Discover hidden costs with Process Mining
Process Mining uses smart algorithms to discover how your processes are actually executed. Using this technique, you can easily track down and solve hidden costs, inefficiencies, and bottlenecks in your primary and secondary processes. By using Process Mining, organizations can tangibly improve their business processes and IT.
The definition of Process Mining
A short definition of Process Mining: a collection of techniques within the field of process management that supports the analysis of business processes using event logs. Process Mining uses specialized Data Mining algorithms that are applied to event log data with the goal of identifying trends, patterns, and detailed information and storing them in information systems. The goal of Process Mining is increasing process efficiency and creating a better understanding of processes.
(Process Mining Group, Eindhoven University of Technology)
Process Mining broadly works as follows: people in an organization who use a computer system leave traces. Those large volumes of stored data can then be analyzed to expose how processes work, with the end goal of designing them more efficiently.
Process Mining is a field that went under the radar for quite a while. Before Process Mining took off, people spoke of Workflow Mining, for example. Workflow Mining can be seen as a predecessor of Process Mining. How should organizations approach the great potential of Process Mining?
Process Mining in 13 logical steps
Process Mining is an analytical process that ideally goes according to a three-phase plan:
Phase 1: Applicability analysis
Phase 2: Data analysis
Phase 3: Mining analysis
Every phase comprises several steps that logically connect to each other. In total, there are 13 steps.
Phase 1: Applicability analysis
The entire process of Process Mining starts with a clear problem analysis. There are several important conditions, however. There has to be a problem: an undesirable situation that can be improved. This problem has consequences for the organization’s performance. There is also a problem owner who recognizes the problem and is willing to do something about it. They also have the authorization to do this. In this phase, the following steps are followed:
- Drafting a problem analysis
- Identifying the process context
- Determining process data
- Identifying system context
- Determining applicability
Phase 2: Data analysis
In the data analysis phase, the data is inspected, cleaned up, transformed, and modeled in order to extract the most valuable information from the data. This phase consists of the following steps:
- Drafting event log model
- Collecting raw process data
- Cleaning raw process data
- Determining validity of process data
Phase 3: Mining analysis
During this last phase, you’ll work towards the end goal: concrete recommendations for process improvements. You’ll also evaluate the entire Process Mining process, because Process Mining doesn’t stop here: it’s a continuous process. This phase consists of the following steps:
- Applying Process Mining
- Interpreting results of mining analysis
- Recommending process improvements
- Evaluating Process Mining process
Process Mining is NOT Data Mining
The terms Data Mining and Process Mining seem very similar at first glance. Both techniques are directly related to business processes. Both concepts fall under the Business Intelligence umbrella, where users strive to use Big Data to generate valuable insights for the organization. Yet there are most certainly differences to be found between the two concepts.
Look for the hidden patterns
The goal of both Data Mining and Process Mining is to generate important insights into the processes, thus enabling users to make better decisions. The role of Artificial Intelligence (AI) and algorithms is becoming more prominent in both fields. Discovering causal relationship and hidden patterns is the biggest priority here. Hidden patterns are not otherwise visible to the human brain.
Time stamps make the difference
Data Mining is a computer technique used to detect complex relations and obscure patterns in (big) data. Using Data Mining, you can analyze data to discover or predict patterns. For example: which advertising campaign leads to the highest conversion rate, or which product group performs best in the supermarket? There’s another important difference. The input for Data Mining consists of tables of data. The input for Process Mining consists of so-called event logs, audit trails, data, and events from the IT system that are provided with a time stamp.
The three basic forms of Process Mining
The goal of Process Mining is optimizing process efficiency and reducing process costs and complexity. In Process Mining, the event log is always the starting point. Process Mining has three basic forms. The event log always plays a key role.
- Discovery: This basic technique uses an event log as the basis for a model. No a priori information is used in this process. This form of Process Mining is most commonly used in practice. The outcome can be a model that you can use in the next form.
- Conformance checking: Here, an event log is used to confirm whether or not the reality as logged matches the model, and vice versa. This is predominantly applied in the area of procedural and organizational models, declaration processes, business rules and policy, and rules and regulations.
- Enhancement: The underlying idea of this form of Process Mining is to expand or improve an existing process model by using information from event logs in the actual process. Time stamps from the event logs expose bottlenecks, for example, and provide insight into service levels, lead times, and frequencies.
Example questions for Discovery
Process Mining is not an abstract, theoretical exercise. In the right context, it can provide answers to very practical questions in a variety of domains.
Possible analysis questions in the Customer Journey:
- What are the most common payment methods?
- Which factors positively or negatively impact the customer journey?
- What is the most common first point of contact between the customer and the business?
Possible analysis questions in Customer Support:
- How long does it take to process support requests?
- How long does the customer have to wait?
- What possible varieties does the customer support process offer?
Possible analysis questions in Claim Handling:
- How long does it take to replace defective goods?
- Which frequent product problems lead to more complaints than average?
- Which product versions lead to the highest sums of restitution claims?
Five examples of process deviations
Within the framework of Process Mining, conformity checking is an important aid to arrive at a clear process diagnosis. This instrument can help you find out why certain processes are stalling, or why they deviated from the regular path. Concretely, there are five categories in which improvements can be implemented in this context:
- Some activities should not have occurred at all.
- Other activities were executed by the wrong person.
- Yet other activities were executed too late.
- One or more planned activities were not executed at all.
- Two activities were intentionally or accidentally switched.
Process Mining Tools: how do you separate the wheat from the chaff?
Process Mining is becoming increasingly popular. There are now several dozens of vendors worldwide who claim to have developed tools and platforms that perfectly suit the customer needs. These vendors mostly emphasize the power of the algorithm at the foundation of their process discovery tools. For customers, it’s hard to evaluate these claims on their merits. The (perceived) quality differences can be considerable in practice. Functionality is an important selection criterion if you’re looking for the right software vendor. But aspects like data visualization and usability also deserve more attention in general.
Tool selection: Make a longlist based on 7 factors
When judging the functionality of a process discovery tool, consider factors such as:
- User-friendliness of the software: does it have clear, intuitive navigation?
- The number of relevant features: can users still see the forest for the trees? Are there essential features missing?
- Visualization options: flowcharts, decision and connector symbols in every color can quickly lead to choice paralysis.
- Integration with other analysis tools: interoperability and compatibility are the key words here.
- Clear data representation: present the data in the right form, so that everyone can understand it.
- Price: license price per user is still an important factor when choosing a vendor.
- Co-operation options / information sharing: Process Mining isn’t a solo project; it’s all about co-operation.
Also take into account the specific infrastructure in which the tool is to be used. Can it be accessed from any random computer? Is there a mobile application? Is data saved in the cloud? What is the processing time of data? What about protecting personal information and security? How fast (or slow) is the Process Mining algorithm when processing large volumes of data? And so on.
5 clear advantages of Process Mining
Although Process Mining is a relatively new field with a limited number of reference cases, some clear advantages can already be summed up as follows:
- Process Mining fulfills an important bridge function between traditional Business Process Management (BPM) and Workflow Management (WfM) systems, which don’t take into account as much event data end the more modern Data Mining (DM), Business Intelligence (BI), and Machine Learning (ML) systems, which use data as the primary starting point and are less concerned with the end-to-end process models.
- Process Mining enables you to predict processes. With Process Mining Software, you can “re-enact” the process and detect problems in the business process, potentially using Gaming techniques. With this same goal in mind, you can also “replay” Process Mining software processes. The possibility to use Process Mining to simulate processes and use Artificial Intelligence (AI) to predict processes as accurately as possible is very promising. Time stamps play a crucial role in these kinds of analyses, because they can precisely date the chronological sequence of events.
- Process Mining makes intelligent use of Big Data. Process Mining solutions are focused on improving both organizational performance and the compliance requirements employees are expected to meet. Process Mining is also the hinge between process model analysis (simulation, verification, optimization, gaming, etc.) and data-oriented analysis (Data Mining, Machine Learning, and Business Intelligence).
- Process Mining eliminates the bottlenecks. The challenge of Process Mining is turning the enormous amounts of Big Data into valuable insights that say something about the performance of processes and the compliance factors. Process Mining will give you a better understand of the bottlenecks, inefficiencies, aberrations, and risks of failure within your crucial processes.
- Process Mining improves the operational processes. Acquiring non-trivial, process-related insights using data analysis is a relatively new development by PhD students (such as R.P. Jagadeesh Chandra Bose), consultants, and commercial software vendors. Operational processes can be improved and safeguarded using the acquired insights, for example in the form of process models. Applying Process Mining to event data from information systems has been proven to lead to insights and practice, according to scientists. The application of Process Mining on event data that isn’t generated by information systems, but by all kinds of devices such as X-ray scanners, CT scanners, copy machines, printers, etc. is also taking off.
Where can I use Process Mining?
Process Mining is already proving its worth in countless industries. Process Mining is a relatively new concept that’s well-suited to Big Data analysis in a variety of industries: from credit card companies, industrial companies, and real estate, to government institutions and hospitals. Consider the possibility, for example, of an analysis of all the necessary process steps in the treatment of a hospital patient. Or all the actions a housing corporation has to go through before a residence can be presented to a new tenant in compliance with all the procedures and regulations. Process Mining has also proven its value in the case of testing the so-called wafersteppers in the semi-conductor industry.
5 pitfalls when applying Process Mining
Whenever you apply a new technique, all kinds of pitfalls or unwanted side-effects can crop up. In the case of Process Mining, this is no different. The Process Mining pitfalls are related to the human aspect on the one hand, and the data aspect on the other. Here are the five most important pitfalls.
- Pitfall 1: Consultants, auditors, quality managers, and process owners are still relatively unfamiliar with the phenomenon and are largely ignorant of the possibilities presented by the current Process Mining tools. You can get your Six Sigma Black Belt or become a certified internal auditor without ever having studied Process Mining. That says something about the field’s status, or lack thereof.
- Pitfall 2: As a result of the current emphasis on machine learning and AI, people don’t realize that Process Mining is something completely different.
- Pitfall 3: Middle-managers are especially intimidated by the results and insights generated by Process Mining. The increased transparency of the processes can bring to light mismanagement, inefficiency, and compliance problems. That may seem threatening.
- Pitfall 4: Lacking data quality and access to the data can also create an impediment. Realize that the 80-20 rule applies here: the precursor track (locating, selecting, extracting, and transforming the data) takes up 80% of the time. The Process Mining track itself usually only takes up 20%.
- Pitfall 5: Process Mining usually runs into data quality issues that you immediately have to tackle. That’s a distraction.
How does Process Mining fit within data-driven working?
Most information systems continuously save what’s happening in the form of so-called event logs. That data is often hidden somewhere inside of the system or computer. Also, the volume of data stored by traditional workflow systems (and other systems) is growing exponentially. One of the biggest challenges for organizations is innovating, growing, and strengthening their competitive edge by making intelligent use of all the available data. Within this context, Process Mining has natural synergy with the existing concept of the intelligent organization and the growing interest in data-driven working.
3 tips to maximize the returns on Process Mining
Besides avoiding the five pitfalls mentioned earlier, these three tips may also help you implement Process Mining successfully.
- Elephant trails are effective, but undermine the process. The divergent routes people take when striving to realize goals are comparable to elephant trails, from the Process Mining perspective (Van der Aalst, 2013). An elephant trail is an unofficial path that is intentionally and unintentionally created by users of the regular path over time. In organizations, too, people sometimes skip certain steps of the process, because they’re convinced that it leads to faster results. In practice, sometimes the Central Purchasing department is skipped, because people don’t want to have to deal with all kinds of bureaucratic procedures.
- Walk the Happy Path and stay in the Happy Flow. It should be obvious that the elephant trails mentioned above can cost organizations a lot of money, for example because they cause the organization to miss out on discounts that the Central Purchasing department has arranged with certain vendors. In an ideal situation, employees don’t deviate from the beaten paths, and they walk the planned route. Within the field of Process Mining we call this the Happy Path or Happy Flow.
- Realize that business processes are the foundation of your strategy. An intelligent organization continuously monitors its performance by translating the strategy into KPIs and so-called Process Performance Indicators. Process Mining software is an essential component when striving for a data-driven, intelligent organization.
Process Mining: a combination of science and practice
Process Mining came into being thanks to, among other factors, scientific research by the Technical University Eindhoven. The inventor and one of the trailblazers and promoters of this new field in the Netherlands is the professor Wil van der Aalst, who is building Process Mining software with the scientific community and PhD students. These days, there are also several dozen commercial software vendors who support Process Mining with the software solutions. The rise of streaming social media, sensor data from the Internet of Things, and technologies like RFID have led to ever more event data becoming available, which is excellently suited to analyzing using Process Mining software.
Process Mining is a specialist field
In conclusion, Process Mining is the perfect way to visualize and analyze the actual execution of business processing, by using the logged process data in operational/transactional systems, such as the ERP system. Process mining, on the one hand, requires insight into business processes, process analysis techniques, and process compliance. On the other hand, it also requires specific knowledge and skills about generating and validating reliable process data and the application of Process Mining software. Process Mining is a specialist field, in other words. Our Process Mining consultants are eager to help you.