The top 10 causes of robotic process automation failure

Roughly half of all robotic process automation (RPA) projects fail. The truth is, most cases of robotic process automation failure is because of people. People’s perceptions and understanding of automation, and the way they plan and run their RPA projects, are the most commons causes of failure.

Want to avoid the most common RPA mistakes? Knowing what they are will help.

Here are the top 10 reasons why half of all businesses experience robotic process automation failure.

      They have unrealistic ideas about what automation can do

Automation is just one part of a wider digital strategy; the tech itself is just an enabler. But people often forget this, expecting their robots to solve every process problem they currently have. Spoiler: they won’t. These unrealistic expectations are a prime cause of failure.

Take a call centre, for example. No robot can pick up a phone and have a call with a human being (yet). But they can log transcripts and notes, and handle a lot of your post-call processes.

Understanding what RPA can and can’t do is the first step towards using it successfully.

Top tip: Understand how RPA fits into and complements your wider digital strategy

      They focus on quantity of robots rather than quality of work

The cult of volume is a great way to ruin an RPA project. Business leaders get one robot, they want more. I once saw two c-suite members from different countries obsess over who had the most robots. But quantity is not the way to measure automation success.

You have to measure success by the return on investment to a business. An underutilised robot is the same as a human being sat around twiddling their thumbs. Early in my career, we had this expression: ‘Make the robots sweat.’ A robot that was optimised at 80 per cent was a successful robot. 

Top tip: Set KPIs for robot productivity and meet them consistently before investing in more

      They see automation as a quick fix for a few boring tasks

When most people talk about RPA, they’re really talking about RTA: robotic task automation. This is a legacy of the early days, when software vendors just wanted to sell licences. ‘Here’s a task,’ they’d say, ‘let’s automate it.’ This approach is a major cause of RPA failure.

To successfully automate your business, you have to consider the full opportunities and implications of RPA. For example, ask yourself: ‘If we had infinite resource, what would we do that we’re not currently doing?’ This should be the starting point for any RPA business case.

Top tip: Think about what your people and your business could be, not what the tech can do

      They give robots dirty data but expect them to do clean work

Like any human employee, a robotic worker needs the right conditions to work well. Their productivity might not be affected by the level of chat, light, or heat in your office, but robots will still ‘suffer’ in a bad environment. For RPA software, a bad environment means bad data.

Too many businesses start their RPA projects without first sharpening the most important tool a robot worker needs: information. You’ll never optimise a robot at 80 per cent or more by forcing it to work with incomplete, badly formatted, or unstructured data.

Top tip: Ensure that your data is complete, structured, and well-formatted before roll-out

      They use the wrong RPA solution for their programme

In the beginning, most software robots were unattended and worked in the back office. You gave them some rules and let them do a routine job behind the scenes, without supervision. Early RPA software was designed for this, and that suited most projects at the time.

But when robotic automation moved into the front office and became more collaborative, new RPA vendors and solutions emerged. So if your automation programme involves humans working alongside robots, there’s an RPA solution that’s best for that.

Top tip: Do your research into the vendors and solutions to find the right one for your goals

      They don’t listen to the business automation experts

The reason I can write this article is that I’ve been working with RPA for over a decade, long enough to see every mistake on this list made many times over, by many different businesses in many different sectors. I’ve collected my fair share of medals and scars.

If you hire an experienced business automation adviser and then ignore their advice, you’re setting your project up to fail. This is especially true if you’re combining RPA with a layer of intelligence, such as machine learning or natural language processing.

Top tip: Consult an experienced RPA adviser and let them guide your key decision making

      They think automation is a matter for the IT department

Something I hear often in early-stage automation meetings is, ‘Well, I’d better get an RPA developer then.’ This is before any real planning or discovery has been done. Seeing automation as an IT-only matter is a perfect way to make your project fail.

RPA is a whole-business opportunity, one that must be supported by your IT department and fully governed and sponsored by the c-suite. But successful RPA requires more than just developers.

You need business analysts, testers, process controllers and business process consultants.

Top tip: Invest in the infrastructure and people you need to make your RPA a success

      They don’t reassure their people that automation is okay

Say the word robot in a work context and you stir up a mix of emotions, especially fear. Even though RPA involves no physical robots, people will picture them marching into the office to take their jobs away. When businesses don’t address these fears, their projects fail.

Run RPA surgeries with your advisers. Involve people, ask for their thoughts, let them ask you questions. Show that RPA is there to make their jobs better, not to take them away. Publish articles, set up steering committees: whatever it takes to improve understanding. 

Top tip: Carefully consider how best to communicate the benefits of RPA to your people

      They only automate happy paths and forget that things go wrong

Early RPA planning is about finding processes with a high potential for automation. That often involves studying your standard operating procedures, the manuals that tell a human being how to do the job and – crucially – what to do when something goes wrong.

A top cause of RPA failure is that people only show their robots how to complete a process when everything goes right. What we call a ‘happy path’. But what happens if you can’t get to the end first time? What’s the blocker and how would you deal with it? You must automate for these ‘unhappy’ paths as well.

Top tip: Map the full process from start to finish, including all potential failure points

      They mark their own homework (and always get an A)

A prime cause of failure for automation projects is poor governance and security: the people who do the work are often also the people who check the work. This is a recipe for disaster, with the disaster taking the form of a live process that doesn’t work.

You have to have a control room and a proper workflow, where work goes into production from a developer, is reviewed by a process controller, and sent back if something isn’t right. This segregation of duties during implementation is absolutely vital to RPA success.

Top tip: Separate quality control from development and put rigorous checks in place

RPA chipLiked this article? Get the full RPA guide

Our guide to the three stages of RPA success shows you how to plan, test, and implement automation in your business using our tried-and-tested method.

Fill out the form below to receive a copy of the guide.

By submitting this form you acknowledge that RSM may contact you regarding the report as outlined in our privacy policy.

automation now: how to succeed with RPA