The reason that most people hire me is that I help them make complexity manageable. If one cannot make a complex situation manageable, one cannot manage it. Ton de Leeuw‘s management model shown in the picture below helps us understand why. To control a real situation, you need some kind of a (mental) model of what the situation is and how you control it. If the model is too complex, the manager won’t have enough processing capacity to turn information about the situation into decisions on how to steer the situation in the right direction.
So one needs a model that is right for the situation. That is all about finding a balance. A model that is too simple will lead to bad decisions that don’t get the situation closer to the goal. The Beer Game
used for teaching systems dynamics illustrates what can happen. And we have already seen that a model that is too complex will not help the manager. The ideal is a model that is refined enough that it allows effective steering, but is just within the information processing capability of the manager. That may sound abstract, but in practice there is a whole toolbox to achieve this. Business schools and engineering schools teach many of these. Common examples include:
- Breaking the scope down into smaller parts and managing the interactions between them;
- Standardization (processes, procedures, parts, raw materials);
- Statistical process control and associated disciplines;
- Deciding what is good enough (Pareto);
- Finding rules of thumb (heuristics).
Where the right balance lies depends on the complexity of the situation. Large, complex organizations need more complex management models than smaller ones. It is also easy to lose the balance; this is probably a key reason for the failure of many large projects, and of some large companies. Companies scaling up will usually need to refine their model several times as they become larger. Startup ventures have a particular challenge, because the number of questions to resolve is large and the capacity to do so is usually very limited. I believe that is an important factor in the high failure rate.
So is there a silver bullet? Not really: every situation is different, but we can, of course, make use of similarities and analogies. The only general rule is that the model must enable whoever is using it to make effective decisions, being those that will bring the situation closer to the goal. If it does not, then we need to change either the model or one of the other factors.