Asking the Right Questions

Some recent research in big data and statistical analysis indicates that the biggest problem faced by database analysts mistake the type of question being considered.

Without going into more detail about the problems of analyzing large data sets, the same problem often occurs in the process area. In process, the problem occurs because the analyst assumes that a process problem must involve one of few possibilities – based on some model that the analyst has in mind. For example, many analysts with a superficial knowledge of Lean, assume that process problems derive from flow problems, which in turn, derive from undertaking activities that don’t need to be done (Waste!). Such an analyst would begin his or her examination of a process problem by creating a diagram of the workflow sequence and then looking for activities that didn’t need to be performed.

In a similar, someone focused on business rules might begin by looking at decisions that need to be made and then examining the business rules used to determine if they represent a logical and complete way of reaching a correct conclusion. As I pointed out in a recent article on problems in Afghanistan, however, the rules in place would have led to the right conclusion – if they had been implemented. The real problem was that the people ignored the rules, because they thought the solution they had come up with, quickly, was already correct. The problem wasn’t rules, but the management will required to insist that the rules always be followed.

In a similar way, in Afghanistan, the US Army decided that additional training was the solution. It almost certainly isn’t.   The men undoubtedly know that they should follow the rules, and they know the rules – you can’t begin an engagement until you check the coordinates to assure that the target isn’t on the list of prohibited targets. It wasn’t a lack of knowledge that resulted in the failure – it was a failure to do what they knew they should do. Someone should have been there to insist that they check the coordinates. And that isn’t a knowledge problem, as such, it’s a motivation problem. More specifically, it’s a management problem.

Process problems involve getting work done. Work fails to get done correctly for a variety of different reasons and those possible causes then to interact with each other, especially in emergencies or when speed is required. A good process analyst needs a very comprehensive model of what might be wrong, and then he or she needs an approach that starts by looking at the problem, broadly, considering all of the various causes that might be involved in producing the performance gap that is associated with the specific problem.

The analyst needs to consider the question he or she is trying to address very carefully. He or she needs to consider alternatives and combinations of causes that may result in the noticed performance gap. Then, as often as not, some experimentation is required to assure that they cause is correctly identified. The rush to apply a simplistic model to a complex situation typically leads to an inadequate solution.



Leave a Reply

Your email address will not be published. Required fields are marked *