What is process mining?
By Yassine Seghrouchni, founderUpdated
In short
Process mining is reconstructing how a business process really runs, based on the logs in your systems: who did what, when and in what order. It shows where lead time, rework and deviations occur. That sets it apart from a process description, which shows how the process was meant to run.
What it delivers
Process mining answers questions nobody can answer off the top of their head:
- How long does this process really take, from start to finish?
- How many cases follow the standard route, and how many deviate?
- Where is the rework: which steps are done more than once?
- Which alternative routes are so common that they’re effectively the norm?
That last question is the most valuable. A route called an “exception” that occurs in a large share of cases isn’t an exception. And it explains why automation based on the official process description so often fails.
Why it matters in practice
Process mining needs three things that SMEs don’t always have: systems that reliably record events, enough volume to reveal patterns, and a tool plus the expertise to do the analysis.
If part of a process happens in mailboxes and spreadsheets, there’s no usable log. Observing the work and a few conversations then get you the same insight faster. How to tackle such a process next is covered in automating business processes with AI.
When it does pay off
With high volumes in a single system, such as order processing in an ERP, tickets in a service desk or applications in a portal, and when you suspect the process runs differently from what everyone thinks. Then process mining shows what interviews don’t bring to light, because people rarely see their own deviations as deviations.
Frequently asked questions
Do you need process mining before implementing AI?
In SMEs, rarely. Process mining needs systems that log reliably and enough volume, and those aren’t always there. Often, half a day of observing the work and a few interviews give you the same insight faster and more cheaply.
How does it differ from a process description?
A process description shows how the process is meant to run; process mining shows how it actually runs. The gap between the two is usually exactly where the time goes: unexpected loops, skipped steps, files that go back and forth four times.
