Process modelling is an act of exploration, and as such it not only helps in understanding the binding of process and organisation structure, but also helps to discover aspects of the organisation that are not obvious to the naive surface view of a process. Process modelling increases the organisational pot of self-knowledge. Process modelling helps us understand our internal business terrain from the inside, and therefore also the external touch points. It helps us identify business units that are tacitly a part of the process, but not recognisably so from the surface view of a process.
Using modelling as exploration and discovery is more than documenting the process, it also helps to identify points of potential failures and also opportunities for improvement in the process. Modelling allows one to abstract the essential underlying aspects or characteristics of process realities. It allows us to formalise the description, and thus more easy to share. The common language enables discussing of the finer details of the reality amongst the many and varied stakeholders. Modelling is performed with an ulterior motive, either as a tool for understanding the current actions or with an eye to automation and / or optimisation.
For all its benefits process modelling (in its true sense) has not received universal acceptance. Modelling is seen as unnecessary, after all processes are something you just do. Business just know what they want, and do just the right set of activities to that end, and to advocate anything else is just an unnecessary diversion of money and time. This state of affairs is further compounded by modelling tools focusing on the drawing part of the process. The drive to democratise the modelling process has led to a situation where the whole modelling has been dumbed down to the production of box-and-arrow diagrams. Tools in attempting to hide the complexities of modelling (be it simple or intricate and complicated processes), see their main purpose as protecting the modeler from all the messy reality, not unlike nannies.
There is nothing entirely now here, some of us will remember the advent and demise of 3GL and 4GL, and the case tools that generated cobol code at the press of a button. Of course, when the code was deployed to production, you could almost hear the wailing and weeping of the support staff as they tried to debug the resultant failures. We have fallen in love with the idea of wanting to do things without thinking too hard. Tools are meant encourage reflection, like an old friend listen and talk (not throw error messages), and in doing so they should help the user think. Perhaps we are just not yet in a position to build such tools.
Modelling is a necessary cost of understanding the process reality, be it for implementation, automation or simply documenting the process. Being a manual process does not negate it from modelling, every organisation thrives when it recognises its processes and is able to exploit the opportunities offered by optimising them at the right points. Every organisations ultimate aim is to become process driven, this cannot be achieved without modelling and all the messiness that entails, and it is not a place for the timorous. One can continue along the path of least cost, in the belief that a process is for the doing only, and all other activity is waste, but you must also then accept that you have chosen to not be a process driven organisation, but just an organisation with some processes.