CIOs seeking operational efficiency are embracing robotic process automation (RPA), an emerging category of software programmed to execute routine tasks. The technology enables anything from low-level activities such as sifting through and generating response to emails, to orchestrating entire process flows in financial, HR and other corporate IT systems.
Unlike artificial intelligence (AI) and machine learning (ML) technologies, the capabilities of which are designed to improve as training models collect more data, RPA follows scripts imbued with business logic. IT leaders view RPA as a stepping stone for more sophisticated business automation enabled by AI and ML. But experts say that RPA requires special governance to ensure that it doesn’t break critical business processes.
An estimated 40 percent of enterprises will operate RPA centers and frameworks in 2019, according to Forrester Research. But most enterprises don’t know how to create such frameworks, let alone implement RPA at scale.
CIO.com found an outlier in Eaton, a $22.5 billion industrial concern specializing in power management, which opened an RPA center of excellence (CoE), a project for which it earned a 2019 CIO 100 Award in IT Excellence. Since 2018, Eaton’s RPA projects have saved more than 15,000 labor hours on an annualized basis, or the equivalent of 10 full-time employees. Like most companies who have implemented RPA, Eaton is using the software to augment, rather than replace existing employees, CIO Bill Blausey tells CIO.com.
Anatomy of an RPA CoE
Blausey traces the success of the program to 2017, when he appointed Eaton APAC IT leader Vish Krishnamurthi to assess and test new technologies as part of the company’s strategic imperative for reducing costs and human error while improving operations.
Recognizing that emerging technologies require a deeper discovery step to arrive at a set of hypotheses or use cases, Krishnamurthi established a new innovation model that incorporated surveillance, filtering, assessment and adoption. Eaton conducts spurts of rapid experimentation ranging from 4 to 8 weeks to test the viability of technology, followed by a product pilot validation phase of 4 to 8 months, culminating in a business case and large-scale deployment within the enterprise.
This innovation model informs the RPA CoE, which focuses on people, process, technology and governance involving multiple IT, business and operations stakeholders.
People. Gaining sponsorship from participating business functions, Blausey’s department funded the RPA CoE team, composed of 10 members, including program and project manages, business analysts, developers, QA testers and support. The RPA CoE initiated and drove projects, collaborating with business relationship managers interfacing with business functions to define and prioritize the RPA processes.
Process. The CoE created an “RPA Playbook” detailing the cross-functional process. For example, the playbook covered interaction between business functions and IT to identify right RPA candidates. It also captured standards for design, development and deployment, identity and security management, compliance and audit, infrastructure needs and future maintenance for RPA.
Technology. Eaton’s quality assurance team tested traditional regression testing tools to automate repetitive manual work. A quick win via a continuous improvement project led to a proof-of-concept (PoC) with an automation tool, in which the company explored the “art of the possible,” Krishnamurthi says. A subsequent RFP process by finance, HR and IT of five RPA tools recommended by Gartner led to the selection of vendor Automation Anywhere. The RPA COE then embarked on production pilots with three major customers to prove value before scaling to production deployment in 2018.
Governance. The CoE established a multi-tiered governance organization to ensure that the RPA program is aligned to business priorities, has buy-in from stakeholders, reviews the program roadmap to ensure it is on track and recommends course corrections as needed. This committee includes: an executing steering council comprised of senior business and functional leaders who met twice a month to provide guidance and sponsorship; an IT steering council with CoE leaders and business relationship managers who met monthly to align program priorities with the roadmap; a “decision gate” committee, whose leaders met frequently to weigh in with guidance and oversight; and project teams, whose leaders, process owners and vendor partners provided weekly updates via email as they implemented the technology.
In 2018, Eaton completed five major projects comprising 10 bots automating 17 processes. These include a cash application for finance; employee onboarding and interview updates for HR; VAT filing and reconciliation for tax; and invoice reconciliation for supply chain management. In 2019, the RPA CoE is scaling up to twice its capacity to meet the demand for process automation, working on an additional five projects that could yield another 15,000 hours’ worth of productivity gains on a yearly basis.
Additional benefits include streamlined processes, along with 100 percent auditing of bots, which provided new insights into root cause issues. Ultimately, Blausey says the RPA CoE improved Eaton’s processes, collaboration and innovation, helping it stay current on emerging technologies. Krishnamurthi offered the following four tips for scaling RPA.
Identify a candidate. Figure out a good use case for RPA and establish a sound process for execution. Selecting the technology is the last step.
Establish a multi-tiered governance. Incorporating several steering councils communicating and meeting regularly in a bottoms-up approach helped drive Eaton’s success.
Involve QA. Krishnamurthi attributes much of the program’s success to centering it in QA because “they live and breathe automation.” He adds, “They helped us make this an automation factory.”
Centralize and scale. Many enterprises are taking a federated, many tentacled approach to RPA. This is a non-starter. The centralized model helped Eaton consume RPA at scale, says Krishnamurthi.