Process Mining vs. Process Discovery
Robotic Process Automation (RPA) continues to expand across the globe. It’s expected to reach $7.2 billion by 2025, at an impressive 32.6% compound annual growth rate (CAGR), even in the face of the economic slowdown in 2020.
As RPA is increasingly used by global enterprises to automate front- and back-office processes, the question comes up: Which process should we automate next? After all, you don’t want to just automate anything because you can. You want to find the right process to maximize the efficiency gain from automation.
According to “The New RPA Manifesto,” released this year by HFS Research, RPA is not about eradicating manual work. It’s about getting rid of inefficiencies. You must estimate the potential return of investments for a particular process before you automate it. Failing to do is probably why some RPA initiatives fail.
Because of this, many businesses are turning to process mining and processing discovery. Both activities help you map out business processes — existing ones as well as discovering new ones — and technology solutions abound to help you take either approach. But many people are confused about what these terms mean and the differences between them.
Many assume they’re synonyms; they’re not. You’ll get different results depending on which you use — and if you’re hoping to integrate the results with your RPA initiatives, only one really does the trick.
Some basic similarities
Both process mining and processing discovery use automation technology to identify and map processes in your organization. Whether it’s your HR department, finance, or other function, you can use both techniques to find out how your processes are working. But there the resemblance ends.
Mapping digital steps
Process mining maps out the visible digital steps in each process. By analyzing system logs through sophisticated algorithms, process mining tools automatically identify and evaluate automatable work processes. At the highest level, process mining extracts knowledge from event and application logs to gain insight into business processes. The goal is to understand and improve processes based on factual information that is collected during the execution of these processes.
According to textbook by Wil M.P. van der Aalst, process mining can do three things: discover process models (create a new, optimized map of what a process should look like); compare an actual process to an “ideal” process to see how it measures up or conforms; or enhance your understanding of an existing model with new information about performance, cost, or other factors.
Introduced in the late 2000s, data mining is a powerful way to identify, analyze, and optimize business processes.
Exploring the white space
If process mining is what processes are (or could be) in place, process discovery is the discovery of how humans execute business processes. Using newer technologies, such as computer vision and machine learning, process discovery identifies what actually happens whenever a human interacts with a system, as well as in the “white space” between logs.
This white space is the digital representation of the business as created by all the human things that users do that diverge from the business processes at hand. This digital representation is valuable because these activities are generally excluded from traditional forms of process mining.
In other words, with process discovery, you identify the steps in a process through the (human) user’s interaction with systems. The good news is that process discovery allows you to identify the visible and invisible processes.
So, what are the main differences?
Process mining has some limitations. First, it works only for systems that produce logs. But many business processes take place in Excel, on a Teams chat, or other personal productivity tools. Those activities are simply not seen by process mining solutions. Also, legacy applications, productivity applications such as Microsoft Office, and terminal and virtual desktop environments such as Citrix are also invisible to process mining.
Most importantly, these solutions don’t see users’ interactions with systems and, thus, are difficult to integrate with RPA solutions. You generally have to start building processes from scratch — using what you found out with process mining, of course — within your RPA solution.
Process discovery was designed to overcome these shortfalls. Process discovery solutions record user interactions with a broad range of systems, including enterprise solutions such as ERP or CRM, personal productivity applications such as Microsoft Office, as well as virtual desktop environments.
When it comes to RPA, process discovery solutions offer many advantages over process mining. You don’t need logs, databases, or API access — unlike with process mining. A process discovery solution:
- Relieves you of integration hassles with your systems
- Incorporates computer vision and machine learning to reveal the business processes taking place in the white space
- Continuously monitors the processes to ensure real-time analysis and rapid retraining in case you make changes to them
Choosing which solution is best for you
Process mining and process discovery help you understand your business processes. You are generally choosing to go through a process mining or process discovery exercise to automate processes or make them more efficient. Both approaches have their advantages and limitations, but the main differences are:
- Process mining gives you complete log-based visibility into processes into what is happening within your enterprise systems such as ERP, CRM, and the like.
- Process discovery leads to process automation using RPA. It records all user interactions with systems, analyzes repetitive actions, and creates RPA bots to perform those actions automatically.
Process discovery is a natural fit for leading RPA platforms. And Automation Anywhere Discovery Bot, which is now available as part of the Enterprise A2019, can help you make the most of process discovery. Discovery Bot helps accelerate and scale automation across the organization by recording user activities, discovering business processes, identifying automation opportunities with the highest business impact, and generating bots to automate them.