Process Mining techniques are not new. Professor Wil van der Aalst and his team have been leading the research on process mining for the last 15 years. More recently, companies such Celonis, Signavio and Minit have accelerated the adoption of Process Mining platforms by companies from different industries.
Process mining is essentially a set of techniques using mathematical algorithms to analyze business processes based on event logs. There are a lot of process mining algorithms, toolkits, or software, available that are generally based on what I call the “traditional” process mining methods of discovery and conformance checking.
A discovery technique takes an event log and produces a process model without using any a-priori information. An example would be to use process mining discovery on a SAP event log containing information on suppliers billing and payment processes.
In conformance checking, the algorithms looks at the underlying process model and compare it with the process logs built from actual process cases to see if they match and flags you when they don’t.
Then there is Performance mining or process mining enhancement, which is much less commonly applied. It too looks at the underlying process model and helps evaluate and predict performance, that is, how long an activity should take to complete, how long a process should take to complete, where there are holdups waiting for activities to completed before another activity can start, and so on.
I really believe that Performance mining or process mining enhancement will play a big role in the future of BPM as those algorithms and techniques can be applied to any business process model already automated in a BPM platform. If properly configured and trained, these algorithms can be applied to different business processes and can be used to make both predictions and recommendations.
At Bonitasoft we are using a process mining enhancement technique for the platform that evaluates the performance of an individual process case against the performance data produced by that case, plus all previous cases of the same process.
Process performance mining data is useful for predicting process blockages or inefficiencies, and when this information is provided to users, they can take action to prevent or avoid an issue before it happens. And when performance data shows issues, that can point to the need to make changes and even suggest where the process itself can be improved.
