How Process Mining is Revolutionizing Business Process Management
Process mining offers a data-driven and efficient approach to process analysis that resolves the pain points of traditional methodologies.
By David Sizer and Maximilian Holsman
Amid COVID-19 lockdowns, supply-chain instability, and rising geopolitical uncertainty, it’s vital that organizations thoroughly understand their business processes to become operationally resilient to the constantly evolving business landscape. Only by understanding their processes can businesses adapt when volatile external factors impact their performance.
However, the traditional approach to process discovery and improvement relies heavily on subjective interviews and workshops, making it time-intensive, invasive, costly, and — most importantly — inaccurate. Increasing levels of organizational complexity and digitization also mean that business processes are becoming less and less visible to this human-centric approach.
Even in best-case scenarios, the traditional approach can only result in one-time improvements with no ability to measure or monitor the impact of process improvements on KPIs. This makes process improvement initiatives very difficult to evaluate and leaves the business’s performance extremely susceptible to rapidly changing conditions.
What is process mining?
Process mining offers a data-driven approach to process discovery and optimization that objectively and efficiently reveals what business processes look like according to data, as opposed to human perception. It enables businesses to implement improvements that will optimize their processes and performance while quantitatively measuring the impact these improvements have on KPIs.
By enhancing businesses operational resilience through continuous process optimization and enabling impactful synergies with robotic process automation (RPA), machine learning (ML), and data analytics, process mining is defining the future of business process management across all industries. It has the power to transform businesses — helping them achieve their goals and change the world — and is therefore the keystone to Slalom’s process optimization approach.
Leveraging event log data
Processes are the heart of any business. They’re how things get done, how inputs get turned into outputs, and how value is created. Any given business process is made up of individual business cases — such as individual customer orders or invoice requests — that move through the stages of the process from creation until completion. At each stage of the process, a case leaves behind a piece of data in the underlying enterprise systems, referred to as an event. An instance of event data specifies at least three pieces of information: at what time was what case at which stage of the process. For some enterprise systems, instances of event data can include additional information, such as the size of the associated customer order.
As each case moves through the process, the event data it leaves behind at each stage is recorded in the system’s event log. Therefore, the collective event log data from all cases provides an accurate picture of the entire business process and how each case proceeds through it. Process mining software take these system event logs as input and use them to reveal this picture, producing an accurate model of the business process in all its complexity.
Want more ways to unlock hidden value in your organization? Learn how to empower your teams to drive efficiency and focus spend on what matters most in turbulent times. Get the guide.
Benefits of process mining: A data-centric solution to business process management
So why should business leaders care about process mining? What insight can it reveal that isn’t already provided by interviews and workshops with the people who carry out these processes?
Process mining thrives at the pain points of traditional process analysis. Because it produces a model that relies solely on the event log data, process mining captures the reality of a process and doesn’t depend on the subjective view that employees or managers may have of it. It also accounts for every case recorded in the event log, capturing all variances and exceptions within the process and not just the most common or best-case “happy path.”
With this in-depth, end-to-end understanding of their processes, businesses can identify and eliminate inefficiencies that would otherwise remain hidden. For example, a recent Slalom engagement with a manufacturing client used process mining to uncover a major inefficiency in the client’s accounts payable process that resulted in $22 million of wasted spending.
Since new event log data can be continuously fed into the software as it’s collected, process mining can continuously monitor the current state of the process as it changes with the most recent cases. Unlike the non-iterative “process snapshot” provided by the traditional approach, process mining produces a dynamic model that evolves with the surrounding businesses landscape. Through this continuous monitoring, process mining enhances a business’s operational resilience providing the tools to quickly identify and react to the impact that the changing business environment has on its operations. As an example of this, Slalom implemented process mining to provide a consumer goods client with continuous insight on which credit holds would optimize its order-to-cash process, resulting in a nearly 20% reduction of holds.
As most business processes today take place within enterprise systems, process mining can be broadly applied to almost any process or industry. Its wide range of use cases include using it to optimize the order-to-delivery process and maximize on-time orders in order management, increasing free cash flow and reducing incorrect payments in accounts payable. Depending on the objective, process mining can be used for anything from simple process discovery to complex process enhancement and optimization, making it an invaluable tool in all initiatives.
Opportunities for impactful synergies
While process mining itself is already a transformative technology, its power is amplified through synergies with other business technologies, such as:
Task mining
Task mining captures the parts of a business process that take place outside of enterprise applications and therefore don’t leave behind event log data. By using optical character recognition (OCR), natural language processing (NLP), and machine learning algorithms, task mining tools can capture desktop-level user actions, such as mouse clicks, keystrokes, and application logins. They analyze this desktop-level data to understand what tasks users are completing across applications and use them to identify key processes.
Combined with process mining, task mining can produce a comprehensive, end-to-end view of entire business processes, including the parts that take place locally on desktops (as opposed to enterprise applications), giving businesses even deeper insight into their operations.
Robotic process automation
Robotic process automation (RPA) refers to the use of software tools to automate repetitive manual tasks, freeing up employees to focus on more value-adding tasks. Often, the difficulty with applying RPA is identifying routines and workflows that can be automated effectively to create long-term sustainable value instead of just short-term cost savings. By providing a model that captures the end-to-end business process in all its complexity, process mining enables businesses to discover routines that can be automated and determine which automations would be the most impactful.
Process mining can also continuously monitor performance, making it easy to evaluate RPA initiatives and determine ROI. According to process mining provider QPR, leveraging process mining during the implementation stage of RPA initiatives leads to a 40% increase in RPA business value, while decreasing implementation time and project risk by 50% and 60% respectively.
Artificial intelligence and machine learning
Machine learning (ML) generally refers to statistical models and algorithms that learn from past data to make decisions and predictions about the future. ML models thrive on large data sets of past information, using this data as input to inform their decisions. This makes ML an incredibly impactful complement to process mining, as it can leverage the collected event log data to make decisions and predictions about the process. Because of this synergy, ML can be leveraged in several areas across process mining initiatives.
When process mining has identified a problem or bottleneck within a process, ML can be used to diagnose its cause. This so-called “root-cause analysis” uses the event log data to find correlations between certain events and a given problem in order to determine what caused it. Also known as diagnostic process mining, a root-cause analysis might find that whenever a certain material is needed for production, the associated order is very likely to be completed too late. By eliminating identified root causes, businesses can optimize their processes and drive performance.
By leveraging the predictive capabilities of ML, process mining can also use past event log data to make predictions about the future state of a process — a technique called “predictive process mining.” Based on what’s happened in the past and the current process state, these algorithms can predict the outcomes of ongoing cases, resulting KPIs, and what future events are likely to happen next. For instance, predictive process mining can use past event log data to forecast which ongoing orders will be on time and which will be late. Understanding the predicted future state of their processes enables businesses to intervene and prevent negative business outcomes.
Lastly, once predictive process mining has been used to identify possible future problems, ML algorithms can be leveraged to recommend actions that optimize the predicted future performance and avoid forecasted issues. This is called “prescriptive process mining.” By exploring a large set of possible process changes, prescriptive process mining can find the best courses of action based on past event log data. Some process mining tools can even implement these actions automatically through RPA, resulting in constantly optimized process performance.
For example, a prescriptive process mining algorithm might predict that production capabilities will be overloaded once a large incoming order enters production, leading to several late deliveries. This allows the algorithm to prioritize certain ongoing orders to free up capacity once the large order arrives, optimizing overall production efficiency and maximizing the number of on-time orders. These algorithms are also cost-aware, meaning they can optimize between the cost of any intervention and the cost of the predicted problem to identify the best course of action.
Conclusion
Process mining offers a data-driven, efficient, and accurate approach to process analysis that resolves the pain points of traditional methodologies. By leveraging the event log data left behind in enterprise systems, process mining captures an objective, end-to-end view of a process in all its complexity. It enables businesses to understand the reality of their processes, giving them the insights they need to optimize their operations. Instead of just offering a one-time analysis, process mining can continuously monitor a process and its performance so that organizations can react quickly when external conditions impact their operations.
Finally, the data-driven nature of process mining results in several powerful synergies with other business technologies — such as ML, RPA, and task mining — that further amplify its impact. Despite having already disrupted several industries, process mining is still in its relative infancy, and its ability to fundamentally change how companies operate will only grow as technology advances.
Slalom is a global consulting firm that helps people and organizations dream bigger, move faster, and build better tomorrows for all. Learn more and reach out today.
