Graphic of person upset because their automation failed with icons

Three Reasons Automation Fails

The best part of being in the world of technology is change! We get to see changes in the technology stack almost daily, many of which improve operations and service delivery. One of the more dominant trends right now is the usage of automation and while it may seem new, the tech industry has been utilizing automation for decades. Today, robotic process automation and artificial intelligence/machine learning enable organizations to reduce human interaction and costs while improving operations.

The idea of automation is compelling, but it isn’t the panacea for all operational issues. In fact, recent reports state that nearly 50% of robotic process automation (RPA) activities fail. The question and concern to be addressed is why they fail. Here are a few main reasons:


One of the primary reasons automation projects fail is a lack of governance. RPA or BOTS must be managed, built on well-established processes, and treated just like a member of the workforce – monitored, and periodically corrected. 

Automation for the Sake of Automating

Automation should span the enterprise and be evaluated for return on investment and total cost of ownership. Repeatable (sometimes mundane) tasks are primed for automation, whereas automating a task that requires constant updates to the BOT becomes a rain on organizational resources. Not everything can or should be automated – sometimes the juice just isn’t worth the squeeze.

Failure Through Design

Poor design practices are still prevalent and widely applied in the development process. While the concept of “no skills” needed to develop a BOT is great, think of the implications of a poorly designed BOT running with no controls, monitoring, or concern for total cost of ownership. BOTs should be created with the same rigor as any other developed program – incorporated into change control, monitored for effectiveness, and retired when no longer useful.

Automation is a great feature and function to improve operational capabilities and service delivery. Like any other tool, it should be used for the right purpose and with the right expectations. Consider the results – quality of implementation and deployment, overall productivity increase, impact on people and operations. We wouldn’t accept carpentry work where the carpenter drives finishing nails with a 20-pound sledgehammer, and we shouldn’t accept poorly developed workflows, automations, or robotic processes.