Is Process Mining and RPA a match made in heaven?
What do Process Mining and Robotics Process Automation (RPA) have in common at first glance? Correct; both technologies are about processes. Process Mining analyzes processes and Robotics Process Automation (RPA) automates processes. At least, that is the promise of the vendors and the names of the software categories also suggest that. So you might think that combining these technologies is a match made in heaven. Thus, it is not surprising that various RPA vendors have invested a lot of money to add Process Mining functionalities to their automation platforms. What is surprising, however, is the feedback from many customers that in practice it is very hard to find automation potentials quickly and easily with Process Mining. But why not?
To answer that, one should first understand what Process Mining and RPA actually do. Process mining uses digital traces from transactional IT systems such as SAP to reconstruct and visualize actual processes. For example, when a purchase requisition is created, the document is stored in the database with a timestamp and Process Mining translates this information into the "Create Purchase Request" activity. The same is done with all other documents and time stamps that occur in the purchasing process. In the end, this results in such a (simplified) end-to-end process:
RPA, on the other hand, does not automate the entire end-to-end process, but only the execution of individual activities. For example, the "Create Purchase Request" activity may consist of the following steps:
领英推荐
And it is only when the data is saved that the timestamp is created, which is then translated by Process Mining into the corresponding activity. If Process Mining now reveals that this activity takes too long and represents a bottleneck in the process, this is important information. But because the tools used in the front end, such as email or PDF, leave no usable digital traces, there is no information other than the one timestamp of creation that could be used to identify the causes of the bottleneck and the potential for automation.
So the reason why Process Mining and RPA are not a match made in heaven is because the technologies operate at different levels of granularity. Process mining looks at the end-to-end process at the activity level, and RPA automates the execution of each activity by being able to perform the individual steps that would otherwise be performed by the user, provided they are recurring and rule-based.
To better understand what is happening on the front end, however, Process Mining could be followed by Task Mining, which looks over users' shoulders digitally, so to speak, and logs every single action like mouse clicks and keyboard strokes. The sequence of all actions can then be reconstructed from this data and, if necessary, used to create RPA bots.
But before reflexively trying to solve all problems with automation, it's worth asking whether automation can actually solve the problem. But that's a whole other question, as is what is the true added value of Process Mining in combination with automation and RPA. I will deal with these questions in my next article. So, please stay tuned...
Thank you so much for taking the time to follow my thoughts. I would be happy to hear your feedback and happy to read other opinions or experiences.
Estrategia, innovación y transformación | Dise?ador de negocios modernos | Excelencia operativa | Tenor a tiempo parcial
9 个月I wouldn't say it's the better search engine for automation potential, but it's definitely the most objective and automated source for it.?
Entusiasta y mentor en estrategia, procesos (BPM), minería de procesos (process mining) y tecnologías
1 年Hola, queda en evidencia el nivel de desconocimiento y/o desinformación que existe al respecto sobre la disciplina, herramientas de PM y automatización de tareas. De seguro más de un proveedor de uno u otro lado, ha hecho de las suyas con el marketing.
Owner, Konoc Integrated Services Limited
1 年Rudy Khan, great article especially with your end-to-end thought process. What is most important before one chooses what tool to use is correctly defining pain points, how it should be resolved and how the solution should be presented? This will give more guidance on what tools are required. Like you said sometimes automation may not be the solution. For me I identify automation with opportunities that takes away a lot of repetitive processes and on the other hand is used as a corrective action which could require manual intervention or be automatic. Savings must come from implementing automation. Automation comes in handy with RCA resolution to manage bottlenecks and increase conformance. Some of the tasks could be updating integrated systems as a result or a feedback system to desired audience publishing relevant information when required. For me after proper definintion of the issues and probable solutions the process is; Process Mining + Task Mining + Automation/RPA + AI. The above is only ideal for a continuous improvement system working with benchmarking and not assuming a perfect system. Lastly Automation works better with direct integration to the different tools required for day to day activities/tasks.
Great article Rudy and good points. Others have already summed up nicely that RPA is not always the answer, but let's face it, sometime is IS the answer. When that's the case, Process Intelligence (via the various tools out there) have some capabilities that everyone should understand: 1. Most can simulate changes in advance by modifying data within the PI tool. This is a great way of learning that 'fixing' a bottleneck via RPA/automation is just going to create a bottleneck somewhere else. 2. Most can and will do ongoing monitoring to ensure that any changes to the process are not only successful, but also don't deviate in the future so SLAs are met. Of course this one will help with any change to the process, not just one affected by automation. At Novelis our high level approach is Analysis > Optimization > Governance. If Process Intelligence is within Analysis and provides unbiased, data-driven information about how a process is being executed, then Optimization should be the *right* approach to increase efficiency. That can be training, process reengineering (including the introduction of AI solutions as a replacement for existing tasks, or various forms of intelligent automation, and yep, RPA.
Customer Value Explorer | Digital Strategist | Change Leader
1 年I find this discussion very valuable and worthy the deep dive! Thank you Rudy KUHN and Tim Hills for sharing your thoughts. Hope more practitioners will provide their thoughts on this!