- the process had been very powerful and should be used more widely across the business; and
- they were surprised by how much had been achieved in a short time.
Leveraging the management science contribution
The use of independent consultants was helpful in getting the necessary people into a meeting. Having got the right people together, facilitation, and particularly the use of a pragmatic problem structuring approach (hexagon mapping), helped to establish a constructive tone for the meeting. With each participant making a contribution from their own perspective, the group as a whole then had the information needed for a better understanding of the problem.
Influence diagramming provided a useful mechanism for catalysing the group’s appreciation of the dynamics of the problems that were arising. It also prepared the way for the introduction of relevant system dynamics archetypes.
The ‘shifting the burden to the intervenor’ archetype was offered to the group Blue Peter style (here’s one I made earlier) with the disingenuous suggestion ‘we wonder if this is relevant’? Clearly, the preliminary meetings with prospective participants had helped us to anticipate some of the archetypes that might be useful to the group. It seemed very helpful to them to have this simple dynamic structure laid bare.
However, the archetype structure was not imposed on their subsequent causal loop diagramming (though they referred to it) and neither were they obliged to include all the factors highlighted in their earlier discussion. In fact, with the changed emphasis upon the dynamics of the process of interaction between client and contractors, some new insights emerged at this stage. These included recognising shortfalls in the information the client organisation collected about contractor performance, and the extent to which contractors had a greater understanding of the project risks than did staff in the client organisation. In spite of now being on new territory, the group’s causal loop diagramming went quickly and easily. They were beginning, quite naturally, to formulate ideas for the more effective management of technical problems that might delay projects. The momentum generated carried them through the review of what they wanted from an improved contract and project management process and on to the diagramming of what this might look like.
Our process design for this first day borrowed a number of elements from David Kreutzer’s documented approach (FASTBreak) for accelerated entry to systems thinking in a group. Things went reasonably well and the group’s diagrams provided a good starting point for model building: in particular, they identified which factors would be central to a useful representation of the contract management process. Even with a streamlined and tested methodology for getting this far, it was something of an experiment for us to set up an expectation of a model to be produced on the day following this one-day meeting. We believed that a model could add value to the group’s thinking, but we lacked the confidence to build the model in collaboration with them. The ‘time out’ solution was proposed as a less stressful route. Even then, as ‘the modeller’ in our team, I felt suitably challenged by the limited preparation that it was possible to do beforehand and the very firm deadline that existed!
For obvious reasons therefore, the two system dynamic models (using iThink) that were produced were kept simple. They shared the same basic structure and were tailored to represent the current and redesigned project/contract management processes. The models had four stocks the levels of which varied as indicated below.
||INCREASED by a flow linked to
||DECREASED by a flow linked to
|1. Potential incidents leading to project delay
||The probability was maintained at a constant level throughout (generated by a Poisson distribution)
||Level of contractor's ability in relation to technical problems
|2. Contractor's ability in relation to technical problems
||Pace of technology change
Turnover amongst crews
|3. Satisfaction with project engineering performance:
i. Current system
ii. Redesigned system
|Low costs through delays
|High costs through delays
Benchmarking activities & assessment of contractor ability
|4. Accumulated technical know-how & learning in the total system
||Investment in learning
||Pace of technology change
A number of the qualitative factors in the models were represented by simple indices and some crude calibration ensured that the derived costs associated with project delays were in the right ball-park.
In introducing the models to the group, we first talked them through the structure using a projected image of the computer-based stocks and flows diagram. Being a simple model with uncomplicated connecting logic, it was easy for the group to engage with and they readily accepted that the logic reflected their own thinking as articulated at the first meeting. Second, the behaviour of the model was explored - initially by trying it out in an environment where the technical challenge and risk of a project delay was constant, and then in an environment with step changes in the risk. The model of the redesigned management process was then introduced and run under a range of conditions. As each computer run was carried out, the group watched key parameters being plotted over time, reviewing the results and comparing them with their expectations.
The results stimulated some discussion and some fresh insights: for example, the group noted that the current system was slow to respond to change and left the client organisation vulnerable in an environment where the pace of change is increasing. Moreover, the improved contract management process as formulated was not always as effective as the group had expected: it reduced, but did not contain, the increased costs of delays under more challenging technology. Thus, exploring the behaviour of the models led to the group questioning their original thinking and going further. This was in line with what Jay Forrester would have predicted for a group exploring their understanding of a dynamic system through a model of their thinking: he suggested that many organisational policy failures have occurred because people’s expectations about the behaviour of a dynamic system do not match what is known about that system. In this case the group formulated new ideas and suggestions that they wanted to explore through a modified model. Some small changes were carried out and tested successfully on the spot; some other changes were beyond the scope of what could be tackled in the time available.
The group’s feedback has been given above. The session reinforced my belief that even very simple system dynamic models can add value to thinking about systems. Thus I’d want to encourage others to take the modelling step even when time is short and only the simplest of models is possible. I suggest that modelling does not have to be either time-consuming or expensive. In fact, making a distinction articulated by Arie de Geus in his foreword to ‘Modeling for Learning Organisations’, a model designed to support learning rather than one intended to have predictive capability might be more effective for being simple and transparent in its logic. For pragmatists, it is reassuring to recognise that there is no simple linear relationship between ‘development time, model calibration and complexity’ on the one hand and ‘usefulness in application’ on the other!
First published to members of the Operational Research Society in OR Insight January - March 1999