Process Mining: the other side of the coin — 5 attention points to keep in mind while investing for a Process Mining solution
We explained in a previous article what Process Mining is and what are its advantages.
Let us now focus on its weaknesses or at least the elements to consider before investing in this technology.
This article is not based on a particular product of one of the leading providers of Process Mining solutions, but rather on the main challenges most of you will face while moving forward with it.
We need to start on this obvious premise: If the wrong quality of data is coming in, no valuable result can be obtained… The quality of data to import in a Process Mining engine is determinant, and several issues must be faced:
- Data gathering: Because Process Mining is based on events stored in information systems, the gathering of those in a unified manner is one of the cornerstones of the technology. To accomplish such ambitions, at least one of the two challenges below must be faced:
Either, you have access to a big range of events from various systems (as JIRA, SAP, Oracle, Salesforce, etc.). In this case, their exploitation and interpretation aren’t always a piece of cake: you have duplicates to handle, different structures or irrelevant data.
Or, you lack of data to model your processes. Some parts of the processes aren’t always stored in IT systems under structured format (Oral exchanges, phone calls, unstructured mail exchanges, etc.). How can a process be modeled based on events if they aren’t identifiable by the system?
Technologies such as the “Task Mining” have been developed to counter this last issue. Concretely, this technology consists in installing a “spy agent” on your employee’s computer.
All actions from the process actors will then be collected, taking into account the systems they are interacting with. Even if this is quite promising, it also brings new challenges related to privacy in addition to other challenges, like: Data interpretation.
1. Data interpretation: To perform Process Mining, the data to integrate in your system must be classified in a defined way. To do so, the several people that are aware of how processes are applied in the organization (either normative or descriptive processes) have to be implicated to provide the data context. Without knowing the particularities of your sector activities and the big picture of all the processes applied behind, it is hard to ensure that the delivered outcome will effectively represent your process… This is quite a risk that must not be discarded too quickly.
2. Specific standard processes based on Major IT systems: As mentioned in the first section, the data gathering and interpretation are essential aspects and are often complex. This explains why most Process Mining engines are mainly based on specific standard processes (P2P, etc.) relying often on major IT systems like SAP, Oracle, Salesforce, ServiceNow, etc. In this context, events can easily be extracted and interpreted by most Process Mining engines. Nevertheless, the increased use of micro-services, the rising amount of inter-dependencies amongst various systems, the emergence of new technologies and the specificities of some sectors are elements that will complicate the use of this technology.
3. Not a cheap “Plug & Play” installation: This technology needs to be set up properly with support from experts and experienced Process Analyst(s), who know or can clearly manage your own company processes. The first outcomes will be delivered after a certain amount of time (at least several weeks) and probably expensive set-up costs (including licenses, potential tool experts) even tough there are free Process Mining engines, with limited capabilities.
4. A great tool… for Process Analysts: The added values of Process Mining engines is certain. Nevertheless, this technology should be considered as an additional tool in Process Analyst’ toolbox. As discussed previously, you must have the internal capability to perform a proper data gathering and interpretation. But you must also be capable to properly define how to solve the raised pain points and/or how to evaluate the suggested BPI areas. Putting such a technology in the wrong hands could lead to lesser performances and even counterproductive changes. Even if you can sometimes simulate the expected outcomes of some automated parts of your modeled processes, the implementation of this solution must also be further investigated. What will be the efforts required in terms of change management? How many stakeholders would be involved in the development of this solution? Are the targeted activities of the process making part of larger end-to-end process ?
In other words, the system will tell you which are the pain points to tackle (if the process has been properly modeled) and could potentially suggest you which solution should be implemented. Nonetheless, the review of these pain points and the best way to solve them will always require a more in-depth analysis from someone with BPM skills who knows the context of your company.
Conclusion
As explained, Process Mining technology has great advantages for your company. In order to handle this technology properly and retrieve the maximum of its benefits, then: You need to have people with serious process analysis skills and clear affinity with data management. They also need to be aware of the specificities of your organization. … “Don’t use a bazooka to kill a fly”
