Robotic process automation: lessons learned from the early adopters

Robotic process automation is set to see massive growth in the coming years, so what can newcomers learn from the companies that have been experimenting with RPA for nearly 10 years?

With annual growth of 57%, the market for robotic process automation software is set to reach $680 million in 2018, according to Gartner, and hit $2.4 billion in 2022. So, as new projects gather pace, what can they learn from early adopters, asks Lindsay Clark.

Robotic process automation is a tough sell. Although software robots promise to help with repetitive administrative tasks, the people already doing the work feel threatened, according to Cristian Paun, head of business transformation and procurement at Lombard International Assurance. Meanwhile, managers at the top of the business have other concerns.

"They want to know about return on investment and how fast they will see their money. We had a lot of discussion around that," he told a recent London conference hosted by RPA software provider UiPath.

He was able to help business leaders understand the strategic benefits of RPA by discussing the stark economic reality many organisations face.

"I don't know of any companies that project negative growth. And for next year, most businesses have regulatory and other strategic projects. But they don't want to hire all the people they might need. If you don't want to hire people or give up projects, then you need something more strategic. You need something else, and for us that something was RPA. It was a bit of journey for us to realise it was not a nice-to-have, but more like something we required to survive," Paun said.

While Gartner has pegged the market for RPA tools at around $680 million in 2018, other analysts think it will be higher. Research group Horse for Sources says it could reach $1.7 billion this year, an increase of 17% on earlier forecasts. It attributes the lift in part to the shift in the focus of RPA from job elimination to augmenting talent, digitising processes and extending the life of legacy IT systems. Many operations executives are keen to fast-track RPA training programmes and invest in broader intelligent automation strategies, even though most enterprises are still in the "tinkering phase" of RPA development, the research firm says.

Insurance firm Direct Line Group started working with RPA in 2015 and in 2018 reached capacity for 250 concurrent robots to run in its estate and can automate about 70,000 transactions a week, according to Christian David, business services director.

He told the UiPath conference: "We started out tactical: It is quite important to learn about the organisation and how it is going to accept and adapt to automation. But you can only go on like that for so long. You have got to take those learnings and use them as a blueprint for your future operation."

One lesson they learned was that they were trying to automate too many processes, but not going into enough detail on each one, he said. "If you only take part of the process, that's ok, but you are effectively breaking the process. We got to understanding that you need to take an end-to-end view. We are doing fewer processes, but we have been far more impactful."

The early implementations of RPA also suffered from over inflated expectations, says James Hall, Accenture managing director of automation engineering services and operations. "People had far too great expectations of what they could achieve. Executives tended to think half the workforce could be bots in 18 months. That is unreasonable."

In high transaction situations, working on legacy systems, automation might save around 20-30% in efficiency. But working on modern applications, bots are only likely to create a saving of 10% or less, he says.

The more advanced organisations in RPA are not only looking to automate existing processes, but trying to combine RPA with management of workflow, process re-engineering and machine learning so that bots can exercise some ‘judgement' based on earlier decisions and bring more value to the business, Hall says.  

But those companies just starting out may ask if they can catch up. As investment in RPA continues to climb, businesses face the challenge of grappling with technology, business processes and people expectations. Those learning lessons from early adopters will be best positioned to succeed.

Admiral Group: turning RPA all the way up to 11

During UiPath's recent London conference, Neil Davies, automation manager with insurer Admiral Group gave a light-hearted review of the what his team had learned about RPA.

"We first got robots in 2016. We were a little bit smug and thought we didn't need anyone to help us," he said.

Here are the highlights of his lessons learned:

1.         People are not scared of robots, but they are scared of the robot team. This matters because the robot team sit very close to the people who the robots are impacting. Everybody has views about robots, but they mostly care about what senior managers say.

2.         Size does not matter. A manager said a competitor had automated 35 processes. But you have to think about what a process is. Are you comparing apples with oranges? The number of processes doesn't matter.

3.         Just because you use a robot to do that stupid thing, you are still doing that stupid thing. People come to us with some crazy processes they want us to automate. We say no. They're still doing the crazy processes, but we're not. 

4.         Give me smart people and I'll give you smart robots. We've recruited completely internally; we have no external consult. We've trained people from our sales team, customer service team and new business team.

5.         You look at things in far greater detail than you have done before. A lot of people are talking about process mining and process discovery. You need them before you build robots. Business analysts are great, but you need technology to give you the detail you need to deploy robots fast.

6.         Slice and dice your processes. Go live in steps. We have made mistakes. We've taken a couple of processes together and gone too big and the developer struggled. Break down processes where you can.

7.         Exceptions are a robot's best friend. If you hit an exception, it's great. The business says they will handle it. But when we automate 80% of the work and they struggle with the 20% because we are going so quickly, then they ask us if we can automate the exceptions too.

8.         Not all processes can be automated. You have to admit when you are not ready.

9.         You will condense aspects of person's role. People might be used to spending an hour a day not thinking, just typing. If you take that away, do you expect them to do the intense, problem solving part of the job all day?

10.       You can't stop experimenting. There will no fixed end point with automation. We won't run out of things to automate.

11.       Automation is exiting and full of surprises. Governance, terms of reference, oversight and audits are all important, but in automation, if want it to keep working, you've got to keep the bit that lets you make stuff up as you go along. If you stop doing that, then you just become another IT department.