Leslie Willcocks, London School of Economics, John Hindle; Knowledge Capital Partners and Mary Lacity, University of Arkansas
Innovation is the introduction of something new that creates value for the organisation that adopts it. Innovations can be characterised as: incremental (a series of small changes); radical (large transformative change); or revolutionary (game changing). RPA has generally been pitched very much in the ‘incremental’ camp, as what we will call operational innovations—changes in technology, software, work and personnel activities, designed to improve but not radically change how work is accomplished. Business process innovations, on the other hand, change the way the business operates in some important ways. For example, fundamental changes to business processes and relationships with customers, brought about by implementing CRM; using cloud-based processes as a service rather than on-premises solutions; IT-based billing system innovations that create new linkages between accounting, maintenance, service fulfilment and customer reporting. Then there are strategic innovations that enhance significantly a firm’s product and service offerings for existing customers, or enable entry to new markets.
Tantalisingly, looking across the hundreds of deployments we have studied for our new book, Becoming Strategic With Robotic Process Automation, we see many glimpses of RPA being used for all three types of innovation. However, frustratingly, organisational RPA efforts to date have focused overwhelmingly on operational gains. Much more can be achieved. Let’s look at some examples.
RPA Innovation Cases
Four cases illustrate what is possible. We deliberately choose some early stage examples to show that, if deploying RPA strategically (as discussed in an earlier blog paper) innovation is possible from the beginning and throughout—not necessarily just a final outcome after several years.
Case: Pre-building Trains
HA major Canada-based rail company developed strategy and governance for an automation program based on a Centre of Excellence approach, using a semi-federated model and a core team of developers (see chapter 6 of Becoming Strategic With Robotic Process Automation for more details). Rail operations need 24x7 support in many areas, and the company is always looking to improve efficiency, and find smarter operating models. The team automated five processes within five months, while ensuring the solutions interacted with custom in-house software. The processes included daily error reporting for intermodal train issues. The most complex process automated how the rail company pre-built trains in the rail yards. This was deliberately chosen to show to the wider business the power of RPA—prebuilding is, organisationally, widely recognised as a highly technical process. This success responded to the need for a solution that was relatively quick to deploy, highly scalable and, crucially, reliable. Here we are reporting early on a case which, nevertheless, saw freeing up FTE resources, and a 95 percent decrease in processing time for reporting queue errors. Note also how this project is being managed, with many of Rogers’ success factors being acted upon.
Case: Identity Theft
A major US-based utility began its utilisation of RPA in 2017 and, by 2019, was using RPA and cognitive technologies in numerous processes. Its vision for RPA was not limited to just automating the existing processes. The utility wanted to use RPA in conjunction with other digital initiatives—like AI and Machine Learning—to do more than would have ever been humanly possible.
One common challenge utilities face in North America is identity theft, whereby a person subscribes to and receives one or more services using a fake identity or someone else’s identity, but then does not pay. These contract accounts accrue arrears which are ultimately collected at less than 10 percent, as it is often difficult to positively identity the perpetrator of the ID fraud and penalise them under legal provisions. Minimising revenue loss from these contract accounts involves identifying the ID theft early so that the build-up of arrears can be minimised. Until recently, this energy company had no mechanism to proactively detect instances of identity theft. They were reliant on the victim of identity theft calling in to initiate an identity theft investigation.
To stem revenue loss from identity theft, the company’s Advanced Analytics team developed a comprehensive machine learning based model to proactively detect identity theft based on a myriad of predictors such as arrears, consumption pattern changes, duration at a premise etc. The model is able to identify, with a high degree of certainty, the contract accounts which are potentially identity theft. Despite the high degree of accuracy, the predictions of this model ran the risk of being relegated to a merely theoretical exercise because of the enormous manpower required to determine with certainty that the account is actually ID fraud, as identified by the model. Because of the enormous size of an energy company’s customer base, even a highly accurate model would flag more customers than it had resources to investigate manually. The time to analyse all the accounts would have potentially eliminated any potential value of identifying them sooner.
To solve this, the company decided to employ RPA with the machine learning model to automate the investigation of the cases. The RPA developed worked in conjunction with the machine learning model to perform a preliminary investigation of these cases, thereby ensuring that the fraud investigators only got to work with a set of contract accounts with a near 100 percent certainty of being fraud. Just RPA or the machine learning model by itself would not have generated enough value. However, leveraging the two technologies together resulted in a significant business value.
The combined solution was completed in mid 2019. The company anticipated the savings on lost revenue of greater than US$3 million each year. The machine learning portion of the solution is continuously assessed against parameters, such as prediction accuracy. The company also hope to refine the outcomes and improve accuracy with each iteration of the model. Encouraged by these positive outcomes achieved so far, by mid 2019 the company had already started working on a similar solution for energy theft—one of the other key contributors towards revenue loss for any utility.
>Case: Rent-a-Robot Service
This international BPO service provider offers banking operations and also a separate Rent-a-Robot service for its customers using an RPA platform. As at 2019, companies wanting to benefit from robotisation of their office work must acquire their own robotic skills, which is not economically viable for smaller companies. To make the service inexpensive, fast and attractive, the service provider built a robot factory based on mostly automated processes of development and maintenance. Robots can be delivered like agencies deliver temporary human work, but with no upfront cost, and with full maintenance and retraining offered. The high level of automation means robots can be delivered within six weeks, with robots testing other robots and being created by humans with little IT background.
The BPO service provider applied strict software engineering principles and created a team consisting of two capabilities—IT Excellence, which extended the RPA platform with their own custom tools, and Robo Shepherds, whose work is based on the Kanban agile methodology. The tools created included configuration manager, monitoring manager, and release manager, as well as a set of best practices, for example automated testing—all the robots are always tested by other robots, making manual testing unnecessary.
The work of the Robo Shepherds team is based on a robot ownership principle. That is, the same people are responsible for design, development, testing and maintenance of the robots throughout whole robot life cycle. The robotic factory covers a variety of processes on different platforms and applications (including SalesForce.com; Temenos T24; Altamira; SAP; IFS; and Uniflow). Some of the most successful processes are: transfers handling; costs settlement; insurance handling; collection support; closing of loans and deposits contracts; loans data calculation; and verification of customers’ loan applications. By way of example, the robotic factory has been successfully used to robotise operations for two clients. In one bank, robots-as-a-service achieved 19 percent running cost efficiency on the team of 82 people. At a second bank, robots are doing the equivalent of the four FTEs in a team of 15 people. The cost of the robot is a fraction of the cost of human employee doing the same work, and the Rent-a-Robot pricing model contains no up-front cost, so benefits appear after a month of deployment. Other benefits are higher quality, very fast deployment—3.5 weeks since start of cooperation and on average two new deployments of robots a week—and full maintenance provided, so customers do not need to develop RPA skills internally.
By late 2019 RPA had reached an interesting phase. It was experiencing exponential growth from an admittedly low base. All too few organisations had grasped RPA’s strategic and innovation implications, or seen it as part of a larger evolution of technologies that, over the next five years will providing the platforms and enablement for digital business. As a result they have missed what the leaders have not—that far from rushing past RPA and on to the next technological fix, the real business value lies in combining technologies right through the RPA, cognitive and AI continuum, as well as combining these with social media; mobile; analytics; cloud; blockchain; augmented reality; Internet of Things; and digital fabrication. One of the emerging strengths and strategic implications of RPA is its role in producing combinatorial innovation. Part of this is producing new business value and solving business problems by combining existing technologies in new ways. Our three cases illustrate some of the opportunities already being grasped by some organisations.
Leslie Willcocks, Mary Lacity and John Hindle are co-authors of ‘Becoming Strategic With Robotic Process Automation’—the fourth in a series of service automation books published by SB Publishing, and available to purchase directly from www.sbpublishing.org.