This site runs best with JavaScript enabled

Robotic Process Automation - a practical tool for an imperfect world

small logo
Robotic Process Automation— A Practical Tool for an imperfect world

Every organization has lots of repetitive jobs. Even though these jobs may take place in clean, air-conditioned offices with excellent coffee available, they are complicated, boring and repetitive tasks, that involve following rules (e.g., opening an application, copying some data, putting that data into another application, running a report, and distributing it to the right people).  These tasks are not fun, and do not create an engaged, happy, creative workforce.  But they exist, because no company has perfect systems.  Wasn't ERP supposed to change this?  But even those organizations that have fully embraced ERP software have legacy applications and systems sitting around that run important (if unglamorous) parts of the business- things like Accounts Receivable, Order Entry, and Payroll.  In the real world, there will always be awkward old systems that may be outdated, but are key to making business work.  For example, many banks still require multiple manual workarounds to maintain controls and quality due to incomplete upstream data feeds, poor quality reference data, multiple sources of data, evolving regulation and legacy IT infrastructure that is tactically updated rather than strategically reviewed.

Back office functions in almost every industry can take advantage of Robotic Process Automation (RPA) to minimize the repetitive jobs.  Not just because using RPA to automate these jobs decreases cost and increases quality, but because they are jobs that no one takes pleasure in doing.   Michiko Matsubara, PSS Group Manager of Japanese IT providers NRI put it well: “Investment banks are looking to create the most efficient environments possible, and freeing up their teams from grating and time-consuming tasks through innovative technology solutions is the first step towards upgrading a workforce.”  Investment Banks are not alone—everybody should want to free their teams from grating and time-consuming tasks.

Robotic Process Automation lets organizations do that, letting machines do the low-value processes and freeing up staff for higher-value (and usually more interesting) activities.

What is RPA?

Robotic Process Automation is the practice of configuring a software robot to mimic human behavior. i.e. a virtual person. This allows the robot to perform rules-based transaction processing-- for example, logging into and collating data from multiple applications, without the need to reprogram the ERP.  It best suits processes with the following characteristics:

*are rules-based with digital structured inputs,

*have repetitious transactional processing,

*require process adherence/quality

*involve multiple system interactions

*are scalable

*are stable

*where the cost of errors is high

*have a high number of FTEs doing repetitious work

*are complex/mission critical processes

*experience a fluctuation in demand or backlogs

These characteristics describe many critical, unglamorous tasks that every modern organization has to manage.  We most commonly see RPA in these departments that usually run well enough, but are rarely a priority for top management attention.

Accounting and finance: Accounts payableAccount reconciliation, Accrual support, GST management, and Treasury management

Customer Service: Customer on-boarding, Customer record management, Order tracking and fulfilment, Accounts Receivable, CRM system updates and downloads

Logistics / supply chain: Order processing automation, Multiple source data transferring to ERP

All of these processes are critical to running a business, and errors are costly.

Where does RPA fit into Automation?

There are three types of Robotic automation.

The first and simplest is Scripting, also known as Robotic Desktop Automation (RDA).  It supports repetitive work of a user using their IT credentials to access and interact with applications. RDA uses a software robot configured to login and interact with applications on a user’s desktop.  The robot behaves as an assistant to the human operating the desktop.  It works very well, for example in customer contact centers as a productivity aid to CSRs, but the typical feature of configuring these robots by “macro-recording” user’s tasks makes them more technical to modify and limits scalability.

More sophisticated is Execution Intelligence, a/k/a Robotic Process Automation, where a repetitive process flow of tasks and rules is replaced with a robot. The robot behaves as a virtual employee, using its own IT credentials to access applications.  In RPA the robot usually lives on a server, and task “objects” can be more easily reused to allow scale.  It offers richer exception handling than RDA, allowing it to operate in “lights-out” mode, requiring little human support once it is configured and running correctly.  These tools can often be configured more easily, with tools no more complex than a flow chart defining the keystrokes, expected outputs and workflow.

Beyond RPA lies Artificial Intelligence, which offers rich decision support, and has a huge amount of potential, but relatively few practical use cases for most organizations (currently, at least). Watson for example is remarkable technology, and may make a big difference in many business segments (e.g., Medical Diagnosis), but solves a different sort of problem than RPA does.

What are the Benefits?

Cost: The most obvious benefit is cost—software robots can do many routine information processing tasks at dramatically lower cost than onshore or offshore people—usually 1/4 the cost of an offshore person.

Availability: Robots will work 24/7, 365 days per year, without a break, sick time or a holiday.

Accuracy: Human processes have a typical error rate of 1-2%. So processes requiring high reliability often have multiple layers of supervision. Robots can eliminate the cost of errors, the cost of supervision and the cost of error remediation.  For certain repetitive processes (e.g., mortgage approvals), the benefits of eliminating the error rate can change the economics of the business.

Speed: We regularly see significant improvements in service levels from using robots driven by speed.  Robotic software can dramatically improve the speed at which processes are performed leading to significant increases in customer service levels and improved staff productivity.

Audit/Compliance: Robots can provide detailed audit logs enabling advanced business analytics and improved compliance.

These benefits compound-- some BPO providers have leveraged RPA technology to deliver productivity improvements of between 30% and 70% and improvements in accuracy of between 70% and 90%.  For example, automating pre-matching in a stock borrowing and lending process led to 80% of the manual effort being eliminated and a 70% reduction in trade fails through standardized approach and narrations.  You can see how there is a compelling case for most organizations.

How Big Is It?

Compared to its impact, it’s a small market, but growing very quickly.  The global market for RPA Software and Direct Services reached $271 million in 2016 and is expected to grow to $1.2 billion by 2021 at a compound annual growth rate of 36%. The direct services market includes implementation and consulting services focused on building RPA capabilities within an organization. It does not include wider operational services like BPO, which may include RPA becoming increasingly embedded in its delivery.


What are the risks?

While RPA is a back-office solution for almost every industry, the gains are not trivial to capture.

RPA will be affected by a quickly changing technology landscape and will require updates and bug fixes.  It is also sensitive to any changes in the user interface (UI) on which it operates, so like any application, any proposed changes to the IT environment must be communicated to the owners of RPA and tested to ensure the RPA does not break upon the production release of the new environment.

Limitations with existing infrastructure, automation and data ingestion-- RPA will successfully automate a process performed by humans, but it will not magically resolve data quality issues or existing platform limitations. If a process or its reliant data is fundamentally flawed or broken and this is converted from human-owned to robot-owned, the underlying data quality or process issues that existed in the old world will also exist, and indeed be significantly amplified, in the new.

And obviously, as with any new technology, there is a risk of not following through on execution. As we have seen with many IT breakthroughs, it takes a while for the organization to absorb them.  Without a rigorous focus on execution and communication, you won’t deliver on the promise of RPA.  It’s critical to develop a Center of Excellence for RPA inside the organization, which allows you to maintain the skills required to not just automate one or two functions, but to launch waves of transformation throughout the back office.  It also frees you from being entirely dependent on consultants.

So How Do I Get Started?

Typically, an organization will perform a short diagnostic which indicates which functions in the organization are (a) a good fit for RPA and (b) material enough that using RPA will make a measurable difference in Cost, Quality, or Time. It should not surprise you that many back office functions do not have KPIs that are rigorously tracked.  It’s worth investing time up front to establish a baseline to make sure you deeply understand how value is created by the function, how it is performed today, and the cost of under-delivery. This baseline can then be used to track the progress of the RPA program.

Before starting, the organization needs to undertake a few weeks of IT pre-work, to make sure the infrastructure of the development and test environments including “sand-boxes” are established.

Next is implementation of the First Wave of Ideas (2 or 3 of them) and the establishment of a Center Of Excellence for RPA. The COE enables the organization to maintain improvements as they are being made and keep deploying more robots without outside assistance. The typical implementation wave takes between 8 and 10 weeks.

More waves of implementation follow as necessary

What should we watch out for?

Some common mistakes we see in starting the RPA journey:

Looking at RPA as purely a “Headcount Reduction” initiative. There is clearly a headcount reduction opportunity from deploying RPA, and it should be taken seriously.  There is no point in preserving positions that are not needed.  But in almost every case, we have seen RPA freeing capacity to focus on higher value activities that help the company more, and are frankly more interesting to execute. And as much benefit or more often comes from increasing accuracy and speed as from improving staff productivity.

Not being rigorous enough in understanding how the team really spends its time.  Tools like Value Driver Trees and Day in the Life Of logs can help significantly to calculate benefits explicitly (e.g., How many people use how many hours to complete this task at what accuracy rate?).  They also make it easier to get organisational buy-in, as everyone is operating with a common well-understood baseline.

Having a bias for a particular software provider.  As the landscape of software providers above shows, there are meaningful differences between providers, and One Size Does Not Fit All.

Thinking of RPA as an IT Project.  IT needs to be an ally and enabler, and core IT disciplines must be followed for a successful implementation, but the push must come from and benefits owned by the Line

What’s Next?

While RPA may eliminate the need for humans to perform certain manual and repetitive jobs, this is not to say that robots are going to replace humans. RPA allows organizations to automate low-value tasks and focus business teams on the tasks that customers genuinely value.  As Andrew McAfee and Erik Brynjolfsson, authors of The Second Machine Age explain, 'the world is not ready to give up on human labour. Humanity is entering a second machine age. The first, spurred by the industrial revolution, was mechanical; this one is digital.  The first augmented our muscles; the second, our minds.'

We are in the early days of automation—while we now can effectively automate regular processes. In most cases, we are addressing each process standalone, not as part of a pervasive new way of thinking about the organization.  (In automotive terms, we are in the era of Ford and the assembly line, not Ohno and Lean Manufacturing.)

But even in this early stage, RPA offers you a chance to take the repetitive jobs out of your staff’s time and free them up to do work which is more interesting, creates more value, and makes more a difference to your customers.  We have taken our own medecine-- PIP has automated many functions in our own operations and have been very satisfied with the results.  If you’d like to talk about it with us, please get in touch.