What to do About Non-Targeted Discovery in Process Mining
Process mining is like therapy. Go in to talk about insomnia, leave with a whole closet of childhood trauma laid out bare. What one finds may not be what one’s looking for. The same goes for process mining. By analyzing traces of data left behind in IT systems, process mining objectively reveals the truth. Even if only 15% of the truth is what one’s after. The other 85% of explorative results may be interesting, shocking, and also valuable, but non-targeted process mining discovery can be distracting and a threat to project success.
Here’s how to better understand process discovery and how to approach non targeted findings with the 30/70 rule.
What is process discovery?
Process discovery lives within the discipline of process mining and is mainly concerned with extracting process knowledge from IT systems and analyzing event logs. The component which makes process discovery different from other areas of process mining, such as conformance checking, is lack of an existing process model. Process discovery is doing exactly what it claims to do — discover, not compare.
Because process discovery is based on a fixed methodology, the accrued information is 100% accurate and true. Unlike manual or second-hand process mapping, information isn’t selective. Information is complete. Process discovery reveals the entire truth of operations, childhood trauma and all. The very nature of “discovery” is to observe or find something unknown, making non-targeted process discovery unavoidable. However, the innocent seed of “knowledge is power” may be the root that derails an otherwise hopeful process mining project.
So what is one to do with all that truth?
Apply the 30/70 rule
Niyi Ogunbiyi of Deutsche Bank (UK) has a solution. Apply the 30/70 rule. Process discovery must strike a balance between explorative analysis and finding answers to defined questions. Ogunbiyi suggests a split of spending 30% of one’s time on non-targeted discovery — following data where data leads, and 70% on stated project goals — targeted exploration to answer questions being asked. This shared focus on non targeted and targeted process discovery allows data scientists to stay focused on delivering what’s anticipated while not ignoring potentially useful, but also potentially irrelevant, data.
Process mining practitioners and data scientists view data more holistically than most business or sales managers. Department heads are susceptible to taking on a pigeon hole view of business intelligence. Razor sharp focus on their KPIs, their goals and their sphere of influence, while taking only a passing glance at other department’s data. However, a good process miner knows that the knee bone's connected to the hip bone, so to speak, and a holistic view of data offers more than siloed data centers. The 30/70 rule allows the entire data picture to emerge while avoiding a rabbit hole that may bog down an otherwise successful project.
Targeted projects result in wins
Process mining isn’t only about powerful technology, it’s about communication. Emerging technology capable of bringing game-changing results will be met with some skepticism; change is intimidating. That makes communicating and achieving goals more important than ever. Don’t let non-targeted process discovery tempt you away from delivering what’s been promised. Targeted process mining projects which are both high in visibility, yet small enough in scope, are ones best situated to result in wins — especially as pilot projects.
Involve process owners, managers, and stakeholders in defining questions before commencement of a process mining project. Then, as a process mining practitioner and believer in the value-add capabilities of process mining, answer those questions.
How to unsee the seen (ie: keep 70% focused)
Now that process mining has discovered a rich, interactive process map chock-full of nuances, statistics, variants, exceptions, and bottlenecks, what next? How can one move forward with a project when so many risks and opportunities were accidentally found through process discovery?
Success is measured by goal achievement. Achieving defined goals for the approved process mining project will position the organization for further process mining success. Stay focused on delivering the promised value and use this success to accelerate the next project. With a foot firmly in the door and a solid pitch, that 30% spent on non-targeted process discovery can blossom into the next great project.
Process discovery has the potential to create more questions than answers. Remember to stay focused and not be the organization’s cat which gets killed by curiosity. Be the explorer who sets off in the unknown sea of data to return with valuable solutions.
Michal Rosik Chief Product Officer & Product Visionary at Minit
05. 02. 2019