The Challenges and Opportunities of RPA and Process Mining
We all know RPA automates Business Processes. But, not all Processes. Some Processes are broken and are not fit for RPA. These processes either do not have an SOP manual to refer to or are performed by multiple people or are executed without any regard to process rules for the sake of time and convenience. All this is probably even okay until you have something like RPA being introduced in the enterprise. An unstable and unstandardized process unequivocally leads to RPA failure!
The main failure causes for RPA from a process perspective are –
- Dynamic processes
- Changes in target UI
- Complex business exceptions
What are the Factors that work for RPA?
- Maturity and Stability of Business Processes
- Application/ System landscape
- Availability of third-party integrations, Plugins, and enhancements
- Employee empowerment
- Process governance and application governance
Automated Process Discovery and Process Mining
Process Discovery is an interesting feature in a fully Integrated Intelligent Automation Platform. Automated Process Discovery is the automated reconstruction of Business Processes by mapping and analyzing the current state of processes.
Process-mining is a subset of Automated Process Discovery that makes uses of Event data — (System Logs, Application Logs, and UI logs) and transaction data to map, analyze, and reconstruct business processes.
Event logs are generated by enterprise systems such as ERP, CRM, HCM, and SCM and even small applications. Process mining is a key enabler when it comes to RPA initiatives. Like any technology in its infant days, process mining has its pros and cons; its challenges and opportunities.
Process Mining Challenges:
- Availability of Event data and Log Data:
Process Mining is not effective without task mining. Task Mining is the use of log data (also called UI logs) of how human employees interact with their computer system when running a task of a business process to find and create automatized routines that can subsequently be executed using programmed bots. We aren’t delving into the challenges of task mining but not all applications write out UI logs. Even simple activities like opening a file, deleting a cell, etc. must be captured. Also, events happen within a specific process context. The process context most of the time that cannot be extracted from event logs are not captured through algorithms and will have to be coded as a pre-condition or business rules. (e.g. Formatting a cell value in MS Excel)
- Pre-processing of Log Data:
Automatically captured UI logs need to be filtered for Noise i.e. events unrelated to the process that is performed by the user. E.g. replying to an email or checking a site in the web browser for information. Redundant event data must be removed. E.g. Pressing keys multiple times that have the same effect. For each Input UI log, there will be a corresponding ‘Task Trace’ which represents an outcome of an executed event/action. ‘Task Traces’ not only validates the execution of a process step but also provides details on process step variations and limits. Segmenting logs into ‘Input events’ and ‘Task traces’ is intricate and requires the workings of refined algorithms.
- Creating Automatic Routines:
Task Traces are converted into a set of actions; i.e. they are converted into a specification that translates into an RPA workflow. Theoretically, just from the specification, it should be possible to automatically create a bot. We don’t see that happening in real life as there are always changes, exceptions, and data validations that require the services of a workflow designer/ studio. Business Rules and Process triggers will have to be explicitly coded into the workflow.
- There are factors outside the processes that impact process mining:
For example- the technology changes, regulatory changes, customer demands, etc. Process mining also seems to disregard human ingenuity to effect process changes for optimum business impact. No matter how much technology advances, Business Process Change will be administered by humans. Process mining can aid the process design, but never trigger the change management of a business process.
- Process mining might bypass the Business Analyst/Consultant and put more onus on the developer who is not a process expert. This is not a good situation and such inadequacies may crop up in the process automation effort.
- Process Mining is not a basic feature of the RPA tool set. It may be a stand-alone application or may be offered via integrations. It increases the cost of automation and is unlikely to be considered at the Piloting stage. Ironically, it is Process mining that aids process discovery and increases the chances of RPA success and not the other way around. The chances of organizations implementing Process Mining after the full adoption of RPA is low. Agreed, the technology, and its products are in some state of flux, yet Process Mining makes sense and is more value when adopted with RPA.
From the above-mentioned challenges, it is quite evident that it is critical to understand what steps of a process to record and at what granular level event and UI data need to be collected. Too many variations in process steps will automatically affect process discovery.
The Opportunities of Process Mining –
- If event data is available and the process is very stable, the end-to-end process automation specification is quite possible.
- There is high process visibility. Process variations can be identified, normalized, and standardized to tasks that can be easily automated using RPA. Redundant process-steps can be weeded out.
- Process errors can be identified, and the root causes can be analyzed.
- Process mining aids Business Process re-engineering efforts. Process paths generated by event logs of several process runs can be aggregated and viewed at a higher level for optimization opportunities.
- For complex processes or a high number of processes, using process analysis with process mining is cost-effective. It is also quantifiable and can be used for comparative analysis with the work of human employees on key parameters such as a) Accuracy, b) Efficiency, and c) Duration.
It can be used to monitor RPA performance based on process metrics and can be used to benchmark robot performance against other deployed robots.
- Process Mining accelerates RPA by discovering processes for the RPA pipeline and prioritizing it for automation more effectively and faster than business consultants. Broken processes are immediately disqualified or can be fixed for RPA with the help of Process Mining thereby saving a lot of undue effort and drain on resources.
- It improves RPA implementation and determines RPA-scalable processes.
- Process mining provides the basis for data-driven process governance and can also detect how process changes over time and suggest to some extent to how it must evolve.
Process mining provides a comprehensive end-to-end view of process execution steps that occurs across the various applications and systems of an enterprise. End-users can ‘Play’ the logs to assess the ‘AS-IS’ state and opt for an application-recommended optimum process path. With Process-mining, the ‘AS-IS’ state is captured with a proper time-context. It has evolved into a distinct discipline and is challenging the traditional notions of Business Process Management and Business Process Re-engineering.
Change is hard to achieve, and Process Mining is not just a technological change, but also a cultural one. It enables not just RPA, but even a larger undertaking of digital transformation by providing pivotal process intelligence. The utility and functionality of process mining are even more amplified with RPA technology.
The business world is on the cusp of a more automated future — a future of Hyperautomation and Process Mining technologies that will have a great deal to do with the success of the hyper-automated world.