At a glance
On the website of Automation Anywhere, a robotic process automation (RPA) software platform, you’ll find its Bot Store that, among other things, makes it clear how much time employees waste doing repetitive, manual tasks during the average business day.
The business supplies bots – autonomous programs that can interact with various systems. These can, for example, take customer details from a sales system and add them to a different customer relationship management (CRM) system. Other bots can add new customers to QuickBooks, add suppliers to Oracle enterprise resource planning (ERP) accounts, or automatically convert files from one format to another.
Some bots can carry out relatively complex tasks, such as analysing sentiments on Twitter towards a particular brand or product, creating or updating supplier bank details across disparate systems, or detecting a foreign language in an Excel file, then translating it to a target language.
Finally, among the hundreds of offerings is software providing “digital workers” that exceed a bot’s function by conducting workflows that imitate those of human workers in particular roles.
Among digital workers there is the digital accounts payable clerk, for instance, which extracts invoices, reconciles them with a purchase order, then processes the invoice and the payment. The digital talent acquisition specialist can create an offer letter, request a signature, and check for offer acceptance, then create assets for a new employee.
Complex or simple?
Welcome to the brave new world of RPA, claimed to potentially reduce manual operation costs by up to 40 per cent while generating almost immediate return on investment (ROI) – all without having to purchase new systems that integrate with each other.
“RPA is either a very complex story or a very simple one,” says Tim Ebbeck FCPA, managing director of Automation Anywhere. “Why go for complex when simple is available?”
Ebbeck explains, “People work by doing different tasks, different processes they have to follow to get their job done. When they get some time outside these tasks, they are able to apply some thinking, rather than just doing. At the end of the day, with RPA we’re trying to augment humans with digital workers to create a workforce of the future.”
RPA simply replicates the “doing” piece, taking away the mundane, repetitive tasks and allowing the human to do the cognitive work that only a human can do.
“Scanning a piece of paper into a system digitises it, but doesn’t release the value of the unstructured data,” he says, “so the human capability that sits on top of the RPA capability is essential for cognitive capability.”
Look before you leap into RPA
One of the most interesting facets of RPA is that it is technology agnostic. Organisations are regularly told that for the very best outcomes, their systems should work together as one, be fully integrated and produce the same type of data that is collected in a central data management system.
Then along comes RPA to say that massive expense won’t be necessary. Your old systems can have a stay of execution; there’s a new way to achieve efficiencies and you’ll see a return on investment within 12 months.
We are constantly told that things that appear too good to be true typically are, so is RPA the solution to every organisation’s operational woes or is it high-tech snake oil?
It’s neither, experts say. It is an excellent solution for specific challenges, and it’s vital to be sure that RPA won’t sink your ship before you bring it on board.
“At this stage, these bots are just mimicking human actions, yet they are evolving to attain cognitive capabilities,” says Mohit Sharma, executive chairman of Mindfields Global, an RPA advisory firm. “In conventional automation, you have to do some tough coding. You have to change things in your current systems. With RPA, nothing has to change at that large scale and at that large cost.
“But right now there are tasks RPA is not good at. If you are looking at digitising text from hand-written documents, for example, accuracy is only 60 per cent to 70 per cent. I am an accountant and I will never implement an RPA process unless there is a solid business case behind it.”
If an internal process is too complex prior to automation, Sharma says, then the process will likely require re-engineering before RPA can be successfully introduced. “We might also have to put extra controls in place, particularly in areas such as accounting, because the work is now being done by a bot, not by a human. These automated processes are also subject to audit, hence we need to have proper controls.”
The early planning part of the process, Sharma says, is known as “automation enablement”. During this period a business case must be developed. If there’s no business case, automation should not be attempted.
Several vital factors must be considered, Sharma explains. The first and most obvious is monetary saving, and the next is the reallocation of staff or enhanced customer experience as staff can now better focus on the customer.
Then there is error reduction. What is the current error rate and how much is that costing the business? What percentage of errors will be avoided via automation and what savings will result?
There are also non-monetary benefits that must be considered, including enhancement of employee experience, as staff no longer need to do so many mundane tasks.
Seek IT specialist advice
Farid Jarrar, chief information and digital officer at CPA Australia, agrees that RPA is not a one-size-fits-all solution. Currently, it is not good for anything that requires cognitive, nonrule-based processes.
Having worked in the RPA industry, Jarrar has seen several organisations make the wrong decisions by employing technology for technology’s sake.
“When new technology comes around, it’s easy for people to see it as a silver bullet,” he says. “People fail to understand how to implement it. They don’t actually understand that the technology is one component, but there are a lot of business process changes that will also be required.
“If you’ve run a pilot program it still doesn’t mean you’ve got the technology ready to go and ready to scale. Getting the technology production-ready and exception handling-ready requires a level of additional effort.
“Finally, the technology itself is not cheap, so if you don’t have a worthwhile ROI and you don’t have the volume, then don’t bother. It’s probably cheaper to get a human to do it.”
Success with RPA, Jarrar says, is about developing a deep understanding of how the technology will be deployed, how it will be leveraged within the business and what will be required for the best outcome to be achieved. Although it is a business-wide project, it is vital to involve the IT department at the earliest possible stage.
Put simply, RPA works in an environment that has a large number of transactions, a large number of people, a hybrid environment of different technologies and a lot of repetitive, rules-based work, Ebbeck says.
Where businesses go wrong with RPA
In a way, it is far easier to identify the benefits of RPA than it is to recognise the dangers. If you run a large business with numerous customer touchpoints and a bot is introduced that can automatically take information from one system and drop it directly into another in real time, it will likely save customers wasting time as customer service staff will have the customer’s information readily at hand. It will be good for staff engagement as it means people can spend more time solving problems and creating excellent customer results. It will make the brand more customer-centric.
There are, however, hidden complexities in implementation. What does such a system mean for current processes? What requires re-engineering for the new efficiencies to be fully realised? What training will be involved? Will customer service staff require different skill sets? How can staff talent be better focused now there is less repetitive, manual work to be done?
An EY report, Get ready for robots: Why planning makes the difference between success and disappointment, says 30 to 50 per cent of initial RPA projects fail “This isn’t a reflection on the technology; there are many successful deployments. But there are some common mistakes that will often prevent an organisation from delivering on the promise of RPA,” it says.
Automating the mundane work
When RPA is utilised well, it can be of enormous benefit to a business. Sharma describes a client who previously employed 40 people largely to manage change-of-address requests across several systems. The successful implementation of a bot that transferred new address details between systems meant only two people were required for exceptions handling.
This also led to automation of cross-selling on the basis of address change from “posh” postcode to “non-posh”, and vice versa. Other staff could be moved on to greater-value roles that helped the business grow.
Ebbeck relates RPA to accounting by describing the downloading of bank transaction statements and the uploading of those statements into the business’s accounting system, all processes that could be handled automatically by a bot.
“People have to sit over these tasks just to do a simple bank reconciliation because they’re getting data from different environments,” he says. “That’s the stuff that people don’t like doing. It’s the reason that engagement scores in many organisations are nowhere near as good as they should be.
“We automate the mundane so people can get back to being customer-centric decisionmakers, to being innovators. RPA can free up the capacity of the human workforce to do what humans are good at by letting robots do what robots are good at.”