The field of process mining has grown rapidly in recent years, and with the increasing adoption of digital technologies and data-driven business processes, it has become increasingly important to understand the tools and techniques available.
It is crucial to be able to differentiate between the various process mining tools and technologies and to choose the right one for your organisation’s needs based on its goals and use cases to maximise intended value. There are two major categories of mining: “process” mining and “task” mining.
Processes are end-to-end groupings of logically related work, such as order-to-cash (OTC). Most of these end-to-end processes can be divided into subprocesses that group this logically related work into functions. In an OTC, subprocesses may include sales or accounts receivable. Each of these processes contains several tasks.
Tasks are the groupings of work within a process. For example, activities in the accounts receivable subprocess may include recording sales, generating invoices and tracking customer balances.
Process mining
Process mining is a technique designed to discover, monitor and improve real processes (i.e., not assumed processes) by extracting readily available knowledge from the event logs of information systems. Process mining extracts process-related data from various digital sources such as enterprise resource planning (ERP), customer relationship management (CRM) and other operational databases that are readily available in today’s information systems.
With this information, process mining can be used to identify bottlenecks, inefficiencies and areas for improvement to enhance enterprise performance.
The tasks in Figure 1 (e.g., create sales order, schedule sales order) represent the “happy path” of the process, when there is no delay and no back-and-forth between tasks. However, this is generally not the case, and many business outcomes are impacted due to delays, assumptions and not adhering to the process.
Each task in the business process in Figure 1 has an associated event log. In the example, they are entries in an ERP system, a CRM system and a supply chain execution system. The task “notify customer” is manual with no event log attached to it; if this task shows a lot of variance in execution, it should be investigated.
After examining the outputs of the process mining algorithms, you can see what happened during the execution of the process to understand if there are areas for improvement. These could include discovering hidden aspects of the process, bottlenecks, conformance checking, or even finding opportunities to automate and enhance the process.
Task mining
Task mining is a method of extracting valuable insights from low-level event data recorded in user interface logs or obtained through computer vision. The data captured includes individual steps performed by a user, such as keystrokes, mouse clicks and data entries, and is interpreted by utilising natural language processing (NLP) and optical character recognition (OCR) techniques.
Task mining complements process mining by providing a deeper understanding of the activities and actions taken within a process. The starting point for any task mining initiative is a user interface log. Each event found in a log refers to a step that is related to a particular task.
The events belonging to a task are ordered and can be seen as one sequence or routine of a task. For instance, consider the scenario where a user is responsible for pulling in invoices from an external system and navigates through multiple applications and screens to accomplish the task.
The steps that the user takes are to first download invoices from the directory of incoming invoices. The details of the invoices are then uploaded into an Excel sheet. Then the user logs into the accounting software – let’s say it’s an SAP system that requires the user to navigate through three different screens before the invoice data can be added into SAP.
All of these steps are completed by an employee at his or her desktop. Task mining will analyse the data collected during the execution of this task, collect variants and try to discover hidden aspects of the task and opportunities to enhance the task.
Process mining or task mining?
The main differences between process mining and task mining are that process mining targets end-to-end business processes (such as procure-to-pay and quote-to-cash) or parts of these processes, where different resources work together to deliver the process outcome – a product, service or information. In this case, the most granular unit is a task. Process mining is mainly interested in the order in which individual tasks are executed or evaluated, usually called the control flow.
Task mining targets a task consisting of different actions or steps. The granularity is a mouse click, keystroke, an entry of data or a desktop operation, such as copy/paste. Task mining is not only interested in the control flow (i.e., the sequence of the steps), but also in the data flow.
When looking at the data requirements of process mining compared to task mining, process mining begins with an event log, which includes events representing activities in a process and associated with a specific process instance. The event log requires an ID of the case, a description of the activity and a timestamp to determine the event sequence. Additional data such as amount, location and client can be used for more comprehensive analysis.
From a data perspective, task mining requires step information and timestamps for sequencing, but it lacks a unique identifier such as process mining. To determine the control flow, task mining first segments the captured data to identify repetitive routines and then analyses the data introduced and its composition. It requires more comprehensive data than process mining, especially for process discovery or robotic process mining use cases.
Gartner regards process mining as a special case of data mining. Unlike data mining, process mining focuses on the process perspective. It includes the temporal aspect and looks at a single process execution as a sequence of activities that have been performed.
Process mining bridges the gap between data mining and the modelling, control and improvement of business processes. Going forward, Gartner expects process mining and analysis will evolve to process intelligence that combines development and runtime software tools to analyse, model, and monitor business processes.
Task mining combines process mining techniques and specific algorithms for case segmentation and data pattern recognition. NLP may be used to identify and extract key data, and computer vision and screen capturing may be utilised in robotic process automation (RPA) use cases to create or improve user interface logs.
This article is an excerpt from Gartner’s ‘Innovation insight: Process mining and task mining’ report by Marc Kerremans and David Sugden.
