Process mining aims at gauging the quality of business processes and improving them by analyzing
recorded events as typically found in
log files,
database tables, mail/communication archives, transaction logs etc.
Importance of recorded events
Because of the importance of the
(data) quality of the recorded events, it is crucial to ensure that the analyzed event logs are complete, factually correct and meaningful.
These requirements turn out to be somewhat problematic in many organizations because the log files are often viewed as (only) an aid to developers for debugging purposes.
In addtion to the stated requirements for recorded log events, it is also necessary that the recorded events can be assigned to well defined process steps.
Challenges
In addition to the
data preparation challenges, we note that log event data typically records facts about a
processed entity rather than the
process itself. Such events need to be mapped into process steps.
When analyzing log data, we need to take into account explainable data variations that are not systematic within the analyzed process but rather in its environment. Among such explainable factors are: staff shortage or new regulatory constraints that lead to longer processing time.
Not only can a changing environment affect activity duration, it's possible that the process itself changes while being analyzed (often referred to as concept drift). If such changes go untedected, the results of a process mining study are very possibly misleading.
Then, there are also an unknown number of unknowns that very possibly would reveal quite a siginificant amount of process information if we could tap it. For example, if it were possible to find out what communication took place ourselves between an organization and its customers and suppliers, and what semantically was exchanged between them, we might find valuable pain points in the process.
Unfortunatly, we often don't know if and where such information exists. Even if we knew it, it might not be possible to interpret the information in a meaningful way.
Non-techincal challenges
We feel we should point out that there are also human related rather than technical challenges.
Process mining creates transparency that some people are not easy with and is therefore not always welcomed. It's now possible to identify employees that are underperforming or have a hidden agenda.
Sometimes, its would be too easy to assume that mistake lies with the identified person, however. In many cases, the organization put in place some incentives that fostered a behaviour among its employees that is detrimental to the goals of the organization. Thus, the results of a process mining project should be used to correct the incentivers rather than to try to correct the employee.
Care must be taken to ensure from the beginning of a data process project to ensure that its findings are not used as an excuse for finger pointing. It should be clear that the essential goal of process mining is to improve