Implementing Intelligent Process Automation (IPA): Four mistakes to avoid
Intelligent Process Automation (IPA) and the less-pretentious sister RPA (Robotic Process Automation) have been at the center of attention over the past few years. The promise is obvious: repetitive tasks that are currently carried out by humans can relatively easily be done by software robots using the “human” user interface to interact with applications. The addition of machine learning and natural language processing to the bots' capabilities have even augmented the business potential.
The IPA software market has become extremely competitive, backed by investors who are aware of the importance to stay ahead of the pack to ensure that you are one of the winners in the end. Features include a better interface for robot developers (in many cases close to the so-called citizen developers where business people create their own robots); better ways to deal with documents through optical character reading; the inclusion of AI techniques for documents and pictures recognition; machine learning to learn from cases where human intervention is required; and tools to support the management and orchestration of a true workforce of software robots.
According to a recent Forrester research, 90% of all organizations with over 500 employees have encountered some form of RPA of IPA over the past five years. That sounds logical: all organizations have some form of repetitive work inside their office walls, that was just waiting to be robotized. This does not mean that all organizations have successfully implemented IPA and RPA to the extent that the digital workforce is really making a difference to the bottom line of the organization: the road to success is long and winding, as the Beatles already knew.
When comparing many projects that we know in-depth, we notice that the ones that take much longer than expected to succeed have several characteristics in common. We could clearly distinguish four factors that should be avoided to make the IPA venture a success.
Mistake 1: Think too small
Many initiatives fail because they have not worked from the very start on a complete business case for a large-scale implementation of a digital workforce. We have seen many examples of companies that started their IPA journey by executing a Proof of Concept to “see what happens”. By itself, this is a totally viable option to investigate if a relatively new technique can work.
But if this small Proof of Concept – made easy because most vendors provide really low-entry desktop environments to do so – is not getting the attention from senior management who can really decide to scale-up, then the risk materializes that the focus is shifted to some other new way of improving business value.
Simply stated: one employee is the owner of a small Proof of Concept, that worked out quite well by itself. However it does not get anymore the attention it deserves. The initial momentum is gone, the budgets to scale up are not in place, and the company loses 6 to 12 months before a new attempt can be done.
“Thinking too small” may bring you to a standstill.
Mistake 2: Think too big
The opposite can also happen. There are examples where the CEO dreams of an army of 200 robots, enabling to reduce salary expenditures by 50%. The company announces that Intelligent Process Automation is a life saver and that the digital workforce will make the company profitable again. Within 12 months.
IPA and RPA can indeed contribute significantly to the bottom line. But three mistakes fall under this “think too big” header. First, a sound data-driven business case should be created, based on facts, not on dreams. Second, it should not be underestimated that a significant effort in time and money is required to develop a digital workforce. And third, there should be the awareness that not all human work can be replaced by robots, in spite of what some visionaries may assert.
“Thinking too big” may lead you to deception.
Mistake 3: Work without an implementation partner
Our research clearly shows a correlation between having an implementation partner and moving forward fast. Especially in the start-up and ramp-up phase of Intelligent Process Automation, a specialist in IPA and RPA can bring a significant amount of knowledge and experience to get organizations through the first couple of hurdles that are common across all implementations. They speed up the learning curve. In many cases it shows that this is both important for the adoption of Intelligent Process Automation in the company as well as financially the better choice in the long term. An implementation partner is also important from a security perspective, as many of them are specialized in bringing in the necessary experience to make sure that the robotized processes fit within the compliance framework of the organization.
Let's share here an example from the Financial Industry. They were absolutely not afraid to invest and even used two implementation partners next to their own developers and analysts. Within 9 months from starting the IPA project they had over 50 robots running and were saving about 20 FTE in an average week, spiking up to 50 in some weeks of the year. And all of that was only possible because they chose to invest in implementation partners as well as their own staff. What was impressive from this project was the vibe that existed in that company after nine months, people from all different departments came to the developers with new potential process on a constant basis.
Mistake 4: Let the implementation partner do all the work
Having learned from the previous point that it is advisable to have an implementation partner, it is not recommendable to let them do all the work. Intelligent Process Automation is all about the organization’s business processes, and who knows them better than the employees who have been working in the same organization?
The organizations in our survey claiming to be the most successful with IPA are the ones that have set up their own CoE: Center of Excellence, that bundles all the organizational knowledge on IPA and RPA. They started the IPA/RPA effort in a joint project with an implementation partner, with the strong goal to do it mostly on their own after several iterations. This requires hiring the right people, or re-skilling part of the original workforce. Hybrid models are also visible, where the CoE is partly staffed with consultants from the implementation partner (bringing in lateral ideas and providing demand-flexibility) or where robot management is carried out by a third party. In the end, IPA/RPA is a mix of advanced technology – typically not at the core of an organization – and the business processes of an organization – which by definition are at the core. Mixing these requires a delicate balance between do-it-yourself and let-it-be-done by others. The truth lies, as so often, in the middle.