What to do after the quick wins of RPA?

Updated: Sep 24

Robotic Process Automation (RPA) has been a great trigger for massive business process automation, and the latest reports (1) suggest that it will keep on growing to a 200 billion USD market - with a little adjustment to the description of it.

The benefits are expected to be massive - already using the current technologies. For example, McKinsey (2) has estimated that “~50% of current work activities are technically automatable by adapting currently demonstrated technologies”.

At this point, I see people coughing behind the screen that we’ve automated all the reasonable processes with back-office robots already. Based on my experience, this typically happens when the automation rate is around 10-15% and a couple of departments are covered. Compared to the potential of 50%, we are talking about a gap of 35-40%.

Scaling RPA is hard

These findings are in line with the Capgemini RPA Market Survey 2020 (3) that states that “the most important question for many is ‘How can we scale?’” and “there are many large corporations which are still in very early stages of their RPA journey”.

It does not matter if you are using Automation Anywhere, Blue Prism or UiPath. The big question is how to close that productivity gap.

The common feature/problem with the RPA back-office robots is that they require a decent amount of standardized transactions in a standard process. Finding these quick wins is hard: it takes time and effort. And even after finding some quick wins, RPA people typically hit the wall when the backlog items are around 0.1-0.25 FTE. In addition, there are lots of other ideas but those lack the business case.

Scaling RPA can be done

I see that there are three paths an RPA CoE (Center of Excellence) can do to go further:

  1. Discover expanding automation rate of current automation with Intelligent Automation

  2. Use automated automation potential analysis to uncover additional quick wins and shared tasks

  3. Introduce citizen developer programs to automate the long-tail of small tasks and activities

Options for scaling RPA

Since I’m a huge fan of automated automation potential analysis, I’ll focus on the pain points of the traditional methods when aiming to close the 35-40% automation gap. By traditional methods, I mean expanding current RPA with intelligent automation and launching citizen developer programs.

The common problem of the traditional methods is the lack of possibility to scale. Fine-tuning the current processes without understanding the real potential is only fine-tuning the problems we already know.

Just to be clear: naturally, there are benefits to intelligent automation. If you already know which new quick wins it will bring, then it is not a new quick win - it is already in your automation backlog. With automated automation potential analysis, you’ll get a view of the total business case of e.g. OCR (Optical Character Recognition) and all the processes and tasks where you could utilize it.

The long-tail of Robotic Process Automation

Similar problem is being faced also in long-tail automation with citizen developer programs. Long-tail automation is about automating lots of small manual activities and tasks that are typically:

  • Done only by a few people

  • Done by multiple people, but we don’t know who they are. For example similar transaction flows are done in a logistics team in Australia and in a finance and accounting team in India.

To get started, people need to first train themselves on their chosen tool, then understand what they could automate, acquire a license, implement, run and maintain it. The problem is that it takes quite some time and big scale implementation to gain real benefits - even with the leading citizen developer RPA tools.

Automated automation potential analysis

The other approach is to use automated automation potential analysis with the following steps:

  1. Identify the modular and manual activities that many people share - regardless of the process name or organizational silos

  2. Build a custom script or use a low-code tool to solve the problem. E.g. Microsoft Power Apps. Use OCR, chatbots, or ML algorithms where suitable.

  3. Deploy the script/automation and push it to all the people benefitting from it. If you’ve automated the analysis work, you have a long list of people benefitting from the same solution.

And always remember, that the best solution might not be the one you are using - or any technology at all.



(1) Siliconangle: https://siliconangle.com/2020/08/08/rpa-competitors-battle-bigger-prize-automation-everywhere/

(2) McKinsey: https://www.mckinsey.com/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages

(3) Capgemini: https://www.capgemini.com/fi-en/article/new-report-on-rpa-reveals-how-organizations-in-finland-are-riding-the-automation-adoption-curve/

Follow us for the latest insights:

©2020 by Workfellow.ai