Robotic Process Automation: The good, the bad, and the how to make it better.
RPA stands for Robotic Process Automation. It’s an emerging technology designed to automate the repetitive, high-volume-but-low-value tasks employees are doing on their desktops — by getting software "robots" to do them instead.
By automating tedious admin like form-filling, copy-pasting and systems-integrating, RPA theoretically frees up people to concentrate on more high-value work — or enables businesses to cut headcount.
As companies everywhere wrangle with how to make their processes better, RPA has been hailed as a potential solution.
But while it’s a useful tool to help integrate disparate systems and automate specific manual tasks, its end-goal is automation for its own sake, rather than to optimize for business outcomes.
Used in isolation, it’s more likely to hamper your processes than improve them. In fact, 30-50% of RPA initiatives fail (or perhaps fall foul of some of the bolder marketing claims made about the technology.) Results tend towards ephemeral improvements rather than long-term ROI — and while the promise of quick wins is tempting, it hasn’t proved game-changing for most. "It is clear that the first act in the 'robotics evolution’ has not been a slam dunk for many," note Alex Edlich and Vik Sohoni, both senior partners at McKinsey.
To achieve lasting impact — and deliver real value — RPA is best used as part of a greater set of tools in a holistic process improvement plan.
Here’s why — and how.
Part of the reason RPA is an attractive proposition is that it’s easy to get started — initial implementation is both cheap and fast (adoption slightly less so).
It is a good way to streamline user interaction tasks — automating the kind of tedious human activities nobody likes doing, reducing staffing costs and the possibility of human error.
It also lowers the barrier for system integration, making it easier and cheaper for people without technical knowledge to integrate data from different systems — especially legacy systems without designated APIs.
But while these benefits are undeniably useful, they’re very much undermined by RPA’s drawbacks when used in isolation.
“Automation applied to an efficient operation will magnify the efficiency. Automation applied to an inefficient operation will magnify the inefficiency.”
— Bill Gates
Adoption tends to be much slower — and costlier — than the initial implementation promises. According to Deloitte, 63% of RPA projects take longer than expected to implement, and 37% of cost more.
Scaling is more difficult than expected, too, with only 3% of organizations actually managing it.
Contrary to popular belief, you also need to supervise your bots — you can’t just let them run riot on your processes. They require upkeep — think maintenance and regular updates, an often unforeseen additional cost. (You don’t want a security breach because you didn’t keep an eye on your bots, do you?)
More importantly, bots often lack the flexibility to change alongside the systems they’re interacting with. And these are systems that change all the time — an updated interface, the latest security upgrade, or a new addition to the tech stack, and your bots are done.
"At Siemens, automation rule number one is to take a process-first approach."
— Gia-Thi Nguyen,
Head of Operational Excellence, Siemens
Part of the issue also lies in identifying the right candidates for automation. Not all high-volume, lower-value tasks should be automated, and often these may have an unexpected impact on other areas of the business.
Take customer service tickets for example. You can get bots to fill them in faster than a human can — but these may then also build-up faster than your service team can cope with. Cue delays, dissatisfaction and a drastic drop in NPS scores.
Automating the wrong process — or worse, a process that’s already broken — creates more work in the long run, where maybe removing a step in the process, or tweaking some of the rules regulating it would have been a better way to drive efficiency.
Some areas of the business actually benefit from having people involved — Accounts Payable will always have invoices that are more likely to need human review because they concern certain vendors or material categories, for example. Automating their approval would just cause roadblocks further down the line.
Ultimately, RPA optimizes for efficiency in isolation.
That’s fine if you’re resolving localized pain points — simple use cases encompassed in rule-based logic.
But for any business interested in truly addressing inefficiency end-to-end and optimizing for real business outcomes, RPA’s main value is as only one of a number of point-solutions in a much bigger process improvement toolkit.
“RPA is not a best-practice business process platform… RPA should be regarded as one tool in the broader process automation toolbox.”
That’s why rather than thinking about RPA as a first step, the most successful companies start by investing in tools that help them understand their processes, and by setting goals for process improvement — like increasing productivity, reducing rework costs, or improving on-time delivery.
Then they deploy RPA as one of the tools to help them get there.
In other words, RPA solves small problems. Process Mining and Enterprise Performance Acceleration software (EPA) actually achieve business outcomes.
Process improvement is about accelerating and achieving business outcomes — not just by increasing productivity and reducing costs, but also driving exponential growth. It’s about identifying sources of operational friction in your processes and eliminating them — only occasionally by leveraging automation.
Where RPA can match POs to invoices and goods receipts and execute on payments, for example, Process Mining can:
Consolidate ERPs by automatically comparing and standardizing PO processes
Improve cash flow and revenue by reducing order rejections from backorders and rescheduling
Lower order processing costs by identifying and eliminating unnecessary order blocks, credit-related or otherwise
Increase working capital by optimizing invoice payment dates
It’s a different ballpark altogether.
There are seven possible steps to achieving world-class processes.
First, visibility. Processes in practice rarely look anything like processes in theory. True process excellence can only be achieved if you have a real, live picture of what your processes actually look like over time.
Process Mining gathers both business data — time-stamped, event-log data from your every system — and user interaction data — the steps your people take to complete their tasks — to reveal what your processes actually look like in practice.
With a live picture over time, you can easily spot both deviations and inefficiencies and tackle them — in whatever way is best.
Sometimes it’s just a question of fixing a broken process, and training your employees to do things differently.
Remove or automate unnecessary steps in the process to increase efficiency, either in the way systems work or the way your people complete their tasks. The latter is where RPA can come in.
Leverage AI to find the best possible path for your process and optimize, optimize, optimize.
Leverage AI to find the best possible path for all your processes as part of a holistic whole, focusing on how these processes work cross-functionally across systems, people and departments.
Measure the impact of your enhancements over time and tweak accordingly and automatically. Your processes become effectively self-improving.
True process optimization is like finding the bit of string that pulls the whole knot apart. Except the string is actually an elastic, and catapults your business into a state of continuous process acceleration.
Because removing friction from your processes and turning it into flow has a knock-on effect on the rest of the business.
We call that becoming a Superfluid Enterprise, and it’s the key to delivering the world-class experiences that your customers — and your employees — expect from you.
RPA by itself does little to advance anybody’s business goals.
But put RPA in the service of Process Mining software, and it can help streamline user interaction tasks — as one of the possible steps an enterprise can take on the way to world-class processes.
Process Mining provides the context and planning and orchestration needed to make sure that RPA benefits rather than hinders your business. Deploy RPA as part of a more thoughtful process excellence initiative, and you’ll reap the benefits.
Download the RPA Data Sheet to learn more about Celonis and Robotic Process Automation.