Differences between Process Mining vs. Process Discovery vs. Task Mining

Let's look at 3 terms which are easily confused.
So what are these technologies on their own, what do they have in common, and how do they differ?

When companies engage in digital transformation initiatives, they stumble upon the problem of having too much choice. Too many "different" technologies, too many vendors, too many consulting companies all talking about different things. Not only that, terminology within the industry gets confusing as well. 

Today, we will demystify the differences between three popular terms, which are easily confused. Hopefully, we will make your life easier by explaining those and providing as straightforward a comparison as possible. 

Process miningprocess discovery, and task mining are the trendiest terms when it comes to discovering business processes. We could say that all these three technologies are used as the first step in most automation initiatives nowadays. Since they help map the processes, the results you get greatly assist in decision-making. Based on the derived knowledge, companies might undertake the subsequent actions, one example for that being RPA. So what are these technologies on their own, what do they have in common, and how do they differ?

What is process mining?

Process mining is a powerful process mapping tool used for end-to-end process optimization. It uses data from event logs that are available in IT systems. Based on that data, it builds the as-is process and then compares it to the "desired" process. By doing this conformance checking, companies can find process deviations and thus improvement opportunities. 

What is process discovery?

Process discovery is a combination of different techniques, including Machine Learning and computer vision, to map the processes. These results are then used as a prerequisite for automation initiatives. It essentially monitors and captures user interactions with different information systems and applications, recording employees' digital traces. Process discovery uses various computational and statistical methods to get valuable, applicable information from the data it gathered. 

What is task mining?

Task mining is another process mapping tool focused on task optimization on desktops. Much like process discovery, it monitors the digital traces of the users. Using character recognition, natural language processing, and other tools, it analyzes the data gathered and finds patterns that could be interpreted as improvement opportunities. 

This might sound confusing at first, but comparing them step by step would definitely make it easier to grasp. Let's have a look at the following 5 criteria. 

Process discovery vs. task mining? 

It's easy to feel overwhelmed by the presence of hundreds of definitions of the same thing. In fact, different companies might use their own definitions for some of the terms, making it even more complex and blurring the boundaries between them. 

Let's start with process discovery and task mining because they have been used a lot interchangeably. These tools are very similar in terms of the technologies they use and how they reach the desired goal. Nonetheless, task mining and process discovery are used for different end goals. Let's look at the more prominent service providers. There's a distinct pattern that process mining companies use task mining as their complimentary service, while process discovery is used by RPA vendors. 

Both technologies monitor user desktops to record users' interactions with different apps and IT systems. While task mining does it to identify process inefficiencies, process discovery needs the data to find automation opportunities more effectively. As a result, although they're very close to each other, they might slightly differ in terms of the outcome they bring.

We could pause here and be done with process discovery, but the world of terminology is too complex for us to do so, so it is used in another case. Process discovery is also used to describe the first step within process mining (process discovery, conformance checking, enhancement).

In a nutshell, process discovery might be used as

  1. Complementary technology for RPA.
  2. One of the steps within process mining. 

Process mining vs task mining?

Task and process mining often go hand in hand and get used together for knowledge work discovery. While both are pretty powerful, they bring better results and higher work coverage when used simultaneously. To fully understand what each of them does, let’s discuss the difference between processes and tasks. 

As discussed, process mining brings visibility to end-to-end processes. The process is a logical chain of individual events with a clear starting and ending point. Processes can be divided into subtasks that make a complete set of related work. Purchase-to-Pay, for example, is a large process that runs through multiple teams. 

On the other hand, tasks are smaller components of work that occur between different processes and subprocesses. Copy-pasting the data, uploading and downloading the files are typical examples of manual tasks.

Process mining vs task mining
Processes vs tasks

Typical process mining use cases:

Process mining can be used in many different fields for various purposes but here are some of the more popular use cases. 

  • Auditing & compliance
  • Order management
  • Purchase-to-pay
  • Order-to-cash
  • Lead-to-order
  • Logistics

Typical task mining use cases:

  • Data entries
  • Invoicing
  • Reporting
  • Reconciliation

Differences between process mining, task mining, and process discovery. 

Let's briefly discuss how process mining, discovery, and task mining differ. Since the latter two interrelate a lot technology-wise, we will be using them interchangeably in this section. 

Origins of data

Process mining uses the event log data found from different information systems such as Salesforce, Oracle, or HubSpot. These event logs contain information about the performed activity, the ID of the activity, and its timestamp. 

Process discovery and task mining, in contrast, collect the data through monitoring and recording how users interact with the computer, capturing all the processes with the help of software agents. 

How the data is gathered 

The data has different origins, so it is gathered differently as well.  

IT support and upfront back-end development are required in the case of process mining to make integrations with the monitored information systems. 

For process discovery and task mining, software agents must be installed on the users' computers. Agents are pieces of software that work on the user's devices continuously in the background. These agents "record" everything that has been done within business software and applications. 

Completeness of the collected data 

It's worth mentioning that some software and apps do not produce event logs, considerably limiting the possibilities of process mining. For example, suppose the goal is to map the current e-invoicing process and find improvement opportunities. In that case, process mining analyzes the distinct steps within whatever e-invoicing platform is used. If the invoicing expert has to use that platform and some other apps, let's say Excel, to execute that process, then the steps in Excel get overlooked, thus leaving more space for bad outcomes. So it can only capture the discrete data in the particular steps of the process and leaves the white spaces between logs that are out of the scope of its discovery. 

Process discovery and task mining can collect the data from log-producing information systems and other productivity apps and software that employees use, such as email, Microsoft suite, etc. This makes them a perfect complementary tool for Process Mining and RPA because they can provide slightly more information on shadow activities otherwise missed. 

Analysis of the data

Process mining, once the data is collected, cleaned, and structured, re-constructs how the current process looks and compares it against the ideal process - how the process should be. This is called conformance checking; multiple data analytics, mining, and data science methods are used. It finds the possible bottlenecks and suggests improvements based on those deviations. 

Process discovery first records everything on the users' desktops to capture processes just the way they are - with all the random deviations and flaws. It presents the process in a way it was actually executed by humans, creating a metamodel of the process using computer vision, machine learning algorithms, and AI tools. This way, root causes and various bottlenecks can be identified easier. 

General limitations

Process mining is limited to the steps of the process within particular IT systems. It ignores the human factor, overlooking the users' digital footprints in other applications, such as personal productivity apps, documents, etc. 

Although process discovery and task mining bring the human touch that process mining might miss, they work the best with shorter and smaller tasks. Because analysis methods used are computationally heavy, it might start bringing false results when analyzing longer processes and fail to identify the importance of a task in a larger context.

Common points

Process mining, task mining, and process discovery are powerful tools that have been used by companies for a few years already, bringing good results and helping even the largest enterprises to find their automation potential. 

Which one is right for you?

Depending on the company, the use case, and other factors, one or any combination of these can be more suitable than the other. Process mining is generally better at harmonizing the distinct processes and conforming them to the end goal, especially if they are done within extensive IT systems. Process mining is good at seeing the reality in heavy systems like ERP, CRM, and Supply Chain Management, and optimizing the processes within them. Task mining is a good alternative if the process is done across different apps and those small activities in between more significant steps matter. It's pretty good at showing the human-machine interactions within the task. Similarly, process discovery is mainly used to replace manual process discovery techniques so that finding automation opportunities is faster than with existing old methods. 

However, as we've discussed, all have certain flaws that might not work out for some companies. These can work in tandem for a more complete analysis of the processes. Still, they might still not cover the full spectrum of business work-related activities in an anonymous way. Work intelligence could be a perfect solution that combines the best of both worlds and beyond that. It helps to see what needs to be automated in the first place before jumping into the pool of many automation-related tools.

Where can work intelligence bridge the gaps?

Workfellow's work intelligence is an all-in-one tool, everything that your enterprise might need to start its digital transformation journey. 

Discover, explore and plan your process optization with Workfellow

Get a free introduction to Workfellow through a demo or free trial. Hit the button below and let's get started!

Get started