Authors’ note: This is the second of a five-part series, in which we explore how financial services firms are using robotic process automation (RPA) to support accounting, compliance, financial risk management and other operations. In the first installment, we offer a list of seven deadly sins that can derail RPA adoption.
In our work with clients who are at various levels of automation maturity, we interact with automation strategists, analysts, technologists and evangelists. From these interactions and our deep experience across the financial services industry, we’ve identified seven steps that banks can take to automate efficiently and derive long-term value from their RPA investment.
- Establish leadership-level commitment and a sound CoE. It’s critical to gain both executive support for the RPA initiative and operational leadership’s buy-in to deploy RPA technology. This backing is needed to identify top candidates for automation, create an organizational change management plan and scale RPA across the enterprise. Appointment of an automation champion will help provide enterprise-wide visibility, credibility and funding, while evangelists situated throughout the various business units can identify critical on-ground challenges and opportunities. Establishment of an RPA center of excellence (CoE) can provide a focal point and governance framework for key aspects of the RPA initiative, including the organization’s evaluation of potential projects; the people, process and technology required; and monitoring of program impact and change.
- Introduce a bot implementation even before a proof of concept. Some organizations choose to begin RPA deployment by automating simple processes with high potential value in terms of reach and volume. This tactical proof-of-concept (PoC) approach can provide quick wins that build support and increase the potential for project funding. This measured approach, however, should be preceded by another step: a “Hello, World!” implementation. This is similar to learning a new programming language, when it’s critical to understand the overall environment setup, not just the syntax of the language. For example, an organization could create a bot that logs in to an internal Web app, searches for the company name on the first page, and creates a spreadsheet log of how many times the name appears on the first page. This kind of introductory bot implementation would help the business understand the complexities of the deployment environment, as well as uncover the organizational, procedural and infrastructural needs that need to be sorted out in order to streamline subsequent deployments of real bots.
- Always optimize and then automate. Automating a broken or inefficient process is a recipe for disaster and can jeopardize the automation program. Very often, processes are changed over time to accommodate organizational boundaries and system limitations, necessitating manual interventions that put unnecessary strain on the interacting system and often result in integration limitations. Automation introduces the opportunity to re-imagine the process in the context of the larger business transformation. Once a process becomes an RPA candidate based on potential return on investment, it is important to determine whether it can be improved and aligned with larger business objectives, and then proceed to do so before commencing automation. While this intermediate step might be viewed as slowing the introduction of automation, fine-tuning the process first is likely to provide substantial, rapid returns. For example, we’ve seen payment and retail banking processes that, as they evolved, required multiple manual interventions. The main contributing factors were the use of inflexible, vendor-supplied core systems and/or a recent merger or acquisition.
- Design to-be processes from the perspective of employees and customers. Processes are commonly designed from a systems perspective – inputs, outputs and required process elements. These elements can be complex and difficult for humans to understand. Designing RPA processes from the point of view of customers and employees, rather than from a systems perspective, can help increase the efficiency of human-process interactions and enhance the user experience. For example, we worked with a client on a process that used numeric codes rather than user-friendly names or descriptions. This created many data entry errors as information moved from one system to another. The RPA initiative gave us the opportunity to simplify this process in a way that allowed users to understand and enter required data in simple text, without changing the core system.
- Think beyond a single RPA platform. To maximize automation benefits, businesses need to look beyond simple rule-based automation and start identifying opportunities in line with larger business goals. This approach demands advanced AI technologies such as natural language processing and machine learning. No single platform can serve all these needs; instead, businesses need to carefully assess each platform player’s roadmap, investment and funding plans to determine which ones fit most of their needs. While a few platform providers are clearly investing in future-proofing their products, others are more focused on the current valuation of their company.
- Work as much as possible with “normal” scripting languages. Remember when mobile development was new, and there was a flood of native, hybrid and cross-platform development tools? What happened to strategies that involved investing in one all-encompassing tool? They generally failed. Likewise, RPA platforms are still evolving. Many of the platform players in the market will undergo consolidation, and some may not even exist in a few years. Relying on a particular RPA development tool can result in vendor lock-in. While RPA tools do need to be selected, limiting their use can help avoid such complications. Open source languages such as PowerShell, Python and VBS have the power and flexibility to handle much of the work of RPA data extraction and integration within the ecosystem, while RPA tools can be used to orchestrate process, monitoring and scheduling. By taking this approach, businesses can avoid being at the mercy of a single vendor’s capabilities and roadmap.
- Monitor and measure RPA performance and impact. When a robot goes into production, the surrounding systems will feel it. What a human does in an hour, the bot does in a fraction of the time, dramatically increasing data volumes that hit interfacing systems. This is a key opportunity for IT to help determine thresholds on various systems and configure robots to suit existing environments. By doing so, the business can avoid disaster on the first day of deployment due to the unexpected surge in transactions. Using a continuous monitoring program, organizations can keep track of both the bot’s performance and its impact on interfacing applications and IT infrastructure, initiating actions when necessary. In one case, a bot we were implementing was going to interact with a mainframe system that was on Variable Workload License Charges (i.e., the organization’s costs were based on the peak rolling four-hour average utilization across a month’s time). By anticipating the higher workload due to the bot’s faster rate of execution, we were able to foresee and avoid the higher mainframe licensing costs by spacing out execution without impacting peak workload.
Step Up to RPA Success
Three factors are critical to RPA success. First, as noted earlier, executive management sponsorship of the initiative is essential. Second, internal RPA specialists and business subject matter experts will need to work together to steer the organization through the fundamental changes brought about by automation. Finally, knowledgeable consulting and implementation support can help the organization plot the automation strategy and capture greater value from RPA deployment.
In our next installment, we’ll share some RPA success stories that are already unfolding and look ahead to what financial institutions can do with RPA in the future. In our fourth installment, we look at what banks should be thinking about when moving from proof of concept-based, task-level automation to overall operations transformation.
Sriniketh Chakravarthi, who leads Cognizant’s North American Banking and Financial Services business unit, contributed to this blog.