“It is a capital mistake to theorize before one has data.” — Sherlock Holmes  


What does Sherlock have to do with one of the most important process mining techniques – conformance checking?

The connection between process analysis and a British detective might seem farfetched at first glance. But with a bit of imagination, one could argue that a process analyst is looking for clues about problems in the process like Sherlock searches for clues about his cases.  

Regardless of whether you view process analysts as some sort of detective, there unquestionably lies some truth in the famous quote. 

One of the main interests of process owners and process analysts is to find out whether the process is executed in the way it is defined.  

Most companies have a clear definition of how their processes need to be executed, often in the shape of a process model. During the actual execution, this model might not always be followed closely, and cases might deviate from it. Finding these deviations is in the interest of each company since they often represent aspects of optimization. 

Accept the challenge 

Finding these deviations is not as simple as it might seem.  

The first impulse might be to call a big brainstorming meeting where everyone shares the challenges they are facing when executing the process. This meeting will surely result in a list of things to improve.  


However, employees are subjective and might have a limited view of the process based on their role. Because of that, it is difficult to retrieve a full overview of the process.  

The likelihood of creating theories about the problems in the process, based on subjective opinions, is high. Here we run into what Sherlock tried to warn us about – we have drawn conclusions without enough data and information. 

Use conformance checking 

Conformance checking provides a solution to these problems by being a data-driven analysis method 

Nowadays, many processes are supported by some sort of business information systemSuch systems allow enterprises to generate data in an IT environment and support decision making. 

These systems also record data about each step taken in the process. This data is the optimal basis to find cases where the process execution differs from its definition.  


Given a process model that defines the process and the data recorded during the process execution, the conformance checking algorithm maps the data onto the process model and highlights cases where they differ from each other.  

The picture above visualizes the results of conformance checking analysis. Blue parts represent a match between data and process model, while yellow indicates a mismatch.  

There are various amounts of conformance checking algorithms available, but in general, they all detect two main types of deviations. 

1. Non-specified steps of a process

On one hand, conformance checking highlights additional steps that have been taken during the process execution but not specified in the process model.

These might be activities that are not part of the process model at all, as well as defined activities occurring in the wrong part of the process. In the picture above, they are depicted by yellow activities.  

2. Skipped steps of a process

On the other hand, conformance checking also shows cases, where steps of the process model were not executed but skipped.

This case is visualized by a yellow arrow, which flows in the same direction as the blue ones, connecting two blue activities. Yellow arrows that connect two blue activities but flow the opposite direction as the blue arrows, usually depict rework since an already executed activity had to be repeated.  

Beyond that, ProcessGold has enriched the conformance checking capabilities by not only analyzing whether the tasks were executed in the correct order but implementing a possibility to specify business rules. 

Which rules are those?  

A company might have business rules about their process such as “Customer questions have to be answered within three business days”.

Or “Purchase orders cannot be approved and paid by the same employee”.

These kinds of rules can be implemented in ProcessGold so that the tag manager tool will analyze which cases violate them. 

Take action 

But what to do with the results obtained from a conformance analysis 

Now that we have followed Sherlock’s advice and created an analysis based on data, the next steps are:

  1. Understand the results in the context of the company
  2. Define actions for improvement 

The goal is to find the reason behind the deviations and to provide a solution for it. Depending on situation it might be possible to find the reason within the data. Although, often it is necessary to talk to the employees executing the process.  

This might reveal special cases where they must perform additional steps due to missing information, or cases where certain steps are not necessary.

To improve the process execution, it is important to learn from the employees’ experience. Providing them with extra information to be able to execute all their tasks or adjusting the process model to avoid unnecessary steps, will improve their performance and avoid problems in the future.  


Based on the conformance checking results, it is also easy to prioritize which problems to tackle first. The analysis shows exactly how many cases violate the process definition, which can be used to rank the deviations. 


In summary, conformance checking is a technique that reveals problems in the process execution. 

In comparison to traditional business process management techniques, it is based on the data collected during the process execution.  

The visual representation of the results helps to spot deviations easily. It serves as a solid base for further discussions.  

Using conformance checking will help to optimize the daily work of a company in a structured way, founded in data.  

Learn more about conformance checking tools of our platform or talk to our team to test these features by yourself.  

Happy mining!  

Anja Syring, Consultant @ProcessGold