How to get started with RPA: 5 projects to consider
As organizations increasingly ramp up their plans to pilot and phase in rollouts of new robotic process automation (RPA) projects, many enterprise leaders are on the hunt for the best use cases to get started with RPA.
Deloitte says that 72% of organizations plan to either embark on or fully commit to their RPA journey by next year. As newbies seek to prove the viability of RPA for their organizations, they should seek out RPA-ripe candidates with five defining features, said Imran Sabir, automation lead for consulting firm OZ.
There could be a lot of candidate processes for RPA, but the best candidate tasks have stable processes, repetitive tasks, and very clear steps, and they are well structured and well documented, Sabir said. "Automate simpler tasks first."
Here are some common use cases that fit that mold, making them great candidates for starter RPA projects.
These are tasks where humans essentially respond to alerts or prompts from one system, then take some of the data and take action in another system. Not only is systems- or function-switching fundamental here, but it's done in a repetitive, repeatable fashion.
Aniket Maindarkar, vice president and general manager of the business process services practice at DXC Technology, explained the ideal candidate: "So, I receive a file, I open a file, I look on the left, the first field is the purchase order, the second field is a date. I look at a different file. I compare either the goods or services received and compare it to what the invoice was. Then I do a two-way match or three-way match, I look at it and then I go to SAP and I trigger a payment based on that and so on. So it's very repetitive and very standardized."
Experts universally agree that finance is usually the first business function in which to start looking for beginner projects.
"Within finance there are a lot of 'swivel chair' activities," said Tim Kulp, vice president of strategy and innovation at the consultancy Mind Over Machines. One example is collecting data from your operations systems, such as the number of cases worked for a client, and then entering that same data into an invoicing system.
There's no need for double entry. With RPA the process of reading data from operations and entering it into invoicing can be automated. "These repetitive tasks are easy ways to dip your toe in the water with RPA," Kulp said.
Another example might be in procure-to-pay, such as processing inbound invoices, reconciling invoices, and issuing payments.
These are all repetitive tasks, and a safe environment in which to start with RPA because they won't directly touch customers or revenue generation activity but still fill an important need.
Experts say that aside from finance, human resources stands as another universal business function that's full of use cases for RPA beginners. One of the most obvious areas is the employee onboarding process, which usually involves a very defined process that requires a structured set of information that must be entered into different systems, documents that must be scanned, and so on.
There is a lot of effort in RPA spent on onboarding activities, said Sairam Bollanpragada, head of the global delivery center for Micro Focus.
"Sometimes going onboard can take a full day or three-fourths of a day, and it is so exhausting in nature. Automation can help save employees time and save the HR team time."
Take, for example, a firm that requires new employees go to HR to fill out a bunch of forms, submit documents, and receive employee handbooks or other information. This not only requires the employee to be tied up in the process, but also usually needs an HR person to shepherd the process and ultimately input the information, initiate new accounts for the employee, and so on.
That process could greatly be aided by either a kiosk or an online mechanism that scans documents and receives soft-copy inputs into a system that then uses RPA bots to automatically do the back-end work to route information to the correct corporate systems, initiate provisioning of new accounts, build access badges, and send out relevant new employee information to the person based on the information received and the role they're filling. "Otherwise, there is a dedicated HR person who sits and does all of that," Bollanpragada said.
Beyond onboarding, organizations could also look at other elements of workforce management and employee interactions with HR for other potential RPA starter projects. When it comes to HR use cases, DXC Technology's Maindarkar said there are generally two sides of the coin: Back office and front office.
Back-office automation includes not only processing onboarding information, but also things like payroll management and staffing coordination.
Similarly, this area could also include reimbursement claims approval processes, which can eat up a significant amount of time. Many firms have staff eyeball employee scans of receipts and ensure that the totals match up in order to approve the claim.
Now the bot can do that by doing an optical character recognition scan, and for many of the receipts if there is a clean match it can give approval instantly, Bollanpragada said. And then if you still need to do a physical approval because the bill is smudged or not clear in nature you can review those exceptions.
"But that saves a lot of time because if you have a company with 50,000 employees and if even 30% of them are submitting claims regularly, that's quite a lot of volume of work."
Making employees happier
The opportunity for workforce management projects doesn't just have to be on the back end. Maindarkar said the other side of the RPA workforce management coin is with front-office interactions, which he believes may offer even greater value because they provide an opportunity to improve the employee experience.
If somebody is calling HR to say, 'I worked eight hours but you credited me for seven hours, or, 'I took a day off and it shows medical leave but it was something else,' RPA could offer a way to get their questions answered in a more seamless and efficient way, he said. "A lot of times you have these silly things where it takes forever to apply for leave, or get payroll issues sorted out, and that causes frustration among employees."
Using chatbots and other RPA mechanisms, he said, companies can automate the usual 80/20 rule, so that the 80% of questions that are repeated on a daily basis can be answered instantly. Not only does this save time, but it will hopefully track back to measurements for the happiness of the employee base.
The IT team could also stand to look in the mirror to find inefficiencies in its own processes, such as in IT service management, that could provide beneficial pilots and early projects to flex its RPA skills. According to OZ's Sabir, areas of IT that could potentially be streamlined through RPA include server and application monitoring, routine maintenance and systems monitoring, batch processing, email processing and distribution, backup and restoration, and password resets.
That last one in particular offers a clue as to the depth of opportunity for RPA projects to help smooth a lot of the system-to-human bottlenecks in the IT service management (ITSM) space.
Projects that don't make sense for RPA
As organizations survey the field for the perfect first projects to get their feet wet with RPA, they should definitely be discerning about use case candidates right out of the gate. These early projects will set the tenor for RPA across the organization and an early fail could stymie automation efforts later on down the road. And there are definitely areas where RPA is not well-suited, said Tom Wilde, CEO for Indico.
"While its benefits are undeniable, RPA also has well-defined limits. If your business process is flawed, RPA simply automates and accelerates that flaw. It cannot make judgments about information or learn and improve with experience.”
One good litmus test for figuring out projects that don't make sense for RPA is to look for workflows that involve a lot of unstructured data. RPA is best suited for processes that uses structured data with clearly defined values, such as spreadsheets, tables, and the like. Unstructured data such as documents, free-form text, and images that need interpretation are a lot trickier.
Unstructured data requires some level of cognitive ability to interpret—and that makes up over 85% of the data in most enterprises today, Wilde said. "In areas like legal and compliance, sales and support, and finance and operations, this type of data is the foundation of almost every mission-critical business process.”
In these cases, it may make sense to take a light touch with RPA to avoid problems.