Finance organizations expect to increase their adoption of robotic process automation (RPA) by over 12 in the next two to three years, according to The Hackett Group research. But the growth will hinge on a critical step – scaling up from one bot to 10, or 50 or 100. While one bot can work faster and more accurately than a human, for finance to realize the full potential of RPA, it must scale up. And that’s where a lot of organizations get stuck.
Moving beyond proof of concept
Many companies begin their robotics journey almost randomly; they initiate discrete, uncoordinated pilots in disparate areas. To scale up, companies must “loop” back into a holistic RPA strategy, according to The Hackett Group’s RPA Practice Leader, Paul Morrison. And that means building an entirely new business case. It also means thinking of robots as part of a comprehensive governance and infrastructure system, and not just within finance.
The best practice is to develop a “live” list of potential opportunities, curated by an RPA center of expertise (COE) or project team. Morrison says the team should use an iterative process to come back to that list and re-prioritize and refresh it at regular intervals.
Our experience shows that finance functions are increasingly adopting the COE model for the adoption of RPA and other smart technologies. The COE leverages resident (or virtual) resources and acts as a hub for coordinating strategy, designing, implementing – and sometimes even running – the bots.
To scale up successfully, finance organizations should define clear objectives. Is the primary goal to reduce cost? Accelerate cycle time? Improve quality? Or some prioritized combination? The COE can then manage the conceptualizing and enterprise-wide implementation approach by collecting direct input from process owners and participants to define in-scope activities and processes. The COE should develop a pipeline of opportunities. This disciplined process is more likely to identify projects that will be successful. (As RPA goes “enterprise-wide,” the same principles apply.)
“At the core is a structured process for identifying opportunities,” Morrison says.
It’s important to consider that robots are limited in terms of functionality. For example, they can’t handle unstructured documents (without additional intelligent data-capture capabilities). But they can take over transactional activities that do not require judgment and are “fed” structured digital information. If the data is not structured, the project team should consider “diverting” the process toward the path of artificial intelligence and cognitive computing, perhaps in combination with RPA.
With a defined goal and a prioritized pipeline of potential applications, companies can begin to scale up. Implementing one bot can take six to eight weeks. Successful implementation of multi-bot projects can take a year to 18 months, with the first wave lasting three to six months. Best practice is to scale in a rolling fashion. After designing the first one or two bots, start deploying more robots on a staggered basis.
Overcoming the hurdles
Not involving the IT function from the start is the single biggest hurdle to a successful project. “Bring everyone under the tent,” recommends Morrison. That includes not only IT but the process owners and business users as well.
Another potential obstacle is not having a disciplined methodology for identifying and prioritizing the RPA opportunities. The decision about which initiatives should be launched first can be based on what’s easiest: the low-hanging fruit. It can be driven by the readiness of specific user communities. But regardless, factor in what would make the greatest impact.
The technology implementation methodology can become a stumbling block to rapid scaling. RPA is being deployed alongside core legacy systems that have typically demanded the structure of formal stage-gated (“waterfall”) development. That’s why it’s essential that the IT organization build the foundation for RPA in terms of access, control, and deployment. When it comes time to roll out the bots, deployment should happen in controlled sprints and involve close user oversight to adjust and monitor as automation volumes ramp up.
It’s best to start small with one or a handful of robots, demonstrate their benefits, and use the experience to de-risk the process. Most of the issues bubble up at that early stage. Then finance can leverage lessons learned to scale up quickly. “Don’t start with 100 robots unless you have a lot of help,” Morrison says.
For more on emerging technology in finance, see Intelligent Finance Boosts Business Performance.