Dilemma between automation and manual execution is always a challenge for correct management decisions. It is not easy to decide, if an investment to automate a process will bring sufficient benefits and returns, especially, on a long term.
However, decision here should not be voluntary. There exist simple and reliable criteria to clearly distinguish relevance of process automation in every practical case. The criteria is process volatility. Volatile processes are difficult and impractical to automate.
Volatility of the process is easy to measure. Record multiple execution paths of the process. Build an average execution path. Measure deviations of each execution from the average. Measure volatility as median of these deviations. This will give uniform and universal criteria for processes of any nature and scale.
As a rule, volatile processes appear in new and actively developing business areas. With time, these processes become a routine and acquire better formalization. When formalization grows and volatility drops below certain level, it is clear evidence that a process is mature and stable enough to automate.
There always exists a discrepancy between a model of business process, however well designed and accurate, and real execution of this process in a business environment. The reason for this gap is an unforeseen depth and hidden details inherent to any real process. Real business model of organization is ultimately unlimited in its depth. Going from highest management levels, it descends to individual departments, client relations, production units, technical code of equipment and controllers etc. In vast majority of cases, it is impossible and senseless to build a complete model covering all and every fine detail of the business. Omitted lower layers of the model create (pseudo) random fluctuations during execution of the model. Real execution paths of a process never follow its model exactly. However, in case of the correct model, we can expect to see that an ensemble of execution paths statistically converges to the model as to its average path over a significant set of observation...
Comments
Post a Comment