How Process Mining Eliminates the Automation of Faulty Processes

How to identify processes for RPA

Digital transformation is taking place in businesses all around the world. And one of its major aspects lie in RPA or Robotic Process Automation.

RPA provides you with the means for automating simple processes in the daily life of your company. Thanks to so-called “bots”, you can save your employees' time which can be spent on more important, higher priority activities.

Bots are software applications you can deploy into your infrastructure and run them for the tasks that were previously human-driven. A good deal of vendors already provide user-friendly solutions for businesses to deploy such bots, based on your needs.

Before you dive into RPA, be sure to run a thorough analysis

Here comes one serious poke on the shoulder, though. At first sight, it might seem to be a great idea to deploy RPA on all the processes in your company end-to-end.

However a reality check is needed. Not all processes can be automated. Before jumping the gun, you should see which processes are ready for automation, which can be improved, or – and this happens often - you might come to the conclusion that some of the processes are not even suitable for automation.

Therefore you should very carefully think about that before starting the RPA initiative.

How do you know which ones to automate? Even though it is possible to run an RPA project without having a clear picture of the processes, it's usually better to understand them first – where the bottlenecks, deviations, unusual transactions, reworks are present, where human decision-making is needed, and so forth.

With process mining, you'll get a precise data-backed picture of your processes as they are running end-to-end. Instead of using old-fashioned and imprecise methods of employee interviews, the latest technology of process mining solutions will come in handy.

Now, what parameters should you be looking at when considering RPA?

Deciding where to deploy RPA

We've put together a list of questions you should ask before launching an RPA initiative in your company. If at any stage the answer is NO, either consider improving the process before deploying RPA or scratch it out completely as a no-go for automation.

Does it make sense to automate this process?

Think twice whether the process should be automated. And don't guesstimate, always calculate. After getting the results from process mining, you might notice a potential problem somewhere along the way.

For instance, you'll speed up the process of filing invoices into the system. But think about the next step of approval. Once the filing time reduces to tenfold, the authorizing person may get overwhelmed with the pile of invoices and become a new bottleneck. Think about these potential pitfalls in all your RPA candidates.

Is the process standardized?

It's beneficial if a process is standardized, meaning it always follows the same path and steps. Standardization of a process leads to higher efficiency and consistent results. If your automation candidate is not standardized, either don't automate or adjust it. This also applies to cases when the potentially automated process has a vague start or end point.

Does the process occur in high volume?

If you had only twenty invoices per month to file into the system, would it make sense to involve robots? Surely not. The costs of deploying a bot would highly exceed the effects. Even a human worker would do the work much faster while the robot is being deployed. That's why RPA provides the best ROI for processes that occur in your organization in high volume.

Does the process run without any human decision-making?

Bots can do just so much in terms of decisions. As a software application, they are programmed to make simple “yes/no” decisions and follow an “if-then” path. For example, a bot can decide whether your customer has an issue with hardware or software, based on the filled-in field in a support ticket they submitted. Yet, the bot won't be able to solve the customer's problem. Still, the truth is that RPA bots are becoming more and more intelligent, and in the future will be able to make better decisions.

Run a simulation before going live and monitor the bots

Before launching your bot into the production, you might want to test it and check whether it takes all the paths it has been programmed to. For such cases, a simulation run is a way to go.

Once deployed into the system, you should think about monitoring your bot. Why? The process might have slightly changed over time and you’ll need to know. What if a new format of the invoice has been accepted in the process? Or, a new hardware type was added to the CRM ticketing system?

There's nothing worse than investing time, effort and money into an RPA initiative which fails from the beginning. In addition, as robots perform their tasks much faster and on a larger scale, the demand on your IT infrastructure is significant.

Automating a poorly functioning process is not only inefficient, but it might also result in slower performance of your systems or employees.

That's why you should always analyze your processes with process mining tools. Such a process audit will provide you with the right candidates for automation. It will also tell you which ones to optimize first and get them ready for your RPA initiative. In addition, Minit process mining can identify which areas to automate.

Do you know how transparent the processes in your company are, and which of them are automation-ready? If not, approach your operations manager to find out.  

Still not sure how to choose the right processes to automate? Let us help!

Richard Lipovsky

11. 06. 2019