Feature
Process mining may not be part of everyone’s day-to-day lexicon (yet), but it is the billion-dollar operational shift at the forefront of strategic planning right now — and for a good reason.
Simply put, process mining is about optimizing processes by analyzing event data. And with the help of AI, it has become a critical component of digital transformation initiatives, allowing companies to scale faster and more efficiently.
A ‘Limitless’ Technology
Although process mining may feel new, it’s not. The concept hit academia in 2007 and became a commercial product in 2011.
It’s only in recent years, however, that North American companies have started to build budgets around the technology — timing that Yusef Khan, associate partner at IBM Consulting Global Process Mining CoE, said is well-aligned with the adoption and pressure to drive value with AI in 2023 and beyond.
The spike in interest may also be, according to Alan Pelz-Sharpe, founder and principal analyst at Deep Analysis, based on the fact that the technology “has become much easier to use, both to mine processes and to understand where improvements can be made.”
But while typical process mining projects in organizations today include high-value options like back-office finance, procurement and order management, some say the potential — i.e., the ways companies could be using the technology — is in fact limitless.
“A-ha moments abound when you show a client their actual process for the first time,” said Joe Mislinski, principal consultant for Doculabs.
Still, he said, those “outside the box opportunities” are not always easy to find and implement, and may involve different data sets and use cases that are truly amenable to the ‘mining’ metaphor. “You can really find the vein of precious metal in that rock,” he said.
Related Article: How Generative AI Will Level Up Business Process Management
Process Mining vs. Process Mapping
Khan frames the conversation around understanding the difference between process mapping and process mining.
Mapping captures known insights and is subjective, “based on the loudest person in the room,” he said. This approach is limited in scope and quickly becomes stale.
Mining is different: “Mining happens in near real-time, and it uses event and log data from the source systems to document and visualize how a process is executed.” It’s data-driven and objective, Khan said. And it’s freeing analysts from constraints imposed by frameworks, like CRM or ERP systems.
“Process mining tells you definitively how the system-level process is undertaken, and chances are it’s not as expected,” added Pelz-Sharpe.
Process mining also folds in unstructured data to reveal novel, streamlined and often entirely new possibilities — or, as Mislinski puts it, it’s teaching clients how to fish. It’s also teaching them when to throw some of those fish back.
Related Article: What Intelligent Process Automation Can Contribute to the Hybrid Workplace
Revealing Potholes Before They're Reached
Thanks to process mining, a number of major business transformation projects have been stopped in their tracks, said Pelz-Sharpe, potentially saving millions of dollars for companies.
“Projects in education and healthcare, particularly in Europe, where process mining is much more widely used, have been halted as the scale and complexity of the mess are revealed,” he said. “Resources are readjusted and targeted to make positive and meaningful business changes, rather than simply adding more technology.”
Mislinski has seen inefficiencies solved when processes shift from the old-fashioned and nonstandard apprenticeship methods to process mining.
“Behind the scenes, what's happening is a Wizard of Oz set of hand waving, chicken wire and duct tape — 20th-century infrastructure holding it all together. Process mining can reveal that,” he said.
And there are boundless more use cases to explore, as machine learning in processing mining evolves, Khan said. “It has taken us from regression analysis to embedding multiple Large Language Models (LLMs) that take full advantage of traditional and generative AI.”
For instance, enterprise automation may be on the horizon, with LLMs analyzing and extracting insight from extremely large text data sets in conjunction with process mining, performing more sophisticated pattern-matching and eventually eliminating costly, error-ridden manual processing.
“But probably the most novel and exciting approach is the move away from ‘one and done’ mining projects, and instead using the technology for long-term, continuous monitoring and improvement,” said Pelz-Sharpe.
Conference
Sep
16
HR Tech Conference Las Vegas 2025
Register
Conference
Oct
27
Gartner HR Symposium/Xpo Orlando 2025
Register
One caveat: This requires executive sponsorship, which can be hard to come by. That’s because although the benefits can easily be documented and validated, seeing the potential beyond the current project is challenging, regardless of the outcome. The logic, according to Mislinski, is either:
- We solved that problem, and we don’t need the tool any longer.
- We don’t like the price tag (the juice isn’t worth the squeeze).
- We didn’t get any value out of the tool.
“A key to propagation is someone in the Center of Excellence who can sell the idea of process mining to the business leaders in the BUs of the org,” he said. And it starts with a strong proof point — a pilot program to support this value.
Related Article: Why the Process Mining Market Is Heating Up
How to Get Started With Process Mining
Companies looking to get started with a pilot program may want to start with a high-volume and high-value process supported by one or two source systems, advised Khan.
“This allows the client to quickly pressure-test known hypotheses about where processes break down and provide a quantified business case for our clients ahead of large transformation or automation programs.”
And understand the data you’re mining first, shared Mislinski.
“Ideally, we’d like to do a detailed Use Case discovery session to understand the business problem we’re trying to solve and get a detailed brief of the business process from the operators themselves. Then, we do a Data Discovery session, where we get a demo of the systems involved and the source data.”
In general, the process follows these steps:
- Initiate Phase: Define Business Goals and Challenges
- Data Selection and Preparation
- Analyze Phase: Process Discovery and Analysis
- Challenge Pinpointing and Prioritization
- Optimize Phase: Deeper Dive and Root Cause Analysis
- Recommend Phase: Develop Business Case, Action Plan, Handoff and Roadmap
- Measurement and Continuous Improvement
Once in motion, work that would have taken three to six months can be done in a few weeks or a month, explained Pelz-Sharpe.
But there’s (another) caveat: “Organizations tend to be too ambitious and take on what they think are relatively understandable processes, only to discover they are far more complex than they first imagined. That can kill a project.”
It is crucial to choose well and trust the process. It’s equally crucial to approach process mining with an open mind and be prepared for unexpected revelations.
Khan views his team as digital truth-seekers uncovering digital footprints and exploring new paths, going where objective data leads them. This will be the biggest hurdle for hesitant companies to surmount. Their hesitancy will come at a cost, as these discoveries materialize at an accelerated rate in competitors’ hands.
