RPA: back from the dead?

Posted on:

In part one of this look at robotic process automation (RPA), we analysed the over-hyping of the technology; asked why RPA projects have been going wrong; and how that’s hurting a mature analysis of the benefits of automation.

And while the death of RPA might have been somewhat exaggerated, problems must be addressed if those benefits are to be realised. We know RPA on its own hasn’t always delivered the value people were sold. Told it was a silver bullet to radicalise processes and agility, it’s not worked out as a panacea at all.

Selling bots into business users in different departments works for vendors at first, and perhaps even for the local business users. But that flawed approach prevents RPA from becoming an enterprise solution. We know it’s hard to stand up RPA projects in complex organisations unless there’s genuine enterprise IT and leadership buy-in (which has been somewhat rare to date).

In that first post, we used some well-known (mis)quotes to frame the problems that have beset RPA. So, here’s one more to introduce a slightly more optimistic take on RPA’s “demise”: 

“Robots can run all the processes some of the time, and some of the processes all the time, but cannot run all the processes all the time.”  - Abraham Lincoln

Where RPA has delivered for clients, there’s a shared understanding that software installation is just part of the picture. The number of bots or the hours saved isn’t the key; the messaging needs to be more nuanced.

The reality is that RPA can deliver a whole host of benefits. Properly implemented, it enables clients to scale processes quickly and deliver new capabilities. That means processes and people must be aligned in ways that allow for the RPA model to actually deliver. Just like ERP, this is a long-term investment – and the ROI comes incrementally as you roll out projects strategically against a much bigger organisational programme. It’s not magic that instantly revolutionizes every process it touches.

The best RPA projects have certain hallmarks.

First, they successfully triangulate the technical, commercial and strategic imperatives of the organisation. That means they’re a key component of a larger transformation programme. They’re properly targeted and justified within a holistic business case that can be delivered over a reasonable time-frame.

Second, they slot into processes that already work. In US healthcare, for example, overhauling claims processes could be a huge automation win. But with so many different variations across locations and departments, getting those right is a huge undertaking even within a single business. It’s a six-plus month job to get the kinks ironed out, not six weeks. That means any ROI on RPA to automate those processes drops from ‘mythic’ levels to just ‘impressive’ – but it’s still valuable, and the processes become scalable.

Third, RPA is treated as part of the core technology infrastructure – not just a departmental bolt-on. RPA needs to interface with ERP; the IT function needs to understand and manage its dependencies – and do training for users and software maintenance.

Even if you get all those elements right, an RPA solution will have limitations – it can’t do all the processes all the time – so successful implementations are also designed with exceptions and fallbacks in mind.


Agents of change

Responsibility for bringing automation back from the dead falls on four groups: vendors, business users, enterprise IT functions and enterprise leadership. So, what do they need to be doing to ensure that automation rises up from the dead – not as a brainless zombie, but as a living breathing entity?

Vendors: beyond the sell

Some big tech vendors have dragged their feet because RPA disrupts their existing licence revenues. They like game-changing software; the low-hanging fruit for business users has often been picked, and genuine automation looks hard for them now, especially given their revenues from other services.

In one sense that’s good news: where automation technology is appropriate and can be wrapped into their wider offering, they are more likely to bring it in as part of a coherent solution. And they’re less likely to peddle half-formed ‘zombie RPA’.

The better news is that there are increasing signs that specialist vendors have moved away from pure RPA. There’s a growing awareness that they need to be able to articulate the end state for the business – and the financial implications, rather than simply trying to sell a discrete tool or a licence for a given number of bots.

Business users: strategy, no tactics

We’re seeing a lot of clients trying to move into a more enterprise-centric approach from a democratised model for their tech. (That’s been particularly important during lockdown, where uniform security, connectivity and scalability underpin working from home.)

But that creates friction. Some business users have been quite happy with isolated bots. When IT comes in to impose control and structure (and finance is asking for ROI data), those users get frustrated. The stuff that ‘just works’ for them is suddenly held to new standards and what seems to be needless scrutiny.

So instead of thinking about isolated processes and possible bot solutions, these business units should think about the customer – and mapping out how value is realised. That defines an automation logic that flows more easily into operations and redesigned processes. Then they can start to articulate how automation enhances them strategically and provides value well beyond simply replacing a human worker’s actions with a bot.

Enterprise IT: keeping control

The other approach is to apply automation to one area, then see how far that process could be standardised throughout the organisation, using feedback from the business to guide the project. But you need the enterprise IT people to take ownership of identifying and solving the common challenges that are revealed – such as more rigorous pre-processing of data.

That enterprise standardisation is a key to securing value – it’s still all about scale – and it also means IT can enhance strategic decisions such as BPR and acquisitions. If you have connected, standardised, intelligent automated systems, when you need to tweak, you can tweak everywhere at once.

The enterprise IT folks also need to be able to interpret small-scale trials for automation. If you roll out an RPA project that’s very limited or disconnected, any feasibility study must be properly calibrated to give a fair idea of what a fully scaled-up programme is likely to deliver.

The template – one that’s been proven with some of our biggest clients ­– is to have a testbed that’s capable of standing up the strategic and technological assumptions, deliver a proof of concept… and then industrialise. The early stages are designed specifically to identify issues that might prevent value being secured in the later stages – not hitting massive ROI out of the park in their own right.

Leadership: pride and prejudice

Many business decision-makers see a value in retaining human headcount for a variety of reasons – not least their remuneration. But the benefits of process standardisation and automation are starting to be appreciated. ‘Designing for data’ in ways that meet clear business needs ought to be a compelling pitch for all but the most stubborn leaders.

To get on the exec radar, high value ROI on a small investment isn’t good enough, however. If your investment is $5m in a broader programme, an ROI of 50% is a big impact – it doesn’t need to be 300%. But leaders must remember: most programmes are a complex blend of winners and losers, of predictable progression and unexpected gains.

Leadership endorsement means a lot. It creates budget. It delivers enterprise alignment. And if leadership has seen a compelling use-case in, say, a call centre, they can be convinced about the benefits of a full-scale, well-supported automation roll-out. (The truth is: as soon as something is clearly successful, senior execs wants to own it.)

We think that creates an opportunity, one where adopting an Intelligent Automation approach – not just the piecemeal, solve-an-annoyance RPA pitch we’ve often seen – could give a big advantage to businesses prepared to learn the lessons from the beginning phase of its popularity.


5 minute read