The term “unstructured” has always bothered me, because without structure you have randomness.  When knowledge workers get things done, it is not random in any way.  They accomplish things in a very structured way, it is just not possible to know ahead of time how it will be structured.

Last week at the BPM & Case Management Summit I presented my talk on how different technology should be brought to bear based how predictable the work being supported is.  There is work on the left of the spectrum that is very predictable, and on the right very unpredictable.

Examples of highly predictable work is that being done at an automobile factory or a fast food restaurant.  This work is predictable mainly because the environment is carefully controlled.  The factory is designed to supply the right things at the right time, and while there may be some (anticipated) variability in the mix of models being produced, one can clearly predict that each car will need four tires, mounted on four rims, attached to the wheel, etc.  A fast food restaurant takes an order, and fulfills it in a few minutes in a very repeatable way.

As you move to the right across the spectrum, we consider shorter predictability horizons.  Integration with other IT systems (the second pillar) means you have to be prepared on a monthly/yearly scale for systems to change.   Human processes (the third pillar) need to cope with people going on vacations, getting sick, learning new skills, and changing positions with a weekly/monthly predictability horizon.   The fourth pillar is production case management where the operations that one might do are well known, but when to do them is decided on a daily basis.  With adaptive case management (fifth pillar) you also have an hourly/daily predictability horizon, but the operations themselves can not always known in advance,and the knowledge worker plays a bigger role in planning the course of events.

Now compare the predictability horizon with the length of the process.  In the case of the fast food, I can predict a month in advance how a particular type of food will be prepared (after the order is received) and it only takes a couple minutes to do the preparation.  We call this predictable because the process is much shorter than the predictability horizon.  The other extreme might be patient care which can take months or years, while our ability to predict is quite a bit shorter than that.  New procedures, new treatments, new drugs are continually entering the market, while a given patient episode might last months or even years.  While treating the patient, decisions are made, and course of treatment can be predicted for certain durations, it is just that those durations are shorter than the overall process.  When this situation occurs, we call it unpredictable because we can not say when the process begins how the process will unfold.

Patient care is not random and it is not unstructured.  Unstructured implies that there is no thinking being done, and that there is no planning necessar and there is no control.  The truth is exactly the opposite; there is quite a bit of thinking and planning being done, there is quite a bit of control of what happens.  The work is not unstructured, it is simply structured while the work is going on.  The planning and the working happen at the same time, and not as discrete phases in the lifecycle of the process.

For this reason I propose the term “late-structured” to explain what knowledge workers do in case management.   They actively plan and structure the work, it is just that they don’t do it as a separate phase.  There are other implications of this:  since you can not separate the planning from the working, clearly both the planning and the working need to be done by the same person.  Knowledge workers must plan, to some extent, their own work.   Also, there is little point in creating elaborate models of the work, since further planning will change that, and it is likely that each instance of the process will be unique.

There is no loss of control.  Late structured processes can still be analyzed after the fact the same way that any process can, and so one can assess how efficient the work was done, as well as whether it complies to all the laws and customs.

When using the term “unstructured,” it is easy to get confused about nature of the work, thinking instead that things unfold randomly in an uncontrolled way.  If you think about it as late-structured work, where the length of the process is longer than the ability to predict what will happen, but prediction and planning still proceed, you gain a better understanding of what is really going on.