What is Task Mining? A discussion with our CEO

There’s a new kid on the block – Task Mining. Our CEO introduces newly developed task mining tech, describes how it works and the differences between it and Process Mining.


Introduction

As companies are looking past RPA implementations and towards Hyper-Automation or Intelligent Automation, they need to understand new process related technologies such as Process Mining. However, there’s a new kid on the block – Task Mining, and people are beginning to ask me what this is all about.

Essentially, Process Mining and Task Mining both have the same goal: to understand business processes automatically without human interaction.  However, they go about it in two completely different ways and are not interchangeable. They have their own set of advantages and disadvantages and are useful for two separate things. Yet, synergy can exist when they are combined together.

Terminology

Firstly, I’d like to take some time to explain the evolving terminology because at the moment it can be quite confusing.

Sometimes new technologies can take some time to settle on a universally accepted name or term. When this is happening, one or more terms can be used for essentially the same thing.  It’s still happening with Process Mining (formerly called ABPD, now becoming increasingly known as Process Intelligence).  

With Task Mining, the names could leave you scratching your head. Task mining is sometimes referred to as Process Discovery – a general term which also refers to the traditional activity of interviewing users to understand a business process.  I’ve also seen it referred to as Process Mining, which is just downright wrong and I sometimes wonder if deliberately misleading.  One company has even started to call it ABPD which is a bit of a throwback.  At Arkturus, we think Task Mining is the right term for what it actually does, which has more to do with the tasks a user performs rather than complete business processes.

What is it?

Task mining is a new technology that watches what a user does and generates data from it.  It then uses that data to form a very detailed understanding of the steps or tasks that a user performs when they do their work.

How does it work?

Task mining requires the installation of a small program on user desktops that run in the background, and monitor everything that a user does.  It captures keystrokes, mouse movements, switching between applications, opening of files, pressing buttons – basically everything a user does.  It also records screenshots and uses OCR to extract information (field labels and field contents) from those screens.

From all this, it generates very detailed data about what the user was doing, and then uses AI and Machine Learning techniques to see through all that data, filter out “noise”, find patterns, group actions together and to work out the tasks that the user is performing.  

For privacy purposes, you can define a list of applications that it ignores or monitors, and some products even allow you to obfuscate certain fields of information.

They are then able to generate a list of events that might look like this:

  • Open File Explorer
  • Navigate to “Orders” folder
  • Open “New Ordres.xlsx” file
  • Open “Salesforce”
  • Click “New Order”
  • Copy from Excel column A
  • Paste into 

What is it used for?

Task mining gives you a very detailed view of the steps users take to perform their work. It captures how long they take to do things, and how they deal with exceptions. As mentioned it also captures screenshots.

For this reason, it is very good for producing a detailed, step by step description of the task, or a Task Description document, complete with screenshots.

This document can be useful for the purposes of understanding the overall process, and so it can be used as input into the discovery of the end to end process that the task was a part of.  The document can also be useful for process documentation purposes, as well as training of users.

Importantly, this document can be used as input to an RPA tool for the purposes of automating that task. In fact, some RPA vendors are seeking to automate the automation by developing task mining capability.

What can’t it be used for?

Unless the process is small and performed by a single user, then Task Mining cannot give you a holistic view of an entire business process, end to end.

In most large organisations that have large complex processes, the tasks that make up those processes can span multiple user roles, systems, and geographical locations.  So Task Mining does not give a full process end to end view.  For this reason, it is (or should be) referred to as Task Mining rather than Process Mining.  It also doesn’t capture all the cases going through the system, unless you have 100% coverage of all users in the organisation with the tool, which is rarely the case.  

Furthermore, Task Mining only generates data once you install it and turn it on, so you can only analyse the data you have from that moment on.  

We hope you found this introduction to Task Mining useful. In a future blog we will discuss the differences between the two technologies more, and describe how they can best be used together.