Skip to content

Home Blog Why do RPA projects fail? Automation is what we make of it

Why do RPA projects fail? Automation is what we make of it

11 out of 12 automation projects fail. Robocorp's COO explores why and outlines a better path forward.


RPA and automation have become a very hot business area and many companies have a lot of expectations for RPA projects. At the same time, a Deloitte study found that only 8% of organizations were at the stage to scale up automation. It is nothing new that a new fast-growing industry looks for its best model and position. Citizen development has been an important part of RPA growth, but it is very unclear if that model can really deliver results.

Citizen development RPA cannot survive contact with reality

Citizen development is a nice idea. It can get anyone to automate their own routine work. It sounds great: you give tools to those in your organization, some people coordinate the work, and then educate people to make their own work better. Can anything go wrong with such an idealistic idea? It is like often said, no war plan survives contact with the enemy, the automation reality and requirements are often more complex than thought.

We can identify citizen development model problems on three levels:

  1. Strategic problems. The real value of RPA projects are not achieved by automating independent routines of individual employees.
  2. Business model problems. It doesn’t make sense to pay for RPA tools for a large number of employees, when the use and value for individual employees is unclear; companies should pay for real use and value.
  3. Implementation and technology problems. Low-code tools are not powerful enough to make proper automation and many citizen-developers don’t have enough skills to make robust automation implementations, especially any mission critical processes.

Let’s look at each problem a little bit more carefully.

Strategic problems with RPA projects

A video rental store doesn’t become a new Netflix simply because employees automate some of their routine work. A high street bookstore doesn’t become a new Amazon when employees automate their routines.

The reality is that to automate some routines or processes can increase productivity by a few percentage points. If you want to really increase efficiency, to make things 10 or 100 times more effective, you must create new processes. And you cannot make these processes only for individual employee’s routines, but how different parts of the organization and functions work as a whole.

When you start an automation project, you must identify how it can give strategic value and make processes better and more effective. Sub-optimization of individual routines is not a sustainable solution. Sooner or later a competitor improves operations by utilizing all opportunities of automation and digitization.

Automation business model problems

Currently, enterprises must buy software licenses for proprietary solutions. These solutions create a vendor-lock, and solutions in closed ecosystems cannot be properly integrated to other systems. And although they are low-code solutions, you need to code to make implementations more robust or integrate them to other solutions.

Does it really make sense to buy automation tools for a large number of employees and educate them to implement automation? If you have professional accountants, lawyers and logistics experts, is their time best used by learning automation tools, especially tools that are quite limited. Of course, they know the problems and needs, but it doesn’t mean they should be the ones to implement the automation project.

At Robocorp, we believe that open source is the way to provide development tools and integrations, and to avoid a vendor-lock. We believe that you should not pay for development tools; that you should pay when you get value from your automation. The consumption model is generally accepted to be the model to use software nowadays. You can do it with public cloud or private cloud solutions.

If you need to pay up front significant license and education costs, it is an obstacle to start automation, and you cannot easily test new things. It is an especially risky investment when you don’t know how many people really need or are able to use those tools.

Problems with RPA implementation and technology

Low-code trend is nothing new. I have personally seen it for the first time in the 1990’s. It always comes with big promises and then fades quietly. Low-code works for some simple tasks of individual employees, like Excel macros work. No-code also works for some purposes, like making web sites from standard components. But if you really want to make professional software implementation, you need professional tools and developers. Why otherwise world leading technology companies pay $250,000 or much more for professional developers?

It is ironic that the robots on factory floors are designed by top-level professionals, but the robots for the more demanding software automations are designed by citizen-developers and low-code tools. At the same time, AI is only growing, and automation tools should really support AI integration too. In practice, implementing RPA projects should require professional programming tools and languages, like Python, to build AI with automation.

Don’t get me wrong; software development tools and models are developing, and many tools can help the citizen developer. The reality, however, is we need more professional software developers and more reliable software. It is the only way to get robust and reliable automation and robotic implementations. If implementation is not robust and scalable, we never see full value of automation.


The promise of low code RPA has many people excited. They are sold on the idea that you simply buy tools, pass them around the organization, and suddenly manual routines are automated. Implement RPA projects, then sit back and wait for more efficient teams. But the next step, to implement automations that can really improve a company’s productivity and bottom line, has proven elusive.

Additionally, as we witness the transition to digital operations and better utilization of AI, automation will continue to be a crucial part of these processes. Automation is a critical component of AI, which requires professional and robust automation implementations. It also means companies must re-engineer their processes or design totally new digital processes; it is not enough to automate individual routines.

Finally, low-code citizen development solutions have their time and place, but it is fundamental to understand that a more professional approach, what we at Robocorp call “Automation Ops,” is now needed. Automation is a key area to improve productivity in small and large companies. Many companies have started the automation journey, but when they realize they need more robust and scalable solutions, they are disappointed by the results or abandon the project entirely. That’s why we at Robocorp want to offer professional tools and scalable business models, and to truly support automation professionals and developers on their journey towards enhanced Automation Ops.

Jouko Ahvenainen is Co-Founder and COO of Robocorp, a serial entrepreneur, and a tech and business pioneer.