top of page
Navigating Process Conformance Checking Challenges

Navigating Process Conformance Checking Challenges

Lessons on overcoming Process Conformance Checking challenges

Authored By: 

Sean Ferguson and Cameron Marro

May 21, 2024

This is a continuation of a previous article – Improve Operational Efficiency with Process Conformance Checking. 


Mature organizations rely on well-defined operational processes to run their businesses. We’ve previously investigated Process Modeling and Process Mining and how they are each used to improve business functions. Bringing them together helps unlock new potential for opportunity within your business. This is the magic of Process Conformance Checking (PCC).  

You can read more about PCC here, but the basic idea is that with well-defined process models and a Process Mining implementation, you can measure your actual process’ conformance against the design. This allows an easy way to determine where breakdowns might be happening in your process. Are activities happening out of order (like receiving an invoice before a PO is created)? Are there unwanted activities occurring (like customers returning goods)? Are SLAs being exceeded based on the business’ requirements? 

The benefits are far-reaching and relatively obvious when you consider some of the value that can be derived. However, there are often some challenges with implementing PCC. Some common issues include the readability of standard process models, the language of process models, and the ability of data to compare the actual process performance to the ideal model. While this list is not exhaustive, we’ll discuss each one of these separately and how they can be addressed. 

Challenge 1: File Readability and Import 

To understand the first challenge (readability of standard process models), consider that a company may have a central repository for its process models. Even in this fortunate situation, it’s quite possible that the documentation methodology isn’t standardized or optimal. And even if it is designed to import business process documents, there may be file formatting issues that prevent the Process Mining software from correctly interpreting the business process model. 

Your Process Mining solution needs something against which to compare the process it mines. In the world of Process Modeling, there are many formats to choose from. Perhaps the most common is a BPMN (XML) file. However, business process models can also be stored in CSV or more proprietary formats like Visio or Nimbus. Most Process Mining programs are limited to running conformance checks or importing models to a specific type of file format – if they can import at all. If your process models are stored in a separate program and created in a format that can’t be ingested, you typically have one of two options. Option 1 is to do the work on the front end by converting your file into something that can be imported and read or copying/rebuilding your business process in an acceptable format. Option 2 is dependent on the Process Mining tool you are leveraging. However, most can design a baseline model within the Process Conformance (or similar) function. Several tools even allow you to mine your baseline process based on actual variants that have been identified. Not only is this feature exceedingly convenient, but you can even leverage it to improve your baseline process model! 


Challenge 2: Language Discrepancies 

The language of process models can create an additional challenge. Suppose you can import your business process documents and the Process Mining tool can read them. This may not be enough to accurately conduct PCC because the language of activity steps in the process model may not match the same language in the Process Mining software depending on how activities are defined. This is obvious in a multi-national corporation where a process model is created in one human language, but Process Mining is deployed in another human language. However, even if the human languages between the process model and the mined process are the same, your Process Mining software may not be able to readily understand that “Cancel Order” in one system corresponds to “Order Cancelled” in the other. Without this consistency, Process Mining suites can struggle comparing processes. 

To circumvent, you again have the option of editing the underlying process model before ingestion/import into the Process Mining tool. Conversely, you can also change the inputs to the Process Mining data model itself to align more closely with the standard business models. Consider that the definition of activities is typically done within the data pipeline during the creation of an event log. By updating or updating the logic used to create the event log, the data engineer can conform to the standard business models designed by the organization. Finally, in the best-case scenario, some tools provide mapping capabilities to account for the translation between the event log and the process model. By leveraging the front-end user interface, you can use this functionality to directly map “Cancel Order” in a process model to “Order Cancelled” in the mined process without changing any of the underlying data upon which other analyses may rely upon. 

Challenge 3: System Data Availability 

Finally, there is the issue of system data availability. Imagine a business process model we are using to manage a purchasing process. I may have a step in my business process model that indicates that a purchaser must first review a purchasing catalogue for standard requisition items before they are able to submit a free text requisition item. With Process Mining, we have no way of knowing whether a purchaser thoroughly checked the purchasing catalogue before submitting the free text requisition item. This is because Process Mining relies on transactional source system data and the timestamps captured within to build the event log. That data is never created when a user is sifting through a purchasing catalog. Therefore, we can’t ever know if they conformed to that part of the process by using Process Mining. 

For this scenario, there is often no quick fix. The limitations of Process Mining mean that our process models need to be simplified for conformance checking. In a future article, we will explore Task Mining and how that can provide some user-based insights that can help business understand how work happens and not just what work happens. Within the realm of Process Mining, though, process activities and decisions that happen outside of the transactional source system cannot be measured for conformance. 

While there are often some challenges, if done correctly, PCC can be an immensely powerful and valuable capability to ensure that processes are done right. The way these challenges are overcome often depends on the Process Mining software in use. By leveraging the tools at your disposal and some practical problem solving, PCC can enable better root cause analysis and help you unlock new efficiencies in your processes as you march toward your ideal end-state process. 

About The Author(s)

Sean Ferguson and Cameron Marro

Sean Ferguson

Sean began his career at TranSigma in 2018 after earning his master's degree in business administration with a concentration in finance from Sacred Heart University. As a member of the TranSigma team, he has helped clients from the Defense and CPG industries across many functions and disciplines including Cybersecurity, Human Resources, Finance, Procurement, and Accounting. Recently, he has become the leader of TranSigma’s Celonis practice. His expertise and passion lie in implementing new and innovative technologies into organizations so companies and people can leverage the power of technology to unlock maximum potential.

Cameron Marro

Cameron Marro is a dedicated professional specializing in process optimization and increasing efficiency. Graduating with distinction from Southern Connecticut State University in 2021, she joined TranSigma to manage back-office operations and develop their process mining practice. She specializes in improving processes in the Banking, Financial Services, and Insurance industry across multiple business functions.

bottom of page