What is business process discovery?

Business process discovery (BPD), or process discovery for short, is the collection of tools, technology and strategies used to identify, outline and analyze an organization’s existing business processes. An essential element of business process management (BPM) and process optimization, the goal of process discovery is to collect and transform process data into process models. These visual representations of end-to-end business processes can then be used to re-engineer and optimize these systems. 

The guiding light in this endeavor is process intelligence (PI), the data generated by informational systems and, in BPD, used to analyze business processes and workflows. 

Though manual process discovery is possible, it can be painstakingly slow and unscientific. To start, traditionally, BPD leaders would rely on interviews with the people involved in executing the process to gather a significant portion of their baseline data. As you can imagine, this approach runs a high risk for error as there’s no real way of telling whether or not a person’s view is objective or fully informed. Secondly, performance data must be pulled from each individual IT system manually. As different systems measure different things, pulling it all together into one cohesive story can be very challenging. 

Automated business process discovery (ABPD), as the name suggests, uses Artificial Intelligence (AI), computer vision and computational logic to automate the entire process discovery lifecycle. Automated discovery solutions, essentially pull data from all of the various applications involved in executing a business process from beginning to end. Using what’s called a discovery algorithm, it then analyzes the data to identify patterns as well as determine process definitions and models. Using this empirical data, ABPD then maps out the entirety of an organizations' current business processes and its major process variations. 

How ABPD Works

Data Collection 

Automated process discovery tools essentially “watch” business users perform normal activities automatically collecting data from user interactions across multiple systems such as enterprise solutions (ERP, CRM, business process management (BPM), ECM, etc.), web applications, personal productivity applications (Microsoft Excel, Outlook, etc.), event logs and databases. Using computer vision and machine intelligence, these solutions can pull millions of data points from a single employee desktop ranging from keystrokes to mouse selections to unstructured data points such as screenshots and documents. By drilling data down to an unprecedented level of granularity, ABPD tools shed light on previously unknown or “dark” processes as well as allow organizations to monitor process performance in real time. 

Existing Process Analysis

Advanced neural networks and AI algorithms link together various data and process attributes to develop a current, baseline business process model. Unsupervised machine learning (ML) algorithms categorize and structure untagged data, identifying meaningful patterns that can be sorted and tagged to classify employee actions with automation potential.

The goal of this process is to create, essentially, a virtual twin of existing processes. In other words, a digital representation of existing processes that maps out all the possible process variations and exceptions. In addition, the ABPD will also use ML or AI to score and rank processes based on their suitability for automation. This helps ensure that no good opportunity for automation goes unexamined. 

Process Simulation

In addition to illustrating how processes currently run, automated discovery tools can also simulate how processes could run under different conditions. Using these tools, analysts can re-engineer and test new processes as well as build and test automation workflows virtually on the “metamodel” first before piloting them in real life. This helps to mitigate the substantial financial and reputational risks associated with automation projects.

Increased Process Efficiency & Operational Performance

One of the major benefits of ABPD is process optimization. By painting a fully formed baseline view of existing, real-world process flows, automated process discovery helps with the identification of bottlenecks, redundant processes, process deviations and other problematic issues. This bird’s-eye view of the entire end-to-end roadmap helps organizations envision new process design workflows and automation opportunities across the enterprise. 

Historically speaking, organizations had to rely on consultants and experts to advise on how they should redesign processes. As mentioned before, automated process discovery solutions not only visualize existing process models, they allow organizations to a virtualize and validate new processes. With ABPD, business leaders are equipped with all of the insights and tools they need to re-engineer processes on their own with minimal risk. 

The granular, empirical process data ABPD produces not only helps guide process redesign and other digital transformation efforts, it also makes it easier for organizations to quantify them. By enabling complete visibility into cross-functional operations, it also enables increased accountability and helps breakdown silos. 

Process Discovery & RPA

Process discovery is a key component of Robotic Process Automation (RPA) success. RPA is software that aims to mimic human behavior and is used to perform simple, rules-based tasks like data entry, invoice reconciliation, scheduling, etc. Unlike artificial intelligence (AI), RPA can’t make decisions or solve problems. It simply performs repetitive tasks faster and more effectively than a human would. 

However, you can’t apply RPA to processes that are not optimized, harmonized and standardized. In fact, according to a recent survey by ABBYY, 38% of respondents said that process complexity was the #1 cause of RPA project failure. In addition, RPA bots require structured input data to function, something ABPD automatically enables. 

Furthermore, the insights gleaned from process discovery help business leaders identify which processes are the best candidates for RPA based on suitability (i.e. whether it’s rule based or requires human intervention) and ROI (i.e. how frequently the process is run, how time consuming it is, etc.). As achieving scalability is key to maximizing the value of RPA, ABPD tools help ensure that organization’s have a pipeline of automation-ready processes as they continue through their long-term digital transformation journeys. 

Traditionally speaking, organizations have used separate process discovery and RPA implementation solutions. In 2020, Automation Anywhere became the first solution provider to integrate these capabilities into one single platform. As described by SiliconAngle, Automation Anywhere’s Automation Anywhere Discovery Bot captures and analyzes employees’ actions to identify the most common and repetitive process steps as they navigate among the various business applications they use. It then prioritizes automation opportunities it finds according to the potential return on investment they offer, before creating new RPA bots to automate those tasks.”

In other words, it automates the entire process discovery to automation lifecycle. 

*Automation Anywhere Discovery Bot Demo

Process Mining vs. Automated Discovery

Though they are sometimes used interchangeably, process mining and automated process discovery are two different disciplines. According to CIO Magazine, Process mining is an analytical discipline or methodology “by which organizations collect data from existing systems to objectively visualize how business processes operate and how they can be improved.” Process mining tools work by extracting knowledge from event logs created from ongoing processes and identifies anomalies, trends, patterns, and even obstacles or challenges. Insights generated from these “mining” activities can help enterprises enhance productivity, profits, and customer satisfaction. While process discovery looks at how people perform tasks and interact with technology, process mining is mainly concerned with how systems applications work and how they can be optimized.

Automated process discovery, on the other hand, are AI-powered powered tools designed to track user behavior and use that data to create a digital representation of enterprise processes from end-to-end. The goal of ABPDis to show how processes within an organization are executed and provide organizations the ability to analyze processes as they are. Unlike process mining, process discovery does not require any logs, databases, API access or systems integration. 

In other words process mining is thought to be more suitable for general business process optimization initiatives. However, according to TDWI, “If automating your processes or undertaking RPA initiatives is your goal, process discovery can help you know which processes  have hidden tribal knowledge and can be optimized for digital transformation.”

Image sourced from “Process Mining versus Process Discovery,” https://www.skan.ai/process-mining-insights/process-mining-versus-process-discovery

Please take 2 minutes to tell us about your automated discovery journey by participating in the survey below:

Create your own user feedback survey

Can't view the above survey? Access it here: https://www.surveymonkey.com/r/XTZXZKY