RPA: “It’s not dead; it’s just resting...”

Posted on:

 We’re living through times made for aphorisms. When the going gets tough, the tough reach for quotations. Indulge us that instinct while we look at where we are with robotic process automation (RPA).

First, some background. RPA has had a meteoric rise. From a standing start – that’s a zero on Google Trends – as recently as five years ago, it saw a steady growth in interest with big jumps in enquiries at the start of 2020 before settling back down just a bit.

RPA Blog Google Search trends

What explains this sudden growth in interest? Well, simply put, RPA has been sold as something of a panacea for businesses looking to boost efficiency, cut costs and solve decades-long problems with legacy systems.

And that’s the problem. Like so many technologies before it – from ‘knowledge management’ and ‘thin client’, to blockchain and augmented reality – the blueprints looked utterly compelling, the sales pitch a dream. And the reality turned out more prosaic.

The Google Trends graph suggests a peak has been passed. (Don’t get too excited: other issues have clearly been prioritised over the past three months…) But, there have been quite a few articles and even some anecdotal evidence that backs up the assertion that the RPA bubble has… well, if not burst, then deflated a little.

Which bring us to our aphorisms...

 “Robots came. They saw. They conquered.”Julius Caesar

The big idea was that companies could pick up slow, intensive tasks, deploy a simple piece of software onto them and suddenly deliver super-efficient processes that would free up personnel. Cheaper, faster, better – what’s not to love?

But the mission for RPA was seldom well-articulated by the clients who bought into it. Far from simply seeing a process and fixing it – Julius Caesar’s “veni, vidi, vici” is a byword for a fast, compelling victory – the bots in RPA need precise fine-tuning. Those who barely described and diagnosed their problem bought licences then wondered where their solution was. Where were the too-good to be true results they had heard about? Why were there so many exceptions?

True, we’ve seen lots of impressive examples of fast deployments of automation during the pandemic lockdown to handle data management, leaving human operators to deal with exception handling. But in reality, a lot of groundwork is required to make RPA successful.

“No, no – he's not dead. He's… he's restin'! Remarkable technology, RPA, isn’t it, eh? Beautiful coding!”Monty Python pet shop owner

Some of this failure to manage expectations has to fall on the heads of RPA vendors. To get their foot in the door, some made bold pitches promising increased capacity and reduced headcount, with return on investment (ROI) running into the hundreds of percent.

Compare that to the sell for a shared services centre or business process outsourcing deal – which demands lots of groundwork, a great deal of project management and some serious reimagining of the enterprise itself. Some RPA firms do make this a key part of their pitch. But others have been guilty of chasing the deal at all costs.

Their promises poison the metrics for success. No-one would evaluate an ERP programme or BPO deal on the raw speed of installation, or the number of FTEs saved, without also weighing up the training needs for the workforce, the implications for supply chain reorganisation or the effect on risk and reporting. Yet RPA, often viewed as a project, rather than full programme, has had a tendency to be measured on the number of bots deployed or speed of roll-out.

The result is frustrated clients asking where their value is while RPA vendors stare at their shoes, regretting the decision to pitch on ROI alone. Offer any CFO a 200% return on investment, and they can’t say no. Fail to prove it after year one, and they’re unlikely to be happy – or ready to double-down on their investment.

“Friends, Robots, countrymen, lend me your ears; I come to bury RPA, not to praise it.”Mark Anthony

Honesty is the best policy: RPA is not a magic wand for transformation. It’s hard, it requires significant technical proficiency and it needs, above all, major workforce realignment.

Common elements we’ve seen in failed RPA projects include:

  • They’re led by mavericks, rather than those with enterprise clout. That hampers buy-in and limits access to people such as enterprise IT architects or the core team of transformation specialists who understand the wider implications.
  • They’re pet projects not taken seriously by business heavyweights – unlike a typical ERP-type implementation, which gets leadership buy-in and proper training across the business.
  • They’re often siloed into outposts, not core business areas. That means gains are always going to be limited.

When that happens, the RPA project may be regarded as a failure; and those evaluating it – from a technical, financial or strategic point of view – will all-too-readily overlook any positive indicators and write off RPA as a niche tech.

A lot of the solutions architects whose buy-in would boost RPA are purists, too. They want to build from the ground up with pristine APIs. But many business users, working almost exclusively in Salesforce for example, love the idea of a quick fix to annoying problems. They buy bots as a kludge or workaround. But to scale and deliver real value, RPA needs to be treated as a strategic programme that is embedded in the back-office under the wing of dedicated solutions and transformation people.

Worse than that, a bug in an ERP module will get the full attention of the systems team and a fix produced. A bug in a localised RPA implementation at best throws out an exception that needs handling; at worst, it just stops a process dead with little technical support because a ‘citizen developer’ from a business unit never had the same standards for the tech or the process design.

And if a bot simply fails, one of the key benefits that was sold to clients – freeing up employees for higher value work – starts to look extremely hard to attain. You can’t redeploy people if you can’t trust their replacement.

But all is not lost. By learning lessons from some early missteps in RPA implementations, we can see a very vibrant future for intelligent automation.

 In part II of this blog, we’ look at the hallmarks of successful RPA and ask who’s responsible for giving automation life in the future...  Read it now!

New call-to-action

4 minute read