October 2021, Paul van Puijenbroek, DPulse
What is Process Mining?
Executing core business processes is tough through rigid and fragmented systems. For example, procurement and sales processes, billing processes, but also production processes.
Over the last 2 decades most of these processes have been automated through ERP and CRM systems. In general, these processes are complicated and the overview of what is exactly happening is often not clear. What are the throughput times why penalties need to be paid for late payment? Are products returned because of quality issues? Why are some cases late?, where are the bottle necks?, what resources are overloaded or under-utilized? What causes the deviations?
In the last couple of years new tooling, big data techniques, called ‘Process Mining’ have emerged, that support the realization of improvement and change of these core processes. Process Mining makes use of the data that can be extracted out of the automated processes/event logs to gain better insight of what is really happening. Process Mining needs a dataset that has at least three necessary elements:
Case IDs,
Activities
Timestamps.
The Case ID is the unique sequence number that identifies a unit going through a process. Data that can be used as Case ID are, for example, customer numbers, client numbers or document numbers, (like purchase orders). The activities indicate which actions have been performed in a process and the timestamps indicate when these actions were performed. With this data a model (digital twin) is made of the real process. Please see picture 1.
Picture 1: Process Mining
Why Process Mining
Process Mining will mainly be used to achieve the below goals:
Reduce costs
Improve quality
Improve productivity
Prioritize sustainability and reduce waste into your existing processes
Compliance check
For Process Mining, all processes with many transactions are eligible, for example:
Billing, This is the process of issuing invoices; Reviewing billing information, issuing the invoice, sending out the invoice.
The Purchasing order process; Identifying a need, supplier evaluation and selection, creating a Purchase order, Delivery.
Manufacturing process, tracking the movement of material in a production line.
If we compare the Traditional Process Mapping with Process Mining, we see in picture 2:
| Traditional Process Mapping | Process Mining | |||
|---|---|---|---|---|
| Process | Subjective, partial, expert | Objective, complete, automated | ||
| Duration&frequency | Long, irregularly | Immediate, regularly, often | ||
| Result | One time understanding | Continuous monitoring |
Picture 2: Traditional Process Mapping vs Process Mining
Traditional process mapping is depending on experts and is subjective, often lengthy and costly and gives a one-time understanding, Process Mining on the other hand is done by software and is objective, can be executed immediately and can be repeated at any time.
As Process Mining is automated and is done real time it gives the organization the opportunity to hand over the responsibility to the department/teams that are responsible for the process. Process Improvement becomes a continuous process allowing us to analyze the flow live, on a daily, monthly or quarterly basis.
The Methodology of Process Mining
Process Mining is based on 4 steps:
Real-Time Data Ingestion (extracting the data)
The automatic reconstruction of processes based on digital footprints i.e. data from events, captured in IT systems. Currently there are over 40 suppliers of Process Mining software.
DPulse primarily use open source technologies in the ecosystem of R, Python or Julia. There are many packages in the open source ecosystem that can be used independently or combined together to perform different steps. E.g. PM4py, ProM, BupaR, edeaR, eventdataR, xesreadR, processmapR, processanimateR, petrinetR, processmonitoR etc..
Dpulse can also help in training or usage of commercial tools such as Disco or Celonis, as the underlying principle and concept of applications are quite similar. Dpulse can help in the selection of tools to optimize the costs for our customers.
Process & Task Mining
The software will analyze the process using various algorithms. Thereafter the software visually displays the actual process.
An example is shown in picture 3.
Picture 3: The ideal process vs the real process
Certainly when you are new to the Process Mining methodology, it is desirable that an expert of Process Mining is involved in extracting and analyzing the data.,
Planning and Simulation
For Planning and Simulation a ‘Digital Twin’ is created. A digital copy of the process gives the opportunity to see the outcomes of different scenario’s through simulation. What happens if we reduce the workforce in this area, where and when should we add resources to resolve bottlenecks. What happens if we buy an extra machine for our production. What impact will this have on capacity and throughput times. If we automate this process what is the impact on resources and throughput times?
Scenario planning and Simulation will enable you to ‘predict the future better’ and hence take better decisions.
Visualization & Daily Management
The gained insight in your processes will enable you to define the right KPI’s and their related goals. Later on we will expand on how this can support autonomous teams and delegation of decision authority.
The PDCA Cycle
Most of us have already heard of the PDCA cycle. A quick summary. The PDCA cycle is an iterative process for continually improving products, people, and services. It became an integral part of what is known today as Lean management. The Plan-Do-Check-Act model includes solutions testing, analyzing results, and improving the process.
With Process Mining and PDCA to Autonomous Teams
As we saw earlier, the advantage of Process Mining is that it can be executed immediately and can be repeated at any time. Therefor the ‘C’ of Check can be done in principal as often as we wish. With the Digital Twin and big data techniques Scenario Planning and Simulation in general became much more powerful and easier to implement.
Insight into what the best process is, is usually not provided by a central knowledge authority (experts), but rather is available locally. Process Improvement must therefore go back to the people.
Feedback and feedforward is crucial for a culture of continuous improvement: you need feedback and feedforward as a human being to move forward. They are the fuel to improve your processes and to keep improvement cycles running. Without systematic feedback and feedforward and a culture of accountability, many of the improvement initiatives in an organization fail prematurely.
Regular feedback and feedforward, focused on what happened recently and the upcoming future, ensure that you can anticipate better. Feedback and feedforward becomes part of an agile climate. Process Mining makes this possible. There will be regularly measurements of the performance of the process (feedback). The scenario planning and simulation module will give the opportunity to predict better the future (feed forward), so better decisions will be taken.
Flexible, autonomous, high performance teams, consisting of various capabilities and experience levels will be able to bring this continuous improvement cycle into practice.
It goes without saying that these processes are complicated and that this requires a new way of working that will have a serious impact on the organization. Both managers and employees need to learn new competencies.
The most important ones:
The manager: problem solver, analytical, visionary, strategic, IT-minded, change-oriented and connecting
The employee: Data literate, flexible and accommodating, eager to learn, enterprising, team player, result-oriented and problem-solving
Your employees will feel empowered and motivated, your competitiveness and results will soar!
We have a lot more tools now than we had before. One of them is Process Mining - the possibility is definitely there. With the right mindset and the right support, we have a bright future ahead. See picture 4 where this process depicted
Picture 4: Process Mining and PDCA enables Autonomous Teams and Continuous Improvement.
DPulse can support organizations that want to take the steps to implement a culture of data-driven and continuous improvement. DPulse proposes to start with defining a data strategy that lays out a roadmap that shows how the different steps and milestones in this transformation process.
One thing is clear for DPulse; if you won’t do it, your competitors will…
