Tips on Scaling RPA from the Philippines
Add bookmark
Last year, SSON posted an article online that championed “operationalising RPA” as the right solution for Shared Services centres in the Philippines. Given that traditional service delivery models in the Philippines were built on low-cost FTEs and linguistic capability, the automation wave engulfing support services meant these opportunities were no longer as compelling as they once were. With the advent of RPA, the Philippines found itself in the midst of a far-reaching ‘reinvention’ of the workplace, based on intelligent processing capabilities that were fast taking over transactional activities.
However, rather than spelling disaster for an industry heavily relying on human resources, RPA also advocates a real opportunity.
That opportunity is growing. Once the proof of concept, or pilot, gains the green light, RPA projects are being implemented quickly and broadly. Indeed, in our 2018 annual industry survey, around 80% of respondents from the Philippines confirmed that intelligent automation was either implemented, in testing, or being planned.
With RPA projects well underway, therefore, Shared Services practitioners are now turning their attention to how to scale this robotic capability across the enterprise. The early pioneers have learned the hard way, in some cases by needing to retrace their steps. For those still at the early stages, there are plenty of learnings to guide them.
Steve Harris heads what is arguably one of the most far reaching RPA adoptions in the region. As Managing Director of ANZ Global Services and Operations in Manila, Philippines, Steve oversees a team of around 3,000 people in Manila, one of 3 centres with circa 10,000 people across the Asia region, all providing process support and value services to stakeholders globally. Across the offshore Service Centres, ANZ has a digital workforce of around 4,000 robots, a number of machine learning (ML) implementations, and a pipeline of many more machine learnings in pilot phase. ‘Scaling up’ has been a top priority for him and his counterparts in other locations.
Scaling is not simply a matter of bringing in more robots, however. It represents a meticulously planned rollout of robotic capability across processes, as well as leveraging new robotic capabilities like machine learning to exponentially drive performance across already automated activities. The single most significant enabler of successful scaling, says Steve, is governance. Here are 5 tips to build scalability into your automated operations.
1. Governance guides growth
Scaling robotics across an organisation as distributed as ANZ is a challenge, but as long as certain ground rules are followed and the right partnerships are formed with technology, this is easily done. "The key to scaling is governance," explains Steve. "It’s the one thing that determines where and how you will deploy robotics, upfront, via a framework, and collaborating closely with technology counterparts."
This means identifying the processes you want to robotise first, with an eye to expanding later, whereby it makes sense to avoid complex processes and, instead, focus on the low hanging fruit – repeatable, well structured activities that can easily be replicated. "Governance provides the guidance you need to plan ahead. It's not just about managing risk, but also about determining which processes are a priority. As long as you choose the right processes, you can scale fairly quickly," says Steve.
2. Freeing capacity to accelerate RPA
At ANZ, capacity is being freed up by robotising the "makers" (previously the person doing the transaction), but still relying on "checkers," as human controls. Robotising the makers is the classic example of freeing up staff to do more valuable work, “however you choose to define that,” adds Steve.
"To scale automation, we are constantly and proactively reinvesting our freed up capacity into accelerating our RPA program," he explains. "This means we can speed up scaling or absorb more volume without hiring extra people; offering more, but with no more resources – which is always a challenge in a cost constrained environment."
Before this newly released workforce can apply itself to scaling robotic automation, however, it needs to be significantly upskilled. "What might previously have been a subject matter expert [SME] for a given process," explains Steve, "is now being coached in three additional fields: human centred design, lean methodology, and agile – the combination of which is a powerful lever.”
The cross-skill training being invested in these resources means they are able to reassess a process, determine where robotics is a good fit, and often design the robots themselves, Steve explains. "In other words, we are able to shift them into the next wave of activity in changing another processes, thereby supporting a ripple effect of robotics scaling across the enterprise."
The fact is that most processes are fairly similar. What cross-skilling does, is allow a process to be reengineered via lean so that it is more efficient, and construct it around human centred design concepts, considering customers’ pain points. “An understanding of human centred design in combination with lean can be applied everywhere to resolve engineering issues that block process performance or flow,” says Steve. "It's good for the customer, of course, but it's also good for our employees. Not only are they better equipped for a lucrative career path within ANZ, but they have gained highly valuable skills in the process.”
3. Empowerment drives implementations
Driving enthusiasm in the production lines to want to automate is key. The strategy at ANZ has been to light what Steve calls "many small fires" by empowering individuals to either develop their own robotic automation, or call in the robotics team to resolve their problems. "The key," says Steve, "is to energise individuals to the extent that they pull automation to the process, as opposed to us pushing automation their way, from senior management. This reflects the lean principles espoused by Toyota, and is far more effective, explains Steve.
There is an inherent danger in this strategy, however. Given the enhanced capabilities of employees, and the empowerment that is encouraged, it may lead to too many tasks being automated across a given process, warns Steve. “The problem is, the more steps that are automated, the more the robot becomes highly complex and difficult to manage. Certainly, it may become too complex to maintain in the long term.” It also exposes you to vulnerability, warns Steve, for example, if the robot analyst leaves; or where business continuity kicks in, if the sister process does not offer the same robotic capability or knowledge to maintain the process in the back up site.
“It’s far better, in my experience, to deploy multiple robots on a task-based approach,” suggests Steve. “We’re not talking super fancy, end-to-end robotic solutions here. In fact, to my mind that would present a flaw in thinking, as it’s better to fix the underlying application than bolting complex robotic solutions on top, certainly for the long-term.”
4. Leveraging the power of robotics through machine learning
Machine learning is often described as a continuum within the intelligent automation range of activities. However, machine learning cannot stand on its own. It requires the automations that have gone before in order to drive the data and the behaviour that the machines learn from.
"We have, to date, implemented a number of different machine learning (ML) opportunities," explains Steve. "In each case we use screen-scraping technology, place a robot at the front end to feed into ML code, then another robot picks up the output at the other end. It's a fairly complex procedure that requires a lot of meticulous design and an upfront ability to feed historical data into the machine learning component to improve the reliability of the automation.”
However once mastered, machine learning offers the opportunity to ratchet up the impact of robotics across a given process. "It's not just about spreading robotic activity across the enterprise, but also about ‘smartening up’ that activity to yield better returns. That's what we are expecting from machine learning," explains Steve.
5. Setting expectations
Scaling robotics, ultimately, improves performance for the enterprise and for its people. In Manila, Steve overseas around 3,000 people directly, who have all become enthusiastic proponents of RPA. “They want more of it. They know it’s for the future, so they are actively helping themselves to build more automation into the processes they manage,” explains Steve. “It’s my responsibility, and that of my leaders, to support this strategy wholeheartedly.”
That means, Steve explains, promoting the industrialisation of processes where it makes sense and setting relevant metrics and goals around desired targets – “whereby it's not about the number of robots in service, but about the outcomes they deliver for our customers."
This is a key factor, emphasises Steve. The focus needs to be on process output quality, not robotic implementations. "If we want quality to improve and turnaround cycles to reduce, then we set that as a target and we encourage the adoption of robotic automation and process reengineering to achieve that target. But it's got to be in that order of priority."
Finally: it's not always about robotics
While RPA is delivering significant wins at businesses like ANZ, these wins are eclipsed by the potential returns on investment delivered by simple process redesign to ensure an efficient flow, says Steve. He is so convinced of this that he is now championing a campaign to re-evaluate processes for opportunities to reengineer them. "We are tapping into our lean, HCD and agile capabilities to review our processes. To do so we deploy small teams or squands – consisting of automation engineers, process/business excellence engineers, and subject matter experts, in three-week iterations or sprints.”
The results have been impressive, he says. "Process engineering is far more time-intensive, of course, and requires more change management and risk management than an RPA fix. And yet the benefits are significant. So while we are strategically planning to scale out the process-based benefits of robotic process automation, there still is, and always will be, the opportunity to ultimately redesign the process for far superior returns.”
