<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=321162&amp;fmt=gif">

Best Practices

Maximize ROI of your RPA with Process Mining


Robotic Process Automation has proved it can deliver optimization of Backoffice processes that are labor intensive, recurring, and rule-based. But it is failing to deliver on the promises and expectations of ROI.

With RPA employees don’t have to waste limited working hours doing tedious activities without added value. Automation both frees up their time and also reduces errors to a minimum. RPA has also proved to be compliant with all IT requirements – it is technologically secure, stable and scalable. These benefits are clear and visible across thousands of projects implemented across various industries – banking, insurance, SSCs, back offices and others. However, the frustration of business customers is rising as their as expectations are not being met. 

Just three years ago RPA was mostly about POCs and exploring the technology and opportunities for its use within organizations. Now the understanding of the technology has grown, and the core question has become one off value – what are the actual gains of implementing Robotic Process Automation?

How Minit Process Mining helps with RPA →

Key problems with RPA

1. A belief that it’s easy to automate fast and at scale

After the initial year of testing RPA, the organization is ready to start an enterprise-wide, rapid deployment of robots. The problem here is not the speed or scale, but the fact that processes within the organizations are constantly changing.

Currently, people in RPA development teams are usually involved through the whole cycle up to maintenance of the robots. Contrary to expectations the last part of the process turns out to be much more difficult than anybody counted on and is usually overlooked when estimating ROI. Since the spending on RPA needs to be effective, teams often end up trying to implement new robots while maintaining the changing requirements of the currently running ones.

In other cases, new transformation initiative where multiple systems are put into a single platform takes place, and all the work done on RPA goes out the window. Whenever a new platform is implemented all processes must be automated anew.

2. Scalability

Regarding RPA scalability, the problem is not with the technology, but in choosing the right processes to automate.

At first glance, there is a ton of potential within processes, but upon closer inspection, there is high variability and many different inputs. So even though thematically correct these processes are difficult to automate.

3. Unstructured Data

Another big issue is that data is not structured, and some of the experts estimate this concerns about 80% of all the data used within organizations. Even with OCR and NLP now being firmly embedded in the solutions, the issues have not gone away.

Robotic Process Automation (RPA) with Minit Process Mining

4. There is no systemic way to identify potential automation candidates properly

Most RPA experts agree that this is more of an art than a science, as the predictions on ROI are mostly based on experience, lottery or crystal ball forecasts rather than numbers. Teams trained within the companies lack long-term experience and therefore estimates on new implementations aren’t accurate, to say the least.

Creating capacity and activity models is very subjective with many vague attributes going into the equation giving wildly different results.

Case in point: One of our partners has done a project on Order Processing. The assumption was that there is a 60% automation potential. The problem was that without knowing the data and the process in detail, the assessment of the ROI potential was based on the average duration of processing of an order which was 4 minutes. There was an assumption of saving 0.5 FTE. A very quick win with a straightforward implementation, right? Once RPA went into production, the wrong assumptions were suddenly apparent. Although the average processing time was 4 minutes, it was mostly due to straightforward orders that the robots automated easily, but also didn’t take much time initially, when human resources did them. The problem was that when estimating ROI, the team looked at the wrong metric. Only later did they find that most of the total time was taken up by the problematic orders that took up to 8 minutes to process. Therefore, the ROI expectations weren’t met. The 80:20 rule applies here - just 20% of the invoices took 80% of the total time.

Michal Rosik, CPO Minit

Companies that skip the analysis and optimization phase of process automation often feel that RPA is like a hammer and wherever you look you see a nail to hit. What we suggest is to take one step back to take two steps forward.

Base your analysis on empirical data – look at the variation of the process, on the inputs on the resources involved. Decide whether to automate at all.

Make sure you are not automating broken processes. The ideal case is to do both optimize and automate at the same time. What processes do you need to transform before they can be automated to meet the expectations?

More obstacles are waiting past the implementation

  • What are the robots doing?
  • How can you utilize them better with the limited licenses you have?
  • How does the robot impact the overall process and resources involved, are there any bottlenecks?
  • Perform conformance checking with the setup, today vs. a year ago. Why do robots stop and are idle?

Another oft-occurring problem is that usually in a process transformation project there used to be a controller and the resources were people. Suddenly here is RPA which is outside of the controllers, and there is a lack of ownership. The question of who is looking at the overall picture and monitoring the metrics of the automated process arises.

Easy use cases are gone, corporations are moving on to other innovative technologies, and RPA is not getting priority status anymore. Pushing the automation forward is now up to the internal CoE developer teams without the budgets that were available before.

Most significant success factor is not the technology or the processes but the change management. That is where initiatives mostly fail on the human level. Here Process Mining can help with the social chart and analyze bottlenecks caused by inefficient communication or relationships between stakeholders.

To learn more about the role of Process Mining and how Minit can support RPA implementation at your company, get in touch with our team. We’re excited about assisting our customers in succeeding with their RPA project.

Watch our session on this topic presented at the RPA & AI Live 2018 now available on-demand 

Richard Lipovsky

02. 07. 2018