2018-09 The role of RPA in digital transformation: De-risking

The role of RPA should be:

  • Enable your company to quickly reduce the effort required to perform business processes, and
  • Do so in a way that makes future transformation work easier by standardising data, engaging staff, and identifying all of the gotchas in a manual process.

Digital transformation is hard because moving from manual processes to automated processes is hard. But the step from one automated process to a different automated process is actually pretty easy.

The diagrams below are the analogy I use to explain the role of RPA.

Digital transformation for a company is a bit like transforming a village into a city. Replacing a dirt track with a freeway is a big change. But replacing a dirt track with a paved road is palatable for the villagers; and then replacing that paved road with a freeway is pretty easy.

Village to city

In this sense, RPA can be used to de-risk digital transformation. Once you have automated a particular task using RPA, you've fixed most of the data problems and addressed much of the change management concerns of the end users.

Village to town to city

RPA helps digital transformation programs for psychological reasons as much as for technical ones. RPA projects are iterative and heavily involve end users in the process as the process is being automated. This means that the end users are involved in the discussions about why standardised data is needed at this part of the process or the discussion about why these two processes could be done as a single process if they made this adjustment in their work practices. As these discussions take place, end users learn the language of automation and, when the change from the paved road to the freeway takes place, are more willing and informed participants.

2018-07 What processes are suitable for automation

In early 2018, IBM released a report on the Evolution of process automation. This report provides some concrete examples of processes suitable for automation; with a particular focus on processes where machine learning can be put to good use.

It is worthwhile reading the whole report but some of the highlights are shown below:

What does an automated process look like.

Insurance claims management is an ideal process for automation. There are lots of transactions and lots of decisions that need to be made along the way. Automation software is used to move the claim through the process, machine learning is used to assist with some of the decision-making.

The process below shows a typical claims process. The first part of the process is performed by a Claims Rep. The second part of the process is performed by an Adjuster.

When automating a process, the one question you need to ask yourself at each step is Do I have the information I need in order to decide what to do next?

Manual insurance claims process

The first part of this process is more suitable for automation than the second part of the process because the data required to make a decision is known at each step.

  1. At Intake, you know who an acknowledgement needs to be sent to,
  2. At Acknowledge, you know what systems you need to query to check coverage,
  3. At Identify, you know how to identify information that is missing and you know who to contact, and
  4. At Contact you know how to contact the client.

This process can be almost entirely automated with very few exceptions needing to go to a claims rep for a decision.

Automated insurance claims process

The second part of the process is less straightforward.

  1. The Investigate step can go in a number of different directions, and
  2. The Determine step can involve a number of considerations that may not be exhaustively documented.

But, just because it is not straightforward, it does not mean it can't be automated. Perhaps you can't fully automate it, but you may be able to automate some of it. So, as a second phase of your automation initiative, you can assess whether you can reduce the number of claims that require the involvement of an Adjuster.

There are two perspectives you should look at this from:

  1. Can you automate some of the activities for all of the claims? For example, instead of automating the entire Investigate step, perhaps you can automate some of the common data gathering work involved in the step, or
  2. Can you automate all of the activities for some of the claims? For example, is there a subset of claims that are simple enough to be fully automated?

In the figure below, we show a fully automated process as light blue. If some of the activities can be automated for all of the claims, we show it as light blue on the left and dark blue on the right. If all of the activities can be automated for some of the claims, we show it as light blue at the top and dark blue at the bottom.

You can see in the figure below that that the Close process was fully automated for some of the claims but that only parts of the Investigate and Determine processes were able to be automated.

Automated insurance claims process - phase 02

What types of processes can be automated?

Now that you have seen an example of a process that can be automated and how to think about the automation of the that process, let's look at the types of processes that are suitable for automation.

The Evolution of process automation report interviewed over 3,000 executives to understand the types of processes they were considering automating.

The common traits across these processes are that:

  1. they are high-volume, and
  2. involve processes where the data required to make a decision at each step is generally known.

As always, at the top of the list are accounts payable and payroll and components of Accounts receivable activities are interspersed through several of the other items in the list. Lower down are items such as customer support, marketing operations and logistics and supply chain.

Processes to automate

Pulling together your list of processes that can be automated

To create a list of processes that can automated within your company, we recommend not attempting to put together an exhaustive list before you take any steps toward automation.

Instead, put together a list of potential target processes and an estimate of the benefits that can be achieved by automating these processes, and then start automating.

As you work through your first automation targets, you'll identify more targets and be able to refine your target benefits and timeframes based on your actual experience. This will allow you to

  1. deliver benefits right away, and
  2. develop realistic objectives for your automation program.

You can contact us if you want to discuss your company's particular situation. We're always happy to have a chat.

Last Updated: 9/3/2018, 11:21:14 PM