Robotic Process Automation (RPA) has the potential to increase productivity and cut costs, but there are also risks, so it is important to choose a supplier carefully
Robots have already had a significant impact in the manufacturing industry where they have replaced blue collar workers from production lines.
They are now poised to have an equally transformative effect on white collar work, not in the shape of C-3P0- like personal assistants but in the form of software ‘bots’ that automate the routine, repeatable administrative tasks that take up so much time, such as moving information between applications and collecting data from different sources.
McKinsey & Company estimates that technology already available could automate as much as 69% of the workload of employees in data processing roles and 64% of the workload of people in data collection roles1.
As businesses aim to improve the customer experience, enhance compliance and tackle the skills shortage and low productivity, more and more are turning to Robotic Process Automation (RPA) to extend the benefits of their existing investment in automation through bots that interact with separate applications in the same way that humans do.
Grand View Research expects the global market for RPA to grow from $199 million in 2016 to $8.7 billion by 2024, at a compound annual growth rate (CAGR) of 60.5%.
Growth of this magnitude is understandable considering RPA’s many benefits. These include:
Reduced labour costs. It is estimated that bots can reduce labour costs by around one third (where they are applied). By freeing staff from mundane tasks, RPA can create a more engaged workforce, potentially reducing churn and associated recruitment costs.
Better use of skilled employees. At a time of growing skills shortages, robots free experienced staff from administration, enabling them to use their skills more productively on revenue generation, problem-solving or customer service. Software wizards that guide a worker through a process reduce training requirements and enable less skilled workers to complete more advanced tasks.
Increased productivity. The Institute for Robotic Process Automation and Artificial Intelligence (iRPAAI) estimates that one software bot can do the workload of 2.5 to 3.5 workers. Nor is there any limit to how long they can work; bots can be programmed to work 24/7.
Greater accuracy/fewer errors. Humans involved in manual processes are prone to make mistakes due to lack of attention. Bots don’t get bored or tired and don’t make errors when copying data from one application to another. Robotic Process Automation (RPA) has the potential to increase productivity and cut costs, but there are also risks, so it is important to choose a supplier carefully
Better customer service. Through its combination of greater accuracy and increased productivity, RPA helps organisations deliver the high levels of customer service demanded by today’s consumers. Many organisations already use software agents, chatbots and virtual assistants in call centres to provide self-service options and help agents resolve queries more quickly.
Enhanced compliance. RPA can improve compliance, by eliminating unnecessary errors, making sure processes are followed to the letter and by cross-checking information. Almost nine out of 10 SMEs surveyed by OnePoll for Ultima said they were considering RPA to improve their IT security and data compliance.
Managing data growth. RPA can help businesses manage and exploit ever greater data volumes. According to the Gemalto Data Security Confidence Index, 65% of organisations are currently unable to categorise all the consumer data they have and only 54% know where all their sensitive data is stored. Bots can help organisations manage the information they have and eventually, through AI and machine learning, mine it for commercial advantage.
So compelling are these benefits that Forrester predicts that by 2021, there will be more than four million software bots doing office and administrative work in claims processing, HR administration, invoicing, procurement, IT service management, customer service/contact centre.
Strengths and weaknesses
Some organisations refer to bots as virtual workers or digital workers. The inference that bots are just another type of employee, almost on an equal footing with sentient beings, is indicative both of RPA’s strengths and its limitations.
In much the same way that temporary workers are brought in to support an organisation’s existing workforce, software ‘bots’ tend to work alongside full-time employees, using existing software programmes, the existing IT infrastructure and, essentially, the same workﬂows.
This makes them relatively easy and cheap to deploy (and scale) and makes it possible for individuals, teams and departments to transform their own processes without the time and expense of a big, centralised IT project.
On the ﬂip-side, software bots, like temps, are limited in what they are able to do: they can follow a process with prescribed steps; they can collect data from multiple sources; they can move information between applications; and they can do this 24/7, with complete accuracy.
However, they are not qualified to make value judgments, for which human input is still required, and are unsuitable for processes with too many steps or too many applications.
Attended or unattended?
There are two types of RPA – unattended and attended. These can be used separately or together.
Attended. With attended (or assisted) RPA, bots work in partnership with human workers. For this reason, they are sometimes referred to as personal assistant bots or software assistants. They run on the user’s PC; work with the same applications as the user; and are activated by the user pressing an embedded button in an application or by a trigger in a process or workﬂow.
This is the entry-level option for many RPA deployments. The close link between workers and robots; its applicability to existing workﬂows and processes; compatibility with the existing IT infrastructure; easily identifiable productivity benefits; and a relatively quick ROI mean that attended RPA is often implemented by a specific department or individual (with or without the close involvement of the IT department).
Attended RPA is typically used to perform routine, repetitive tasks within a process that involves human input. It completes these more quickly and with greater accuracy than a human can and enables employees to focus on parts of the process that require intelligence and judgment.
For this reason, attended RPA is widely used in customer-facing roles or self-service applications. In a call centre, for example, bots can access multiple applications to retrieve data needed to answer a customer enquiry, saving an agent from having to toggle between screens or copy information from one application to another while on the phone to a customer.
Unattended. As the name suggests, these robots tend to work in the background and reside on servers. They can be monitored and controlled centrally and can even request assistance, but typically operate independently of human workers. Because they start automatically (according to a schedule or an event e.g. when a file is received in a watched folder or drive), unattended bots can operate outside office hours, 24/7.
Unattended RPA is commonly used for collecting large volumes of data and distributing it to other applications. The ability to push and pull data makes it particularly suitable for applications like loans processing, invoice processing and account openings.
AI & machine learning
On its own, RPA has significant benefits for automating individual tasks, but leading RPA platforms are increasingly adding machine learning and natural language processing capabilities, which, as well as blurring the lines between RPA and artificial intelligence (AI), are further eroding distinctions between human and machine work.
Combining RPA with intelligence enables organisations to automate not just a process but an entire workﬂow, significantly increasing the potential benefits of automation and the digitisation of business processes.
For example, the integration of text analytics in Kofax KAPOW makes it possible to automate the extraction of data from unstructured and non standard documents, such as forms, email messages and invoices, and can include basic decision-making.
A bot could sense when a new invoice has arrived, use intelligence to find the information it needs in an unstructured document and, if something is missing, e.g. a PO number, look for it in another system. This shows a degree of problem-solving that can significantly reduce the number of exceptions that require human involvement to resolve.
As the technology develops and RPA platforms incorporate more sophisticated AI and machine learning capabilities, organisations will be able to automate more varied activities.
For the time being, there is still plenty of scope for them to implement rules-based RPA, as many organisations have yet to do even basic process automation. In a 2018 AIIM survey, 75% of respondents described process automation as ‘important’ or ‘very important’. Yet 67% admitted that fewer than half their processes were automated. Just 3% have automated more than 90% of their processes. (AIIM Industry Watch report Digitializing Core Business Processes, 2018).
Risk of failure
For all its benefits, RPA is not always successful. In fact, Avasant suggests that as many as 40% of RPA engagements fail to deliver the expected benefits. Gartner suggests the figure could be as high as 72%.
There are a number of reasons why an RPA implementation might fail. The wrong process might have been chosen for automation; it might have been chosen for the wrong reasons (i.e. automation for the sake of automation); there may have been an inadequate cost-benefit analysis; too many people might claim ownership of a project – and fail to act when something goes wrong; vested interests might resist change; the wrong metrics might be used to measure a project’s value.
When intelligent automation specialist Cortex analysed why implementations went wrong, 53% of those surveyed complained of a shortage of people with the necessary skills; 46% said they had not budgeted for the documentation of existing processes; 44% cited cultural resistance from employees fearful that robots were going to take their jobs; and 41% said they lacked process creation expertise.
A good supplier can help businesses avoid these pitfalls. Data Capture Solutions (DCS), a Neopost company, has helped hundreds of organisations cut costs and increase efficiency by automating business and document processes. With its experience in process automation and change management, it is well placed to help businesses of all sizes identify processes suitable for RPA and make sure the project is a success (see box on page 16).
For too long, businesses, their employees and their customers have had to put up with time-consuming, inefficient, manual processes. It is time to put robots to work for the benefit of all.
1. Where machines could replace humans – and where they can’t (yet), Michael Chui, James Manyika, Mehdi Miremadi, McKinsey Quarterley, July 2016