29 September 2020

Three questions to answer before RPA deployment

If you have ever been interested in Robotic Process Automation (RPA), you are most likely familiar with its benefits, which include increasing productivity and efficiency as well as reducing cost.

A software robot never sleeps, which saves a lot of time and speeds up processes; it also relieves people of time-consuming and routine tasks. This results in improved services for customers and the general public as well as higher employee satisfaction.

But even RPA technology cannot do everything. For it to be successfully deployed, some necessary prerequisites must be met. Here are the three basic things to keep in mind before implementing RPA.

1. Nothing is done without input data

RPA needs digital data; it cannot handle paper documents. If you want the software to automate things, such as the processing of written customer complaints or invoices arriving by traditional mail, these documents need to be digitised first.

Ideally, input data should be “structured”, i.e., organised in a predefined way. The data available in companies and organisations (such as emails, audio recordings, videos, and images as well as digitised handwritten forms) is often unstructured, which may seem to be a major obstacle to the effective implementation of RPA.

However, having data in a form inappropriate for robotic processing is not an insurmountable problem. Today, paper documents can be quite reliably “read” using Optical Character Recognition technology. Similarly, there are solutions for transcribing audio recordings from contact centres and turning them into text files.

2. A software robot has no brain

Robotic software can process anything it was programmed to do, but it is not intelligent enough or able to learn on its own. So you cannot expect it to answer questions that have no clear answer to them. Such questions require careful consideration by humans and cannot be outsourced to robots.

For example, when approving loans or any other type of application where an applicant may meet some yet not all criteria, an automatic rejection could possibly mean the loss of business or an unhappy customer. This is why the automated application approval process should be complemented in some cases with the expert opinion of an experienced person which may be partly based on feelings.

However, the need for human judgement and decision making at some stages of the process does not automatically mean that the activity is not suitable for automation. It only means that the automated process conducted by a robot may sometimes require the involvement of a human. This can be done, for example, by a window popping up on the screen of the person in charge whenever the robot is unable or unauthorised to do something on its own.

The lack of RPA intelligence can also be dealt with by incorporating artificial intelligence and machine learning elements. In such a combination, the software robot is able to adapt and learn to make decisions based on the broader context as well as past experience.

3. Necessary integration

For RPA to be effective, it may sometimes be necessary to process large amounts of data in a variety of formats from multiple sources. For example, accounting departments tend to receive invoices from multiple vendors in a variety of formats, with individual data being located in different parts of the document. For the software to be able to read the data, it is necessary to either unify the format or adapt the robot to be able to read the data from any required form and shape.

As with the previous points, the diversity of formats and data sources is not an insurmountable obstacle. It is merely a complication to consider when determining the total labour intensity and making a costs-benefit analysis when deciding which processes are cost effective for automation and where the return on the investment may no longer be adequate.

When considering limitations and possible obstacles to the implementation of RPA, it is worth remembering that a software robot will not fix an essentially faulty or inefficient process. This is the reason why we at Soitron analyse the affected processes and, if necessary, propose how they can be improved using business process management tools before the automation.

To sum it up, when considering RPA, it is essential to find a partner who knows the prerequisites for the successful implementation of automation very well and who provides a wide range of services (including integration and process management) that may be critical for the effective use of RPA.